[{"_1":2,"_3":-5,"_4":-5},"loaderData",{"_5":6,"_7":8,"_9":10},"actionData","errors","root",{"_393":394,"_395":396},"routes/_app.x.$handle.$name",{"_99":98,"_32":44,"_107":89,"_27":108,"_13":14,"_109":-5,"_11":110,"_111":112,"_113":114,"_115":116,"_117":118,"_20":-5,"_119":-5,"_120":-5,"_121":25,"_122":123,"_124":58,"_125":-5,"_126":-5,"_127":128},"routes/_app.x.$handle.$name.branch.$branch.version.$version.diff",{"_11":12,"_13":14,"_15":16,"_17":-5,"_18":-5},"project",{"_42":43,"_32":44,"_37":45,"_46":-5,"_47":25,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":66,"_67":68,"_69":70,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":82,"_83":84,"_85":86,"_87":25},"version",156,"changes",[19],"aiChatInfo","userId",{"_20":21,"_22":23,"_24":25,"_26":25,"_27":28,"_29":28},"file",{"_30":31,"_32":28,"_33":20,"_34":36,"_37":38,"_39":14,"_40":41,"_24":25},"previous",{"_30":31,"_32":28,"_33":20,"_34":35},"deleted",false,"deletedHere","path","PIVOT216_ROLE_FLIP_PROTOCOL_V2.md","latestPath","fileId","12c4ae26-ab24-11f1-ab88-1607ee4eb77e","name","fileType","content","# Pivot v2 role-flip study — frozen before new outcome inspection\nDate: 2026-09-08 JST. User corrected v1: strong breakout -> trade with the break near the pivot, next pivot TP, invalidation beyond anchor pivot; fade only after repeated rejection. User example is a 5m NVDA cash chart with D P broken upward. Resistance becomes prospective SUPPORT after an upward break. V1 always-fade / +/-1ATR exits did not test this method.\n\nResearch only. No account state, orders, signatures, schedules or paid data. Preserve v1 inputs/results. Input blob prefix research:pivot216:NVDA:20260908. New output prefix research:pivot216:NVDA:20260908:roleflip_v2.\n\n## Point-in-time levels\nPrevious completed source-period Traditional D/W/M; discard first partial W/M source periods. CASH_RTH independent Yahoo OHLC vs XYZ_UTC OHLC separately; execute XYZ only except clearly labeled cash control. Actual TradingView feed/period mapping not certified. Reuse frozen historical inputs; do not refresh into the original prefix.\nEntries use default displayed D P/R1/R2/S1/S2, W P/R1/S1, M nearest above/below PRECEDING close. Target search uses full D R1-R5/S1-S5, W R1-R3/S1-S3, M R1-R5/S1-S5 as original next-target code. R/S names do not fix trade direction; ANY pivot can flip roles. Levels within max(0.20 ATR,2 ticks) of anchor are treated as one zone for target search, not independent obstacles. Target = first full-grid level beyond that zone in intended direction, fixed at signal time. It is never moved farther away to improve RR. Price tick=0.01 for these NVDA approximations.\n\n## Frozen numerical interpretation (research assumptions, not supplied numeric rules)\nATR14 Wilder of execution candles; A=previous completed candle ATR; 50-bar warm-up.\nStrong cross long: preceding close<=L, current close>=L+0.25A, bullish body>=0.50A, close in upper25% of current range; short mirrored. New-period level shifts alone never count as crosses: anchor must exist on preceding bar.\nBREAK_NEAR: after strong cross, buy/sell next candle open only if still on break side and within1.0A of anchor. No hypothetical fill at already-passed pivot price.\nRETEST_LIMIT: after strong cross, rest limit L+0.10A (long), L-0.10A (short), beginning NEXT bar; expires in1 hour or when source anchor expires. Require1 tick penetration for intrabar fills, not mere touch. A gap across the stop is filled conservatively at limit then stopped at adverse open, not magically canceled before the gap. No TP credit on an intrabar entry bar because ordering is unknown; SL can occur. If target hit without fill, cancel only after that bar closes.\nREJECT_2: within6h, >=2 distinct failed attempts at same active pivot from same side, separated by at least2 candles AND retreat close>=0.50A away. Touch range within0.20A, close back>=0.10A on approach side. A decisive close beyond0.25A resets attempts. Enter next open on approach side, <=1A from anchor. One continuous cluster is not several rejections. No fade against a simultaneously qualified strong breakout at another nearby visible pivot.\nHYBRID: strong break has priority; if signal close<=1A from pivot, BREAK_NEAR, otherwise RETEST_LIMIT; if no strong cross, REJECT_2. One position or pending entry at a time. No same-exit-bar reentry. Never combine independently tested modes by adding their trades.\nSL: anchor minus0.50A for long / plus0.50A for short. REJECT_2 also places SL beyond observed attempt extremes by0.10A. TP = next pivot, not1ATR. SL rounded outward; TP rounded toward entry. Require target reward greater than assumed6bps roundtrip cost, valid risk>0, no minimum RR in base. Pending orders/trades freeze anchor/TP/SL/A. Max holding6 clock hours for ALL execution timeframes. Cash-only closes before session end; no overnight cash entry. Signals near data end excluded so full entry+holding horizon is observable.\nBoth SL/TP within same execution OHLC => stop-first. Adverse gaps at open; favorable gaps capped at target. Structural levels do NOT move after future daily rollover. Historical Mark/L2/queue/partial fills/size capacity/funding not recreated. No claim these are realized exchange Net profits.\n\n## Evaluation (no parameter optimization)\nBase friction6bps total entry+exit; sensitivities0/12/20bps on same trades. Report both fixed-risk R and equal-notional bps, profit factors for each, target success/SL/timeouts, trade win rate, maxR drawdown, actual RR, pending fill counts, ambiguous bars, long/short and D/W/M and signal-time session breakdowns. No daily dollar target sizing.\nMain28 scenarios: XYZ 5m/15m/1h x CASH_RTH/XYZ_UTC x BREAK_NEAR/RETEST_LIMIT/REJECT_2/HYBRID=24, plus CASH5m four modes. Secondary17: RTH-only HYBRID5m/15m both sources(4); D P-only LONG HYBRID5m/15m both sources(4); CASH_RTH XYZ15m HYBRID one-factor SL0.25/0.75, body0.35/0.75, netRR>=1.0(5); common5m-overlap HYBRID15m/1h both sources(4). Total45 reported strategy scenarios. Additionally20 deterministic shifted-grid placebo HYBRIDs for each15m source=40. Changes after results must be logged, not hidden.\nSplit each full dataset70/30 by timestamp, excluding boundary-spanning trades. This is retrospective temporal stability, NOT untouched out-of-sample: v1 already examined these dates. Weekly block bootstrap1000 draws; intervals exploratory, few weekly clusters and multiple testing caveats explicit. Same-date secondary removes date confounding but not bar-resolution/intrabar ambiguity differences. Result-selected attractive subgroups are hypotheses, not validated strategies.\n\n## Reuse\nPure engine takes normalized bars/daily inputs/config; new symbols must explicitly specify source instrument, price scale, tick, execution/session clock, holidays/DST and fees. NY cash classifier is NVDA-specific. Keep strategy research separate from live ledger/handoff.\n","# Pivot v2 role-flip study — frozen before new outcome inspection\nDate: 2026-09-08 JST. User corrected v1: strong breakout -> trade with the break near the pivot, next pivot TP, invalidation beyond anchor pivot; fade only after repeated rejection. User example is a 15m NVDA cash chart with D P broken upward. Resistance becomes prospective SUPPORT after an upward break. V1 always-fade / +/-1ATR exits did not test this method.\n\nResearch only. No account state, orders, signatures, schedules or paid data. Preserve v1 inputs/results. Input blob prefix research:pivot216:NVDA:20260908. New output prefix research:pivot216:NVDA:20260908:roleflip_v2.\n\n## Point-in-time levels\nPrevious completed source-period Traditional D/W/M; discard first partial W/M source periods. CASH_RTH independent Yahoo OHLC vs XYZ_UTC OHLC separately; execute XYZ only except clearly labeled cash control. Actual TradingView feed/period mapping not certified. Reuse frozen historical inputs; do not refresh into the original prefix.\nEntries use default displayed D P/R1/R2/S1/S2, W P/R1/S1, M nearest above/below PRECEDING close. Target search uses full D R1-R5/S1-S5, W R1-R3/S1-S3, M R1-R5/S1-S5 as original next-target code. R/S names do not fix trade direction; ANY pivot can flip roles. Levels within max(0.20 ATR,2 ticks) of anchor are treated as one zone for target search, not independent obstacles. Target = first full-grid level beyond that zone in intended direction, fixed at signal time. It is never moved farther away to improve RR. Price tick=0.01 for these NVDA approximations.\n\n## Frozen numerical interpretation (research assumptions, not supplied numeric rules)\nATR14 Wilder of execution candles; A=previous completed candle ATR; 50-bar warm-up.\nStrong cross long: preceding close<=L, current close>=L+0.25A, bullish body>=0.50A, close in upper25% of current range; short mirrored. New-period level shifts alone never count as crosses: anchor must exist on preceding bar.\nBREAK_NEAR: after strong cross, buy/sell next candle open only if still on break side and within1.0A of anchor. No hypothetical fill at already-passed pivot price.\nRETEST_LIMIT: after strong cross, rest limit L+0.10A (long), L-0.10A (short), beginning NEXT bar; expires in1 hour or when source anchor expires. Require1 tick penetration for intrabar fills, not mere touch. A gap across the stop is filled conservatively at limit then stopped at adverse open, not magically canceled before the gap. No TP credit on an intrabar entry bar because ordering is unknown; SL can occur. If target hit without fill, cancel only after that bar closes.\nREJECT_2: within6h, >=2 distinct failed attempts at same active pivot from same side, separated by at least2 candles AND retreat close>=0.50A away. Touch range within0.20A, close back>=0.10A on approach side. A decisive close beyond0.25A resets attempts. Enter next open on approach side, <=1A from anchor. One continuous cluster is not several rejections. No fade against a simultaneously qualified strong breakout at another nearby visible pivot.\nHYBRID: strong break has priority; if signal close<=1A from pivot, BREAK_NEAR, otherwise RETEST_LIMIT; if no strong cross, REJECT_2. One position or pending entry at a time. No same-exit-bar reentry. Never combine independently tested modes by adding their trades.\nSL: anchor minus0.50A for long / plus0.50A for short. REJECT_2 also places SL beyond observed attempt extremes by0.10A. TP = next pivot, not1ATR. SL rounded outward; TP rounded toward entry. Require target reward greater than assumed6bps roundtrip cost, valid risk>0, no minimum RR in base. Pending orders/trades freeze anchor/TP/SL/A. Max holding6 clock hours for ALL execution timeframes. Cash-only closes before session end; no overnight cash entry. Signals near data end excluded so full entry+holding horizon is observable.\nBoth SL/TP within same execution OHLC => stop-first. Adverse gaps at open; favorable gaps capped at target. Structural levels do NOT move after future daily rollover. Historical Mark/L2/queue/partial fills/size capacity/funding not recreated. No claim these are realized exchange Net profits.\n\n## Evaluation (no parameter optimization)\nBase friction6bps total entry+exit; sensitivities0/12/20bps on same trades. Report both fixed-risk R and equal-notional bps, profit factors for each, target success/SL/timeouts, trade win rate, maxR drawdown, actual RR, pending fill counts, ambiguous bars, long/short and D/W/M and signal-time session breakdowns. No daily dollar target sizing.\nMain28 scenarios: XYZ 5m/15m/1h x CASH_RTH/XYZ_UTC x BREAK_NEAR/RETEST_LIMIT/REJECT_2/HYBRID=24, plus CASH5m four modes. Secondary17: RTH-only HYBRID5m/15m both sources(4); D P-only LONG HYBRID5m/15m both sources(4); CASH_RTH XYZ15m HYBRID one-factor SL0.25/0.75, body0.35/0.75, netRR>=1.0(5); common5m-overlap HYBRID15m/1h both sources(4). Total45 reported strategy scenarios. Additionally20 deterministic shifted-grid placebo HYBRIDs for each15m source=40. Changes after results must be logged, not hidden.\nSplit each full dataset70/30 by timestamp, excluding boundary-spanning trades. This is retrospective temporal stability, NOT untouched out-of-sample: v1 already examined these dates. Weekly block bootstrap1000 draws; intervals exploratory, few weekly clusters and multiple testing caveats explicit. Same-date secondary removes date confounding but not bar-resolution/intrabar ambiguity differences. Result-selected attractive subgroups are hypotheses, not validated strategies.\n\n## Reuse\nPure engine takes normalized bars/daily inputs/config; new symbols must explicitly specify source instrument, price scale, tick, execution/session clock, holidays/DST and fees. NY cash classifier is NVDA-specific. Keep strategy research separate from live ledger/handoff.\n","createdAt",["D",1788830916590],"branchVersion","branchId","9b5029ed-e69a-483d-8cb3-234853c9a254","id","87f5b472-21e5-4767-9ac3-e191e3944e56","hyperliquid-market-bridge",["D",1788584474524],"deletedAt","legacyEnvVars","privacy","public","httpPrivacy","description","Public market data only; on demand; no trading.","imageUrl","0e061ab1-2aec-4517-be54-467ef7fcb625","pinnedFileId","01a06ff1-37fe-726c-835e-8d7a62a2a226","pinPreview",true,"pinMigrated","screenshotVersion","likeCount",0,"orgId","01a06fee-2d51-773c-94a7-f1125fac90b4","tursoName","87f5b47221e547679ac3e191e3944e56","tursoHostname","87f5b47221e547679ac3e191e3944e56-val-town.aws-us-east-2.turso.io","tursoMeasuredAt",["D",1789491713377],"pinned","userPerm","r","valPerm","blobPerm","rw","emailPerm","sqlitePerm","projectPerm","telemetryPerm","user",{"_99":98,"_106":102},"org",{"_99":98,"_100":98,"_101":102,"_18":54,"_103":104,"_105":25},"branch",{"_13":88,"_32":89,"_42":41,"_90":43,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":44,"_97":98},"isOwned",187,"main","projectId","forkedBranchId","forkedBranchVersion","forkedProjectPrivacy","forkedProjectId","forkedProjectOrgId","projectName","userHandle","kamibaba","handle","displayName","avatar","https://avatars.githubusercontent.com/u/51115889?v=4","tier","free","canRestrictVals","profileImageUrl","branchName","","adoptOrgId",{"_42":43,"_32":44,"_37":388,"_46":-5,"_47":25,"_48":49,"_50":49,"_51":52,"_53":-5,"_18":54,"_55":56,"_57":58,"_59":-5,"_60":-5,"_61":62,"_63":64,"_65":66,"_67":68,"_69":389,"_71":-5,"_72":73,"_74":73,"_75":76,"_77":76,"_78":76,"_79":73,"_80":73,"_81":390,"_83":391,"_85":392,"_87":25},"tree",[295,298,299,300,301,302,303,304,305,306,307,308,309,310,311,312,313,314,315],"files",[295,296,297,298,299,300,301,302,303,304,305,306,307,308,309,310,311,312,313,314,315],"counts",{"_285":62,"_286":252,"_287":62,"_288":289,"_290":291},"contents",[129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147],"upstreamProject","upstreamBranch","emptyEnvVars","flags",[],"hasHttpFiles","grants","orgs","customDomains",{},{"_42":280,"_32":49,"_30":281,"_151":-5,"_40":41,"_39":252,"_153":58,"_37":282,"_13":62,"_33":283,"_34":-5,"_24":25,"_157":-5,"_27":284},{"_42":273,"_32":274,"_30":275,"_151":-5,"_40":41,"_39":276,"_153":58,"_37":277,"_13":62,"_33":20,"_34":278,"_24":25,"_157":-5,"_27":279},{"_42":266,"_32":267,"_30":268,"_151":-5,"_40":41,"_39":269,"_153":58,"_37":270,"_13":62,"_33":20,"_34":271,"_24":25,"_157":-5,"_27":272},{"_42":263,"_32":28,"_30":31,"_151":-5,"_40":41,"_39":14,"_153":25,"_37":264,"_13":252,"_33":20,"_34":36,"_24":25,"_157":-5,"_27":265},{"_42":255,"_32":256,"_30":257,"_151":-5,"_40":41,"_39":258,"_153":58,"_37":259,"_13":260,"_33":20,"_34":261,"_24":25,"_157":-5,"_27":262},{"_42":247,"_32":248,"_30":249,"_151":-5,"_40":41,"_39":250,"_153":58,"_37":251,"_13":252,"_33":20,"_34":253,"_24":25,"_157":-5,"_27":254},{"_42":239,"_32":240,"_30":241,"_151":-5,"_40":41,"_39":242,"_153":58,"_37":243,"_13":244,"_33":155,"_34":245,"_24":25,"_157":-5,"_27":246},{"_42":232,"_32":233,"_30":234,"_151":-5,"_40":41,"_39":235,"_153":58,"_37":236,"_13":62,"_33":155,"_34":237,"_24":25,"_157":-5,"_27":238},{"_42":224,"_32":225,"_30":226,"_151":-5,"_40":41,"_39":227,"_153":58,"_37":228,"_13":229,"_33":155,"_34":230,"_24":25,"_157":-5,"_27":231},{"_42":216,"_32":217,"_30":56,"_151":-5,"_40":41,"_39":218,"_153":58,"_37":219,"_13":220,"_33":221,"_34":222,"_24":25,"_157":-5,"_27":223},{"_42":209,"_32":210,"_30":211,"_151":-5,"_40":41,"_39":212,"_153":58,"_37":213,"_13":62,"_33":155,"_34":214,"_24":25,"_157":-5,"_27":215},{"_42":202,"_32":203,"_30":204,"_151":-5,"_40":41,"_39":205,"_153":58,"_37":206,"_13":62,"_33":155,"_34":207,"_24":25,"_157":-5,"_27":208},{"_42":195,"_32":196,"_30":197,"_151":-5,"_40":41,"_39":198,"_153":58,"_37":199,"_13":62,"_33":155,"_34":200,"_24":25,"_157":-5,"_27":201},{"_42":188,"_32":189,"_30":190,"_151":-5,"_40":41,"_39":191,"_153":58,"_37":192,"_13":62,"_33":155,"_34":193,"_24":25,"_157":-5,"_27":194},{"_42":180,"_32":181,"_30":182,"_151":-5,"_40":41,"_39":183,"_153":58,"_37":184,"_13":185,"_33":155,"_34":186,"_24":25,"_157":-5,"_27":187},{"_42":173,"_32":174,"_30":175,"_151":-5,"_40":41,"_39":176,"_153":58,"_37":177,"_13":62,"_33":155,"_34":178,"_24":25,"_157":-5,"_27":179},{"_42":166,"_32":167,"_30":168,"_151":-5,"_40":41,"_39":169,"_153":58,"_37":170,"_13":62,"_33":155,"_34":171,"_24":25,"_157":-5,"_27":172},{"_42":159,"_32":160,"_30":161,"_151":-5,"_40":41,"_39":162,"_153":58,"_37":163,"_13":62,"_33":155,"_34":164,"_24":25,"_157":-5,"_27":165},{"_42":148,"_32":149,"_30":150,"_151":-5,"_40":41,"_39":152,"_153":58,"_37":154,"_13":62,"_33":155,"_34":156,"_24":25,"_157":-5,"_27":158},"f911537f-535a-4954-b4a3-8912461ebcf9","tmp_reconcile_after_exits.ts","a6902602-aa03-11f1-9940-1607ee4eb77e","parentId",63,"latest",["D",1788706808115],"script","const user='0x75e6e34a78cD19054BD0f298eb47b6DdfA2c4127';\nconst INFO='https://api.hyperliquid.xyz/info';\nconst post=async(b:any)=>{const r=await fetch(INFO,{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify(b)});if(!r.ok) throw new Error(`${r.status}:${await r.text()}`);return await r.json();};\nconst startTime=Date.parse('2026-09-05T15:00:00Z');\nconst recentStart=Date.parse('2026-09-06T14:40:00Z');\nconst [state,orders,xyzState,xyzOrders,fills,funding]=await Promise.all([\n post({type:'clearinghouseState',user}),post({type:'openOrders',user}),post({type:'clearinghouseState',user,dex:'xyz'}),post({type:'openOrders',user,dex:'xyz'}),post({type:'userFillsByTime',user,startTime,aggregateByTime:false}),post({type:'userFunding',user,startTime})\n]);\nconst perp=fills.filter((f:any)=>f.feeToken==='USDC' && /(Long|Short)/i.test(String(f.dir||'')));\nconst gross=perp.reduce((s:number,f:any)=>s+Number(f.closedPnl||0),0),fees=perp.reduce((s:number,f:any)=>s+Number(f.fee||0),0),fund=funding.reduce((s:number,x:any)=>s+Number(x.delta?.usdc||0),0);\nconsole.log(JSON.stringify({positions:state.assetPositions.map((x:any)=>x.position),orders,xyzPositions:xyzState.assetPositions.map((x:any)=>x.position),xyzOrders,recentFills:perp.filter((f:any)=>f.time>=recentStart).map((f:any)=>({coin:f.coin,px:f.px,sz:f.sz,dir:f.dir,closedPnl:f.closedPnl,fee:f.fee,tid:f.tid,oid:f.oid,time:f.time})),strict:{gross,fees,funding:fund,net:gross-fees+fund}}));","label",[149],"f8b8f3da-df4b-422c-b306-7ed819e6a211","structure_levels_readonly.ts","bf984f78-aa6a-11f1-a401-1607ee4eb77e",112,["D",1788751088273],"// Read-only price-level audit. No account identifiers or Exchange actions.\nconst now = Date.now();\nconst targets = [{coin:\"CRV\", reference:0.39935},{coin:\"TAO\",reference:272.94}];\nconst results = [];\nfor (const target of targets) {\n const res = await fetch(\"https://api.hyperliquid.xyz/info\", {method:\"POST\",headers:{\"Content-Type\":\"application/json\"},body:JSON.stringify({type:\"candleSnapshot\",req:{coin:target.coin,interval:\"4h\",startTime:now-60*86400000,endTime:now}}),signal:AbortSignal.timeout(15000)});\n if (!res.ok) { results.push({coin:target.coin,error:`HTTP_${res.status}`}); if(res.status===429) break; continue; }\n const raw = await res.json();\n if(!Array.isArray(raw)) {results.push({coin:target.coin,error:\"INVALID_RESPONSE\"});continue;}\n const bars=raw.filter(b=>Number(b.T)({t:Number(b.t),h:Number(b.h),l:Number(b.l),c:Number(b.c)})).filter(b=>[b.t,b.h,b.l,b.c].every(Number.isFinite)).sort((a,b)=>a.t-b.t);\n const highs=[],lows=[];\n for(let i=2;ib.h>=x.h)&&neighbours.some(x=>b.h>x.h))highs.push(event); if(neighbours.every(x=>b.l<=x.l)&&neighbours.some(x=>b.lx.price>target.reference).sort((a,b)=>a.price-b.price).slice(0,15),supportCandidates:lows.filter(x=>x.priceb.price-a.price).slice(0,8),lastHighs:highs.slice(-8),note:\"Observed confirmed 4h pivots, not guaranteed future barriers; reference prices are for filtering only, not live quotes.\"});\n await new Promise(r=>setTimeout(r,1000));\n}\nconsole.log(JSON.stringify({readOnly:true,source:\"Hyperliquid official /info candleSnapshot\",fetchedAt:new Date().toISOString(),results}));",[160],"d675935f-7133-4668-8231-52375fe35d1e","realized_pnl_audit_readonly.ts","5da8d5d0-aad3-11f1-acd0-1607ee4eb77e",141,["D",1788796021112],"// Read-only Hyperliquid /info audit. No exchange writes, signing, or schedules.\n// Account-specific results are stored only in val-scoped private blob storage.\nimport { sqlite } from 'https://esm.town/v/std/sqlite/main.ts';\nimport { blob } from 'https://esm.town/v/std/blob/main.ts';\nconst END=Date.now(), OFFSET=9*3600000, DAY=86400000;\nconst user=Deno.env.get('HL_ACCOUNT_ADDRESS'); if(!user) throw new Error('MISSING_ACCOUNT');\nconst n=(v:unknown):number=>{const x=Number(v);if(v===null||v===undefined||!Number.isFinite(x))throw new Error('INVALID_NUMERIC_FIELD');return x;};\nconst rnd=(x:number)=>Math.round(x*1e6)/1e6;\nconst date=(t:number)=>new Date(t+OFFSET).toISOString().slice(0,10);\nasync function info(body:Record){const r=await fetch('https://api.hyperliquid.xyz/info',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify(body),signal:AbortSignal.timeout(15000)});if(!r.ok)throw new Error('HL_HTTP_'+r.status);return await r.json();}\nasync function range(type:string,start:number,end:number,depth=0):Promise{const a=await info({type,user,startTime:start,endTime:end,...(type==='userFillsByTime'?{aggregateByTime:false}:{})});if(!Array.isArray(a))throw new Error('NON_ARRAY');const cap=type==='userFunding'?500:2000;if(a.length=24||end-start<2)throw new Error('PAGINATION_INCOMPLETE');const m=Math.floor((start+end)/2);return [...await range(type,start,m,depth+1),...await range(type,m+1,end,depth+1)];}\nconst meta=Object.fromEntries((await sqlite.execute('SELECT key,value FROM strategy_meta')).rows.map((r:any)=>[String(r.key),String(r.value)]));\nconst effective=meta.effective_from_jst||'2026-09-07', START=Date.parse(effective+'T00:00:00+09:00');\nif(!Number.isFinite(START)||ENDstring)=>[...new Map(a.map(x=>[key(x),x])).values()];\nconst fills=uniq(fr,f=>String(f.tid)).filter(f=>!String(f.coin).startsWith('@')&&/Long|Short/i.test(String(f.dir))).sort((a,b)=>n(a.time)-n(b.time));\nconst funding=uniq(fu,f=>[f.hash,f.time,f.delta?.coin,f.delta?.usdc].join('|')).filter(f=>f.delta?.type==='funding').sort((a,b)=>n(a.time)-n(b.time));\nconst oid=new Set(os.rows.map((r:any)=>String(r.oid))),tid=new Set(ts.rows.map((r:any)=>String(r.tid)));\nconst tagged=fills.filter(f=>oid.has(String(f.oid))||tid.has(String(f.tid)));\nconst seeds=new Map(ss.rows.map((r:any)=>[String(r.coin),{q:n(r.signed_size),t:n(r.at_time)}]));\nfunction active(coin:string,t:number){const seed=seeds.get(coin);let q=seed&&seed.t<=t?seed.q:0;for(const f of tagged){if(n(f.time)>t)break;if(f.coin!==coin||(seed&&n(f.time){throw new Error('UNKNOWN_SIDE')})())*n(f.sz);}return Math.abs(q)>1e-8;}\nconst empty=()=>({gross:0,fees:0,funding:0,net:0});\nconst days:Record={},coins:Record={};\nfor(let t=START;t<=END;t+=DAY)days[date(t)]={account:empty(),strategy:empty()};\nfor(const f of fills){const d=days[date(n(f.time))];if(!d)throw new Error('FILL_OUTSIDE_WINDOW');d.account.gross+=n(f.closedPnl);if(String(f.feeToken||'USDC')==='USDC')d.account.fees+=n(f.fee);else throw new Error('NON_USDC_FEE');}\nfor(const f of tagged){const d=days[date(n(f.time))].strategy;d.gross+=n(f.closedPnl);d.fees+=n(f.fee);const c=coins[f.coin]??=(empty());c.gross+=n(f.closedPnl);c.fees+=n(f.fee);}\nfor(const f of funding){const d=days[date(n(f.time))],v=n(f.delta.usdc);d.account.funding+=v;if(active(String(f.delta.coin),n(f.time))){d.strategy.funding+=v;(coins[f.delta.coin]??=empty()).funding+=v;}}\nlet carry=n(meta.prior_carry_in_usd||'0'), base=n(meta.base_daily_target_usd||'200');\nfor(const d of Object.values(days)){for(const scope of ['account','strategy']){const s=d[scope];s.net=s.gross-s.fees+s.funding;for(const k of Object.keys(s))s[k]=rnd(s[k]);}d.carryInIndependent=rnd(carry);d.requiredIndependent=rnd(base+carry);d.remainingIndependent=rnd(Math.max(0,base+carry-d.strategy.net));carry=d.remainingIndependent;}\nfor(const c of Object.values(coins)){c.net=c.gross-c.fees+c.funding;for(const k of Object.keys(c))c[k]=rnd(c[k]);}\nconst focus=['LINK','HYPE','UNI','ZEC'];\nconst episodes=focus.map(coin=>{const fs=tagged.filter(f=>f.coin===coin);const ent=fs.filter(f=>f.dir==='Open Long'),ex=fs.filter(f=>f.dir==='Close Long');const q=(a:any[])=>a.reduce((s,f)=>s+n(f.sz),0);const vwap=(a:any[])=>a.length?a.reduce((s,f)=>s+n(f.sz)*n(f.px),0)/q(a):null;return{coin,entryQty:rnd(q(ent)),exitQty:rnd(q(ex)),entryVwap:vwap(ent),exitVwap:vwap(ex),entryFirst:ent.length?new Date(n(ent[0].time)).toISOString():null,entryLast:ent.length?new Date(n(ent.at(-1).time)).toISOString():null,exitLast:ex.length?new Date(n(ex.at(-1).time)).toISOString():null,...coins[coin]};});\nconst summary={schemaVersion:'realized-pnl-audit-v1',fetchedAt:new Date(END).toISOString(),effectiveFromJST:effective,dayTotals:days,byCoin:coins,focusEpisodes:episodes,fillCount:fills.length,taggedFillCount:tagged.length,fundingRows:funding.length,unmatched:FillsUnmatched(),warnings:['Funding attribution reproduces presence-based project ledger; mixed discretionary/project size in same symbol requires a separate proportional audit.','Original account_checker.ts uses static prior_carry_in_usd for every date: its post-rollover remainingTodayUSD is not authoritative until repaired/tested.','Independent carry arithmetic is accounting only, not a trading target mandate or recovered cash loss.','No order submissions/amendments/cancellations, no schedules.']};\nfunction FillsUnmatched(){return fills.filter(f=>!oid.has(String(f.oid))&&!tid.has(String(f.tid))).map(f=>({coin:f.coin,dir:f.dir,closedPnl:n(f.closedPnl),fee:n(f.fee),at:new Date(n(f.time)).toISOString()}));}\nconst key='project:hyperliquid:audit:'+new Date(END).toISOString()+':realized';\nawait blob.setJSON(key,{summary,rawFills:fills,rawFunding:funding,registeredOids:os.rows,registeredTids:ts.rows,seeds:ss.rows,meta});\nawait blob.setJSON('project:hyperliquid:audit:latest-realized',{key,summary});\nconsole.log(JSON.stringify({privateAuditKey:key,...summary}));",[167],"abd6765d-5f70-451f-a2df-9de62e605476","pivot216_roleflip_v2_engine.js","9216d3d4-ab24-11f1-8480-1607ee4eb77e",155,["D",1788830898322],"// Pure offline research. No accounts, network, orders, keys or schedules.\nexport const VERSION = 'roleflip-v2.0';\nexport const HOUR=3600000, DAY=24*HOUR;\nexport const DEFAULT={interval:300000,mode:'HYBRID',tick:0.01,body:0.5,stop:0.5,maxChase:1,wait:HOUR,hold:6*HOUR,attemptWindow:6*HOUR,cost:6,minRR:0,cash:false,rthOnly:false,dpLong:false};\nexport const mean=a=>a.length?a.reduce((s,x)=>s+x,0)/a.length:null;\nexport function quantile(a,q){if(!a.length)return null;const b=[...a].sort((x,y)=>x-y),p=(b.length-1)*q,j=Math.floor(p);return b[j]+(b[Math.min(j+1,b.length-1)]-b[j])*(p-j);}\nexport const date=t=>new Date(t).toISOString().slice(0,10);\nexport function periodKey(t,p){const d=new Date(t);return p==='D'?Math.floor(t/DAY):p==='W'?Math.floor((t+3*DAY)/(7*DAY)):d.getUTCFullYear()*12+d.getUTCMonth();}\nexport function traditional(h,l,c){const P=(h+l+c)/3;return {P,R1:2*P-l,S1:2*P-h,R2:P+h-l,S2:P-h+l,R3:2*P+h-2*l,S3:2*P-2*h+l,R4:3*P+h-3*l,S4:3*P-3*h+l,R5:4*P+h-4*l,S5:4*P-4*h+l};}\nexport function atr(b,n=14){let s=0;const a=Array(b.length).fill(NaN);for(let i=0;i=n)a[i]=(a[i-1]*(n-1)+tr)/n;}return a;}\nexport function rng(seed){let x=seed>>>0;return ()=>{x^=x<<13;x^=x>>>17;x^=x<<5;return (x>>>0)/4294967296;};}\nfunction hash(s){let h=2166136261;for(let i=0;i>>0;}\nexport function sourcePeriods(daily){const out={};for(const p of ['D','W','M']){const m=new Map();for(const b of daily){const k=periodKey(b.t,p),old=m.get(k);if(old){old.h=Math.max(old.h,b.h);old.l=Math.min(old.l,b.l);old.c=b.c;}else m.set(k,{key:k,h:b.h,l:b.l,c:b.c});}out[p]=[...m.values()].sort((a,b)=>a.key-b.key);if(p!=='D')out[p].shift();}return out;}\nexport function grid(t,src,seed=0){const a=[];for(const p of ['D','W','M']){const k=periodKey(t,p),prev=src[p].findLast(x=>x.key'R'+(i+1)),...Array.from({length:depth},(_,i)=>'S'+(i+1))])a.push({p:v[name]+shift,label:p+' '+name,period:p,uid:p+':'+prev.key+':'+name});}return a.sort((x,y)=>x.p-y.p);}\nconst nyFmt=new Intl.DateTimeFormat('en-CA',{timeZone:'America/New_York',year:'numeric',month:'2-digit',day:'2-digit',hour:'2-digit',minute:'2-digit',hourCycle:'h23'});\nexport function ny(t){const p={};for(const x of nyFmt.formatToParts(t))p[x.type]=x.value;return {date:p.year+'-'+p.month+'-'+p.day,minute:+p.hour*60+(+p.minute)};}\nexport function cleanCash(b){return b.filter(x=>{const t=ny(x.t);return t.minute>=570&&t.minute<960&&x.v>0;});}\nexport function prepare(bars,daily,cashDaily,cfg={},seed=0){const c={...DEFAULT,...cfg},src=sourcePeriods(daily),cache=new Map(),days=new Set(cashDaily.map(x=>date(x.t))),levels=[],sessions=[];let gaps=0;for(let i=0;iMath.min(b.o,b.c)||b.h=570&&n.minute<960?'RTH':'outside_RTH':'cash_closed');}if(gaps)throw Error('missing execution bars '+gaps);return {bars,levels,sessions,atr:atr(bars),quality:{n:bars.length,gaps,first:bars[0].t,last:bars.at(-1).t}};}\nexport function visible(levels,close){const m=levels.filter(x=>x.period==='M'),lo=m.findLast(x=>x.p<=close),hi=m.find(x=>x.p>=close);return levels.filter(x=>x.period==='M'?x===lo||x===hi:x.period==='D'?['D P','D R1','D R2','D S1','D S2'].includes(x.label):['W P','W R1','W S1'].includes(x.label));}\nconst rounded=(x,t,up)=>Number(((up?Math.ceil(x/t-1e-9):Math.floor(x/t+1e-9))*t).toFixed(8));\nexport function strong(b,prev,L,A,side,c){return side*(prev.c-L)<=0&&side*(b.c-L)>=0.25*A&&side*(b.c-b.o)>=c.body*A&&(side===1?(b.c-b.l)/(b.h-b.l):(b.h-b.c)/(b.h-b.l))>=0.75;}\nexport function updateRejections(st,b,prev,L,A,i,side,c){st.events=st.events.filter(e=>b.t-e.t<=c.attemptWindow);if(side*(b.c-L)<-0.25*A){st.events=[];st.armed=true;return false;}const tol=Math.max(.2*A,2*c.tick),touch=b.l<=L+tol&&b.h>=L-tol,last=st.events.at(-1);if(last&&i>last.i&&!touch&&side*(b.c-L)>=.5*A)st.armed=true;if(!st.events.length)st.armed=true;if(touch&&side*(prev.c-L)>0&&side*(b.c-L)>=Math.max(.1*A,2*c.tick)&&st.armed&&(!last||i-last.i>=2)){st.events.push({t:b.t,i,extreme:side===1?b.l:b.h});st.armed=false;return st.events.length>=2;}return false;}\nexport function planFor(candidate,levels,b,A,i,c,type){const {level:l,side}=candidate,zone=Math.max(.2*A,2*c.tick),target=side===1?levels.find(x=>x.p>l.p+zone):levels.findLast(x=>x.px.extreme)):Math.max(...candidate.events.map(x=>x.extreme));stop=side===1?Math.min(stop,ext-.1*A):Math.max(stop,ext+.1*A);}stop=rounded(stop,c.tick,side===-1);const tp=rounded(target.p,c.tick,side===-1),limit=rounded(l.p+side*.1*A,c.tick,side===-1),ref=type==='RETEST_LIMIT'?limit:b.c;if(side*(ref-stop)<=0||side*(tp-ref)<=ref*c.cost/10000)return null;const rr=(side*(tp-ref)-ref*c.cost/10000)/(side*(ref-stop)+ref*c.cost/10000);if(rr=0,hitSL=s===1?b.l<=pos.stop:b.h>=pos.stop,hitTP=s===1?b.h>=pos.tp:b.l<=pos.tp;if(!intrabarFill&&openSL)return {price:b.o,reason:'SL_GAP',ambiguous:false};if(!intrabarFill&&openTP)return {price:pos.tp,reason:'TP',ambiguous:false};if(hitSL)return {price:pos.stop,reason:'SL',ambiguous:hitTP};if(hitTP&&!intrabarFill)return {price:pos.tp,reason:'TP',ambiguous:false};if(b.t+c.interval>=pos.entryTime+c.hold)return {price:b.c,reason:'TIME',ambiguous:false};return null;}\nexport function simulate(data,config={}){const c={...DEFAULT,...config},b=data.bars,a=data.atr,states=new Map(),trades=[],counts={strongEvents:0,rejectEvents:0,plans:0,limitPlans:0,limitFilled:0,expired:0,targetBeforeFill:0,anchorExpired:0,entryRejected:0,noGeometry:0,cashGap:0};let pending=null,pos=null;const start=c.startTs??b[50].t,end=c.endTs??b.at(-1).t+c.interval,signalCut=end-c.hold-c.wait;\n const finish=(out,j)=>{const q=b[j],risk=pos.side*(pos.entry-pos.stop),g=pos.side*(out.price-pos.entry);trades.push({...pos,exitTime:q.t+c.interval,exit:out.price,outcome:out.reason,ambiguous:out.ambiguous,risk,riskBps:risk/pos.entry*10000,grossR:g/risk,grossBps:g/pos.entry*10000,rr:pos.side*(pos.tp-pos.entry)/risk,holdingHours:(q.t+c.interval-pos.entryTime)/HOUR});pos=null;};\n for(let i=50;il.uid));if(q.t>=end)break;if(!(A>0))continue;const vis=visible(levels,prev.c),breaks=[],fades=[];for(const l of vis){if(c.dpLong&&l.label!=='D P')continue;for(const side of [1,-1]){if(c.dpLong&&side===-1)continue;if(priorIds.has(l.uid)&&strong(q,prev,l.p,A,side,c))breaks.push({level:l,side});const k=l.uid+':'+side;if(!states.has(k))states.set(k,{events:[],armed:true});const st=states.get(k);if(updateRejections(st,q,prev,l.p,A,i,side,c))fades.push({level:l,side,events:[...st.events],state:st});}}\n if(i%100===0){const valid=new Set(levels.flatMap(l=>[l.uid+':1',l.uid+':-1']));for(const k of states.keys())if(!valid.has(k))states.delete(k);}\n let occupiedAtStart=!!pos||!!pending;\n if(pos){const out=exitOnBar(pos,q,c);if(out)finish(out,i);else if(c.cash&&(!b[i+1]||date(b[i+1].t)!==date(q.t))){finish({price:q.c,reason:'SESSION',ambiguous:false},i);} }\n if(pending&&!pos){const p=pending;let cancel=false;if(q.t>=p.expiry&&p.type==='RETEST_LIMIT'){counts.expired++;cancel=true;}else if(!levels.some(l=>l.uid===p.anchorUid)){counts.anchorExpired++;cancel=true;}else if(c.cash&&date(q.t)!==date(p.signalTime)){counts.cashGap++;cancel=true;}else if(c.rthOnly&&data.sessions[i]!=='RTH'){counts.expired++;cancel=true;}\n if(cancel)pending=null;else {let fill=false,entry=NaN,intrabar=false;if(p.type==='RETEST_LIMIT'){const atOpen=p.side*(q.o-p.limit)<=0,penetrates=p.side===1?q.l<=p.limit-c.tick:q.h>=p.limit+c.tick;fill=atOpen||penetrates;entry=p.limit;intrabar=!atOpen;}else{fill=true;entry=q.o;if(p.side*(entry-p.anchor)<0||p.side*(entry-p.anchor)>c.maxChase*p.atr){fill=false;pending=null;counts.entryRejected++;}}\n if(fill){const risk=p.side*(entry-p.stop),reward=p.side*(p.tp-entry),cost=entry*c.cost/10000;const netRR=(reward-cost)/(risk+cost);if(risk<=0||reward<=cost||netRR=p.tp:q.l<=p.tp)){counts.targetBeforeFill++;pending=null;}\n }}\n if(occupiedAtStart||pos||pending||q.tsignalCut||(c.rthOnly&&data.sessions[i]!=='RTH'))continue;\n counts.strongEvents+=breaks.length;counts.rejectEvents+=fades.length;let candidates=[],type='';if(c.mode!=='REJECT_2'&&breaks.length){candidates=breaks;type=c.mode==='RETEST_LIMIT'?'RETEST_LIMIT':'BREAK_NEAR';}else if(['REJECT_2','HYBRID'].includes(c.mode)&&!breaks.length){candidates=fades;type='REJECT_2';}\n candidates.sort((x,y)=>Math.abs(q.c-x.level.p)-Math.abs(q.c-y.level.p)||x.level.uid.localeCompare(y.level.uid));if(candidates.length){const pick=candidates[0];if(c.mode==='HYBRID'&&type==='BREAK_NEAR'&&Math.abs(q.c-pick.level.p)>c.maxChase*A)type='RETEST_LIMIT';if(type!=='RETEST_LIMIT'&&Math.abs(q.c-pick.level.p)>c.maxChase*A){counts.entryRejected++;continue;}const plan=planFor(pick,levels,q,A,i,c,type);if(!plan){counts.noGeometry++;continue;}pending=plan;counts.plans++;if(type==='RETEST_LIMIT')counts.limitPlans++;if(pick.state){pick.state.events=[];pick.state.armed=false;}}\n }\n return {config:c,counts,trades,openAtEnd:!!pos,pendingAtEnd:!!pending,signalStart:start,signalCutoff:signalCut};}\nexport function stats(trades,cost=6,bootstrap=false){const r=trades.map(t=>t.grossR-cost/t.riskBps),b=trades.map(t=>t.grossBps-cost),pf=x=>{const neg=x.filter(z=>z<0).reduce((s,z)=>s-z,0);return neg?x.filter(z=>z>0).reduce((s,z)=>s+z,0)/neg:null;};let total=0,peak=0,dd=0;for(const x of r){total+=x;peak=Math.max(peak,total);dd=Math.max(dd,peak-total);}const out={n:trades.length,tp:trades.filter(t=>t.outcome==='TP').length,sl:trades.filter(t=>t.outcome.startsWith('SL')).length,time:trades.filter(t=>['TIME','SESSION'].includes(t.outcome)).length,winPct:trades.length?r.filter(x=>x>0).length/trades.length*100:null,meanR:mean(r),meanBps:mean(b),grossR:mean(trades.map(t=>t.grossR)),grossBps:mean(trades.map(t=>t.grossBps)),pfR:pf(r),pfBps:pf(b),totalR:total,maxDrawdownR:dd,medianRR:quantile(trades.map(t=>t.rr),.5),medianRiskBps:quantile(trades.map(t=>t.riskBps),.5),holdingHours:mean(trades.map(t=>t.holdingHours)),ambiguous:trades.filter(t=>t.ambiguous).length};if(bootstrap&&trades.length){const groups=new Map();trades.forEach((t,i)=>{const k=periodKey(t.entryTime,'W');if(!groups.has(k))groups.set(k,[]);groups.get(k).push(r[i]);});const g=[...groups.values()],rand=rng(216260908),samples=[];for(let j=0;j<1000;j++){let sum=0,n=0;for(let k=0;ks+x,0);n+=v.length;}samples.push(sum/n);}out.weeks=g.length;out.ci95=[quantile(samples,.025),quantile(samples,.975)];}return out;}\nexport function summarize(run,data){const tr=run.trades,c=run.config,start=c.startTs??data.bars[0].t,end=c.endTs??data.bars.at(-1).t+c.interval,split=start+.7*(end-start),before=tr.filter(t=>t.exitTimet.entryTime>=split),breakdown={};for(const [dim,values] of [['session',['RTH','outside_RTH','cash_closed']],['period',['D','W','M']],['type',['BREAK_NEAR','RETEST_LIMIT','REJECT_2']],['side',[1,-1]]]){breakdown[dim]={};for(const v of values)breakdown[dim][v]=stats(tr.filter(t=>t[dim]===v),c.cost,true);}return {version:VERSION,quality:data.quality,config:c,counts:run.counts,signalStart:run.signalStart,signalCutoff:run.signalCutoff,split,splitCrossers:tr.length-before.length-after.length,overall:stats(tr,c.cost,true),early:stats(before,c.cost,true),late:stats(after,c.cost,true),costs:[0,6,12,20].map(cost=>({cost,...stats(tr,cost)})),breakdown,trades:tr,fundingIncluded:false,markRecreated:false,openAtEnd:run.openAtEnd};}\nexport function selfTests(){const pass=[],assert=(x,name)=>{if(!x)throw Error(name);pass.push(name);};const p=traditional(110,90,100);assert(p.P===100&&p.R1===110&&p.S1===90&&p.R5===150&&p.S5===50,'Traditional formulas');assert(periodKey(Date.parse('2025-12-29'),'W')===periodKey(Date.parse('2026-01-02'),'W'),'Monday week year boundary');const d=[{t:Date.parse('2026-01-01'),h:110,l:90,c:100},{t:Date.parse('2026-01-02'),h:999,l:1,c:500}];assert(grid(Date.parse('2026-01-02T12:00Z'),sourcePeriods(d)).find(x=>x.label==='D P').p===100,'No current-period OHLC leak');const b=Array.from({length:30},(_,i)=>({t:i*300000,o:100,h:101,l:99,c:100}));assert(atr(b)[13]===2&&atr(b)[29]===2,'Wilder ATR');const c={...DEFAULT};assert(strong({o:99.5,h:101,l:99.4,c:100.9},{c:99.8},100,2,1,c),'Strong upward cross LONG');assert(!strong({o:99.9,h:101,l:99.8,c:100},{c:99.8},100,2,1,c),'Wick-only crossing not strong');assert(strong({o:100.5,h:100.6,l:99,c:99.1},{c:100.2},100,2,-1,c),'Strong downward cross SHORT');const pos={side:1,entryTime:0,entry:100,stop:99,tp:102};assert(exitOnBar(pos,{t:300000,o:100,h:103,l:98,c:101},c).reason==='SL','Same-bar TP+SL stop-first');assert(exitOnBar(pos,{t:300000,o:98,h:100,l:97,c:99},c).price===98,'Adverse gap actual open');assert(exitOnBar(pos,{t:300000,o:100.5,h:103,l:99.5,c:101},c,true)===null,'No TP credit on intrabar limit-entry bar');const st={events:[],armed:true},prev={c:99};const touch=(t)=>({t,o:99,h:100.1,l:98.9,c:99.2});assert(!updateRejections(st,touch(0),prev,100,1,0,-1,c),'First rejection not fade');assert(!updateRejections(st,touch(300000),prev,100,1,1,-1,c),'Continuous touch not second rejection');updateRejections(st,{t:600000,o:99.2,h:99.4,l:99,c:99.1},prev,100,1,2,-1,c);assert(updateRejections(st,touch(900000),prev,100,1,3,-1,c),'Separated second rejection qualifies');return pass;}\n",[174],"5e22acf1-320f-491d-b6d8-1ac084ae0f83","pivot216_research_run.ts","90c31e7c-ab1d-11f1-b23d-1607ee4eb77e",149,["D",1788827982423],2,"import { blob } from \"https://esm.town/v/std/blob/main.ts\";\nimport { analyze, cashClean, selfTest } from './pivot216_research_engine.ts';\nconst P='research:pivot216:NVDA:20260908';\nconst cashDaily:any[]=await blob.getJSON(P+':bars:cash_NVDA_1d');\nconst xyzDaily:any[]=await blob.getJSON(P+':bars:xyz_NVDA_1d');\nconst config={tf:'1h',interval:3600000,horizon:6,includeCash:false};\nconst bars:any[]=await blob.getJSON(P+':bars:xyz_NVDA_'+config.tf);\nconst compact:any[]=[];\nfor(const [source,daily] of [['CASH_RTH',cashDaily],['XYZ_UTC',xyzDaily]] as const){for(const rejection of [false,true]){\n const id='XYZ_'+config.tf+'_'+source+'_'+(rejection?'REJECT':'TOUCH');\n const r=analyze(bars,daily,cashDaily,{interval:config.interval,horizon:config.horizon,rejection,cash:false,tick:0.01});\n await blob.setJSON(P+':result:'+id,r);\n compact.push({id,quality:r.quality,splitAt:r.splitAt,...r.summaries.overall,earlyR:r.summaries.early.meanNetR,earlyN:r.summaries.early.n,lateR:r.summaries.late.meanNetR,lateN:r.summaries.late.n,lateCI:r.summaries.late.meanNetR95,placeboMeanR:r.placebo.meanNetR,placeboPct:r.placebo.actualPercentile,censored:r.censored,missingEntry:r.missingEntry});\n}}\nif(config.includeCash){const cb=cashClean(await blob.getJSON(P+':bars:cash_NVDA_5m'));for(const rejection of [false,true]){const id='CASH_5m_CASH_RTH_'+(rejection?'REJECT':'TOUCH');const r=analyze(cb,cashDaily,cashDaily,{interval:300000,horizon:12,rejection,cash:true,tick:0.01});await blob.setJSON(P+':result:'+id,r);compact.push({id,quality:r.quality,splitAt:r.splitAt,...r.summaries.overall,earlyR:r.summaries.early.meanNetR,earlyN:r.summaries.early.n,lateR:r.summaries.late.meanNetR,lateN:r.summaries.late.n,lateCI:r.summaries.late.meanNetR95,placeboMeanR:r.placebo.meanNetR,placeboPct:r.placebo.actualPercentile,censored:r.censored,missingEntry:r.missingEntry});}}\nawait blob.setJSON(P+':summary:'+config.tf,compact);\nconsole.log(JSON.stringify({selfTests:selfTest(),fundingIncluded:false,rows:compact}));",[181],"a3e39e71-af4c-487f-be4f-70e5186a9a2f","pivot216_research_engine.ts","85e6d11a-ab1d-11f1-955b-1607ee4eb77e",146,["D",1788827871383],"// Pure research mathematics. Reads no account state; makes no network/exchange calls.\nexport type Bar={t:number,o:number,h:number,l:number,c:number,v:number,T?:number};\nexport type Level={p:number,name:string,period:string,key:number,range:number};\nexport type Trade={signal:number,entryTime:number,exitTime:number,side:number,level:number,label:string,period:string,session:string,entry:number,exit:number,atr:number,atrBps:number,grossR:number,grossBps:number,outcome:string,ambiguous:boolean,signalClose:number};\nexport const DAY=86400000;\nconst nyFormatter=new Intl.DateTimeFormat('en-CA',{timeZone:'America/New_York',year:'numeric',month:'2-digit',day:'2-digit',hour:'2-digit',minute:'2-digit',hourCycle:'h23'});\nexport function ny(t:number){const p:any={};for(const a of nyFormatter.formatToParts(t))p[a.type]=a.value;return {date:p.year+'-'+p.month+'-'+p.day,minute:+p.hour*60+(+p.minute)};}\nexport function utcDate(t:number){return new Date(t).toISOString().slice(0,10);}\nexport function key(t:number,period:string){const d=new Date(t);return period==='D'?Math.floor(t/DAY):period==='W'?Math.floor((t+3*DAY)/(7*DAY)):d.getUTCFullYear()*12+d.getUTCMonth();}\nexport function traditional(h:number,l:number,c:number){const p=(h+l+c)/3;return {P:p,R1:2*p-l,S1:2*p-h,R2:p+h-l,S2:p-h+l,R3:2*p+h-2*l,S3:2*p-2*h+l,R4:3*p+h-3*l,S4:3*p-3*h+l,R5:4*p+h-4*l,S5:4*p-4*h+l};}\nfunction random(seed:number){let x=seed>>>0;return ()=>{x^=x<<13;x^=x>>>17;x^=x<<5;return (x>>>0)/4294967296;};}\nfunction hash(s:string){let h=2166136261;for(let i=0;i>>0;}\nexport function validate(bars:Bar[],interval:number,cash=false){let duplicate=0,bad=0,gaps=0;for(let i=0;iMath.min(b.o,b.c)||b.hb.h)bad++;if(i&&b.t<=bars[i-1].t)duplicate++;if(i&&b.t-bars[i-1].t>interval*1.5&&(!cash||utcDate(b.t)===utcDate(bars[i-1].t)))gaps++;}if(bad||duplicate)throw Error('Invalid OHLC/order: '+JSON.stringify({bad,duplicate}));return {n:bars.length,first:new Date(bars[0].t).toISOString(),last:new Date(bars.at(-1)!.t).toISOString(),bad,duplicate,gaps};}\nexport function cashClean(bars:Bar[]){return bars.filter(b=>{const x=ny(b.t);return x.minute>=570&&x.minute<960&&b.v>0;});}\nexport function atr(bars:Bar[],length=14){const a=Array(bars.length).fill(NaN);let sum=0;for(let i=0;i=length)a[i]=(a[i-1]*(length-1)+tr)/length;}return a;}\nexport function makeSource(daily:Bar[]){const out:any={};for(const period of ['D','W','M']){const m=new Map();for(const b of daily){const k=key(b.t,period),v=m.get(k);if(v){v.h=Math.max(v.h,b.h);v.l=Math.min(v.l,b.l);v.c=b.c;v.last=b.t;v.n++;}else m.set(k,{key:k,h:b.h,l:b.l,c:b.c,first:b.t,last:b.t,n:1});}let a=[...m.values()].sort((a,b)=>a.key-b.key);if(period!=='D')a=a.slice(1);out[period]=a;}return out;}\nfunction levelsFor(t:number,source:any,seed:number){const out:Level[]=[];for(const period of ['D','W','M']){const k=key(t,period),a=source[period];let prev:any;for(let i=a.length-1;i>=0;i--)if(a[i].key'R'+(i+1)),...Array.from({length:depth},(_,i)=>'S'+(i+1))])out.push({p:p[name]+shift,name:period+' '+name,period,key:prev.key,range:prev.h-prev.l});}return out;}\nexport function prepare(bars:Bar[],source:any,cashDays:Set,seed=0){const levels:Level[][]=[],sessions:string[]=[],cache=new Map();for(const b of bars){const d=utcDate(b.t);if(!cache.has(d))cache.set(d,levelsFor(b.t,source,seed));levels.push(cache.get(d)!);const n=ny(b.t);sessions.push(cashDays.has(n.date)?(n.minute>=570&&n.minute<960?'RTH':'outside_RTH'):'cash_closed');}return {bars,levels,sessions,atr:atr(bars)};}\nfunction visible(levels:Level[],lastClose:number){const m=levels.filter(x=>x.period==='M');let below:Level|undefined,above:Level|undefined;for(const x of m){if(x.p<=lastClose&&(!below||x.p>below.p))below=x;if(x.p>=lastClose&&(!above||x.px.period!=='M'||x===below||x===above);}\nexport function simulate(data:any,options:{interval:number,horizon:number,rejection:boolean,cash:boolean,tick?:number}){const {bars,levels,sessions,atr:a}=data;const trades:Trade[]=[];let i=50,censored=0,signalCount=0,missingEntry=0;const tick=options.tick??0.01;while(i0)){i++;continue;}const buf=Math.max(A*0.25,2*tick),back=Math.max(A*0.10,2*tick);const candidates:any[]=[];for(const l of visible(levels[i],prev.c)){const tag=l.name.split(' ')[1];const side=tag==='P'?(prev.c>l.p?1:-1):tag.startsWith('S')?1:-1;if(side*(prev.c-l.p)<=buf)continue;if(b.l>l.p+buf||b.hx.d-y.d);const l=candidates[0];signalCount++;const start=i+1,nb:Bar=bars[start];if(nb.t-b.t>options.interval*1.5||(options.cash&&utcDate(nb.t)!==utcDate(b.t))){missingEntry++;i++;continue;}const entry=nb.o,sl=entry-l.side*A,tp=entry+l.side*A;let exit=NaN,outcome='TIME',ambiguous=false,j=start,bad=false;const last=Math.min(start+options.horizon-1,bars.length-1);for(;j<=last;j++){const q:Bar=bars[j];if(j>start&&options.cash&&utcDate(q.t)!==utcDate(nb.t)){j--;exit=bars[j].c;outcome='SESSION';break;}if(j>start&&q.t-bars[j-1].t>options.interval*1.5){bad=true;break;}const openSL=l.side===1?q.o<=sl:q.o>=sl,openTP=l.side===1?q.o>=tp:q.o<=tp,hitSL=l.side===1?q.l<=sl:q.h>=sl,hitTP=l.side===1?q.h>=tp:q.l<=tp;if(openSL){exit=q.o;outcome='SL_GAP';break;}if(openTP){exit=tp;outcome='TP';break;}if(hitSL){exit=sl;outcome='SL';ambiguous=hitTP;break;}if(hitTP){exit=tp;outcome='TP';break;}if(j===last){exit=q.c;break;}}\nif(bad||!Number.isFinite(exit)){censored++;i=Math.max(i+1,j+1);continue;}const gr=l.side*(exit-entry)/A,gb=l.side*(exit/entry-1)*10000;trades.push({signal:b.t,entryTime:nb.t,exitTime:bars[j].t+options.interval,side:l.side,level:l.p,label:l.name,period:l.period,session:sessions[i],entry,exit,atr:A,atrBps:A/entry*10000,grossR:gr,grossBps:gb,outcome,ambiguous,signalClose:b.c});i=j+1;}\nreturn {trades,signalCount,censored,missingEntry};}\nconst mean=(a:number[])=>a.length?a.reduce((s,x)=>s+x,0)/a.length:null;\nfunction quantile(a:number[],q:number){if(!a.length)return null;const b=[...a].sort((x,y)=>x-y),x=(b.length-1)*q,j=Math.floor(x);return b[j]+(b[Math.min(j+1,b.length-1)]-b[j])*(x-j);}\nexport function stats(trades:Trade[],costBps=6,ci=false){const r=trades.map(t=>t.grossR-costBps/t.atrBps),b=trades.map(t=>t.grossBps-costBps);let sum=0,peak=0,dd=0,profit=0,loss=0;for(const x of r){sum+=x;peak=Math.max(peak,sum);dd=Math.max(dd,peak-sum);if(x>0)profit+=x;else loss-=x;}const out:any={n:trades.length,tp:trades.filter(t=>t.outcome==='TP').length,sl:trades.filter(t=>t.outcome.startsWith('SL')).length,timeout:trades.filter(t=>['TIME','SESSION'].includes(t.outcome)).length,ambiguous:trades.filter(t=>t.ambiguous).length,netWinPct:trades.length?r.filter(x=>x>0).length/trades.length*100:null,meanGrossR:mean(trades.map(t=>t.grossR)),meanNetR:mean(r),meanNetBps:mean(b),breakEvenCostBps:mean(trades.map(t=>t.grossBps)),pfR:loss?profit/loss:null,maxDrawdownR:dd,totalR:sum};if(ci&&trades.length){const m=new Map();trades.forEach((t,i)=>{const k=key(t.entryTime,'W');if(!m.has(k))m.set(k,[]);m.get(k)!.push(r[i]);});const groups=[...m.values()],rng=random(216202609),av:number[]=[];for(let j=0;j<1000;j++){let s=0,n=0;for(let k=0;kutcDate(b.t))),data=prepare(bars,src,days,0);const real=simulate(data,options),split=bars[0].t+(bars.at(-1)!.t-bars[0].t)*0.7;const before=real.trades.filter(t=>t.exitTimet.entryTime>=split);const controls:any[]=[];for(let seed=1;seed<=20;seed++){const sim=simulate(prepare(bars,src,days,seed),options);const s=stats(sim.trades);controls.push({seed,n:s.n,meanNetR:s.meanNetR,meanNetBps:s.meanNetBps,netWinPct:s.netWinPct});}const ctrl=controls.map(x=>x.meanNetR).filter((x:any)=>x!==null),overall=stats(real.trades,6,true);const summaries:any={overall,early:stats(before,6,true),late:stats(after,6,true),sensitivity:[0,6,12,20].map(cost=>({roundTripCostBps:cost,...stats(real.trades,cost)})),byPeriod:{},byDirection:{},bySession:{}};for(const x of ['D','W','M'])summaries.byPeriod[x]=stats(real.trades.filter(t=>t.period===x),6,true);for(const x of [-1,1])summaries.byDirection[x===1?'LONG':'SHORT']=stats(real.trades.filter(t=>t.side===x),6,true);for(const x of ['RTH','outside_RTH','cash_closed'])summaries.bySession[x]=stats(real.trades.filter(t=>t.session===x),6,true);return {version:'pivot216-research-1.0',fundingIncluded:false,markTriggersRecreated:false,originalRsiStrategyRecreated:false,quality,options,splitAt:new Date(split).toISOString(),splitCrossers:real.trades.length-before.length-after.length,signalCount:real.signalCount,censored:real.censored,missingEntry:real.missingEntry,summaries,placebo:{count:controls.length,meanNetR:mean(ctrl),p05:quantile(ctrl,0.05),p95:quantile(ctrl,0.95),actualPercentile:ctrl.length?ctrl.filter(x=>x({t:i*300000,o:100,h:101,l:99,c:100,v:1})));if(a[13]!==2||a[29]!==2)throw Error('ATR test');if(key(Date.parse('2025-12-29'),'W')!==key(Date.parse('2026-01-02'),'W'))throw Error('week key');const daily=[{t:Date.parse('2026-01-01'),o:100,h:110,l:90,c:100,v:1},{t:Date.parse('2026-01-02'),o:100,h:999,l:1,c:500,v:1}];const levels=levelsFor(Date.parse('2026-01-02T12:00Z'),makeSource(daily),0);if(levels.find(x=>x.name==='D P')?.p!==100)throw Error('lookahead leak');return ['Traditional formulas','Wilder ATR','year-crossing week key','no same-period OHLC'];}\n",[189],"123492b4-0682-4796-9adf-45b0e27dc149","pivot216_research_diagnostics.ts","028bb172-ab1e-11f1-98f9-1607ee4eb77e",150,["D",1788828080514],"import { blob } from 'https://esm.town/v/std/blob/main.ts';\nimport { stats,simulate,utcDate } from './pivot216_research_engine.ts';\nconst P='research:pivot216:NVDA:20260908';\nconst summary:any[]=[];for(const tf of ['5m','15m','1h'])summary.push(...await blob.getJSON(P+':summary:'+tf));\nconst compact=(x:any)=>({n:x.n,win:x.netWinPct,r:x.meanNetR,grossR:x.meanGrossR,bps:x.meanNetBps,breakEvenBps:x.breakEvenCostBps,pf:x.pfR});\nconst rows:any[]=[];\nfor(const s of summary){const r:any=await blob.getJSON(P+':result:'+s.id);const split=Date.parse(r.splitAt);const reaction=r.trades.filter((t:any)=>t.side*(t.signalClose-t.level)>=Math.max(t.atr*.1,.02)).length;\n rows.push({id:s.id,RTH:compact(r.summaries.bySession.RTH),outside:compact(stats(r.trades.filter((t:any)=>t.session!=='RTH'))),D:compact(r.summaries.byPeriod.D),W:compact(r.summaries.byPeriod.W),M:compact(r.summaries.byPeriod.M),long:compact(r.summaries.byDirection.LONG),short:compact(r.summaries.byDirection.SHORT),RTHlate:compact(stats(r.trades.filter((t:any)=>t.session==='RTH'&&t.entryTime>=split))),immediateBackPct:reaction/r.trades.length*100});\n}\nawait blob.setJSON(P+':diagnostics',rows);\nconst c5:any[]=await blob.getJSON(P+':bars:cash_NVDA_5m'),x5:any[]=await blob.getJSON(P+':bars:xyz_NVDA_5m'),x15:any[]=await blob.getJSON(P+':bars:xyz_NVDA_15m');\nconst bytime=new Map(x5.map(b=>[b.t,b]));const diff=c5.filter(b=>b.v>0&&bytime.has(b.t)).map(b=>Math.abs((bytime.get(b.t)!.c/b.c-1)*10000)).sort((a,b)=>a-b);\nlet matches=0,maxCloseDiff=0,maxHighDiff=0;for(const b of x15){const a=[0,1,2].map(j=>bytime.get(b.t+j*300000));if(a.every(Boolean)){matches++;maxCloseDiff=Math.max(maxCloseDiff,Math.abs(a[2]!.c-b.c));maxHighDiff=Math.max(maxHighDiff,Math.abs(Math.max(...a.map(x=>x!.h))-b.h));}}\nconst bs=Array.from({length:55},(_,i)=>({t:Date.parse('2026-08-24T00:00Z')+i*300000,o:103,h:103,l:103,c:103,v:1}));bs[50]={...bs[50],o:103,h:103,l:100,c:102};bs[51]={...bs[51],o:102,h:105,l:99,c:102};const levels=bs.map((_,i)=>i===50?[{p:100,name:'D S1',period:'D',key:1,range:10}]:[]);const test=simulate({bars:bs,levels,sessions:bs.map(()=> 'RTH'),atr:bs.map(()=>2)},{interval:300000,horizon:2,rejection:false,cash:false});if(test.trades.length!==1||test.trades[0].outcome!=='SL'||!test.trades[0].ambiguous||test.trades[0].entry!==102)throw Error('next-open/stop-first fixture failed');\nconst qa={alignment:{matchedCashXYZ5m:diff.length,medianAbsoluteBasisBps:diff[Math.floor(diff.length*.5)],p95AbsoluteBasisBps:diff[Math.floor(diff.length*.95)],maxAbsoluteBasisBps:diff.at(-1),aggregated5mVs15mBars:matches,maxCloseDifference:maxCloseDiff,maxHighDifference:maxHighDiff},executionTests:['next-bar open, not signal-bar price','same-bar TP+SL resolves to SL'],fundingIncluded:false};await blob.setJSON(P+':quality_checks',qa);\nconsole.log(JSON.stringify({qa,rows:rows.map(r=>({id:r.id,RTH:r.RTH,outside:r.outside,RTHlate:r.RTHlate,immediateBackPct:r.immediateBackPct}))}));",[196],"0ff43c94-8746-4073-9ed4-d1d10dd48353","pivot216_data_collect.ts","b9209cb0-ab1c-11f1-8233-1607ee4eb77e",144,["D",1788827527840],"import { blob } from \"https://esm.town/v/std/blob/main.ts\";\n// Research-only collector. No account data, secrets, exchange writes, or schedules.\nexport const PREFIX = \"research:pivot216:NVDA:20260908\";\nexport async function collect() {\n const end=Date.now(), out:any={asOf:new Date(end).toISOString(),datasets:[],errors:[]};\n const jobs=[...['5m','15m','1h','1d'].map(interval=>({kind:'hyperliquid',name:'xyz_NVDA_'+interval,interval})),...['5m','1d'].map(interval=>({kind:'yahoo',name:'cash_NVDA_'+interval,interval}))];\n for(const j of jobs){\n try{\n let raw:any,url='',request:any;\n if(j.kind==='hyperliquid'){\n const ms:any={'5m':300000,'15m':900000,'1h':3600000,'1d':86400000};\n url='https://api.hyperliquid.xyz/info'; request={type:'candleSnapshot',req:{coin:'xyz:NVDA',interval:j.interval,startTime:end-ms[j.interval]*5050,endTime:end}};\n const r=await fetch(url,{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify(request),signal:AbortSignal.timeout(16000)});if(!r.ok)throw Error('HTTP '+r.status+' '+await r.text());raw=await r.json();\n }else{\n url='https://query1.finance.yahoo.com/v8/finance/chart/NVDA?range='+ (j.interval==='1d'?'2y':'60d')+'&interval='+j.interval+'&includePrePost=false';\n const r=await fetch(url,{headers:{'User-Agent':'Mozilla/5.0'},signal:AbortSignal.timeout(16000)});if(!r.ok)throw Error('HTTP '+r.status+' '+(await r.text()).slice(0,150));raw=await r.json();\n }\n const text=JSON.stringify(raw);const sha=Array.from(new Uint8Array(await crypto.subtle.digest('SHA-256',new TextEncoder().encode(text)))).map(x=>x.toString(16).padStart(2,'0')).join('');\n await blob.setJSON(PREFIX+':raw:'+j.name,{fetchedAt:new Date().toISOString(),url,request,sha256:sha,raw});\n let rows:any[]=[];\n if(j.kind==='hyperliquid') rows=raw.map((x:any)=>({t:x.t,T:x.T,o:+x.o,h:+x.h,l:+x.l,c:+x.c,v:+x.v,n:x.n}));\n else {const q=raw.chart?.result?.[0];if(!q)throw Error(JSON.stringify(raw));const a=q.indicators.quote[0];rows=q.timestamp.map((t:number,i:number)=>({t:t*1000,o:a.open[i],h:a.high[i],l:a.low[i],c:a.close[i],v:a.volume[i]})).filter((x:any)=>Number.isFinite(x.c));out[j.name+'_meta']={currency:q.meta.currency,exchangeTimezoneName:q.meta.exchangeTimezoneName,exchangeName:q.meta.exchangeName};}\n const ms:any={'5m':300000,'15m':900000,'1h':3600000,'1d':86400000};\n const closed=rows.filter(x=>x.t+ms[j.interval]<=end && [x.o,x.h,x.l,x.c].every(Number.isFinite));\n await blob.setJSON(PREFIX+':bars:'+j.name,closed);\n out.datasets.push({name:j.name,count:closed.length,first:new Date(closed[0]?.t).toISOString(),last:new Date(closed.at(-1)?.t).toISOString(),sha256:sha,firstBar:closed[0],lastBar:closed.at(-1)});\n }catch(e){out.errors.push({name:j.name,error:String(e)});}\n }\n await blob.setJSON(PREFIX+':manifest',out);return out;\n}\nconsole.log(JSON.stringify(await collect()));",[203],"263cfe15-f1f7-4f0a-9ef5-2d2c9d93ad40","persist_audit_only_handoff.ts","d69a5536-aad3-11f1-9061-1607ee4eb77e",143,["D",1788796224033],"// Persist this requested audit boundary. No exchange writes, no trading, no schedules.\nimport { blob } from 'https://esm.town/v/std/blob/main.ts';\nconst currentKey='project:hyperliquid:authoritative-handoff:current';\nconst realized:any=await blob.getJSON('project:hyperliquid:audit:latest-realized');\nconst paths:any=await blob.getJSON('project:hyperliquid:audit:latest-paths');\nif(!realized?.summary||!paths?.summary)throw new Error('AUDIT_MISSING');\nconst user=Deno.env.get('HL_ACCOUNT_ADDRESS');if(!user)throw new Error('MISSING_ACCOUNT');\nasync function info(type:string,extra:Record={}){const r=await fetch('https://api.hyperliquid.xyz/info',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({type,user,...extra}),signal:AbortSignal.timeout(10000)});if(!r.ok)throw new Error('HL_HTTP_'+r.status);return await r.json();}\nconst [main,xyz,orders,xyzOrders,spot]=await Promise.all([info('clearinghouseState'),info('clearinghouseState',{dex:'xyz'}),info('frontendOpenOrders'),info('frontendOpenOrders',{dex:'xyz'}),info('spotClearinghouseState')]);\nif(!Array.isArray(main.assetPositions)||!Array.isArray(xyz.assetPositions)||!Array.isArray(orders)||!Array.isArray(xyzOrders)||!Array.isArray(spot.balances))throw new Error('BAD_ACCOUNT_DATA');\nconst positions=[...main.assetPositions,...xyz.assetPositions].map(p=>p.position).filter(p=>Number(p.szi)!==0),openOrders=[...orders,...xyzOrders];\nif(positions.length||openOrders.length)throw new Error('STATE_CHANGED_RECHECK_REQUIRED');\nconst usdc=spot.balances.find((b:any)=>b.coin==='USDC');if(!usdc||!Number.isFinite(Number(usdc.total))||!Number.isFinite(Number(usdc.hold)))throw new Error('BAD_BALANCE');\nconst now=new Date().toISOString(),archive='project:hyperliquid:authoritative-handoff:archive:'+now;\nawait blob.copy(currentKey,archive);\nconst updated={schemaVersion:'hyperliquid-project-handoff-v22-audit-only',asOf:now,priority:'USER PAUSED TRADE INSTRUCTIONS. Audit/research only unless user explicitly asks to resume. Current exchange facts always override stored snapshots.',userRequest:{text:'Linkも切りました。一旦指示はいらないので、どうしたら勝てるか一晩考えて',noNewTradeInstructions:true,focus:'Improve symbol, direction, entry, SL/TP path and portfolio expectancy; smaller size, more symbols or forced hedges are not substitutes for edge.',backgroundWork:false,overnightThinkingOrAutotradingPromised:false},objective:{dailyBaseNetUSD:200,carryPolicyUnchanged:true,profitOrRecoveryGuaranteed:false,newPermanentNumericRiskCaps:false},accountObservations:{fetchedAt:now,scope:'core and xyz main project account only; no claim about separate bot subaccount',positions,openOrders,usdcTotal:Number(usdc.total),usdcHold:Number(usdc.hold),availableBalanceUSDC:Number(usdc.total)-Number(usdc.hold),linkClose:{at:'2026-09-07T15:42:10.902Z',qty:100,price:12.884,closedPnlUSD:-36,exitFeeUSD:0.521801,manualCloseAttributedToProject:true},dayTotals:realized.summary.dayTotals},audit:{realizedBlob:realized.key,pathBlob:paths.key,last4TradeResults:paths.summary.trades.map((t:any)=>({coin:t.coin,entryVwap:t.entryVwap,exitVwap:t.exitVwap,qty:t.qty,netUSD:t.net,mfeGrossTradeCandleEstimate:t.mfeGrossInner,entryAt:t.entryStart,exitAt:t.exitAt})),findings:['Final HYPE, UNI and ZEC exits match pre-registered SL OIDs, not evidence of Cross liquidation.','HYPE/UNI/ZEC fully-filled holding-period trade-candle MFE was only about8.855/8.78827/3.92 USD. TP-only changes do not demonstrate corrected edge.','At22:00 observation UNI lower highs7.4224->7.3188->7.1914 and ZEC1253.9->1222.3->1212 conflict with treating EMA-UP as unqualified continuation.','UNI SL6.98 above recorded1h support6.965 and ZEC SL1175 above recorded1170.1. They might define short-horizon stops, but were not justified as overnight structural invalidation.','UNI TP7.30 requires passing7.156 and7.1914; ZEC TP1216 passes1212. These path hurdles were insufficiently modeled.','Indicators across nested timeframes are not independent votes. Closed4h EMA context must not suppress newer lower-timeframe reversal evidence.','All-target profit is not expected profit. No calibrated win rate, first-passage probability or profitable out-of-sample strategy has been demonstrated here.','One losing session does not prove all rules have negative expectancy, and reverse-all-trades/widen-all-stops is not a validated repair.','Saying safe to sleep on existence of SL/TP alone overstated what was verified. The read-only bridge is not a 24/7 execution or discretionary exit agent.'],dataCaveats:['1m candles are trade-price extrema, not historical Mark series or executable fills. Boundaries and queue effects matter.','Funding attribution follows existing tag-presence accounting; same-symbol mixed manual/project exposure needs separate proportional audit.','Trade-path audit concerns last completed long cycle for each ofLINK/HYPE/UNI/ZEC, not the entire day for HYPE/ZEC.'],accountingDefects:[{status:'LINK_RESOLVED',detail:'3 manual-close TIDs registered in strategy_tids; exit loss must count against project, not be excluded as discretionary.'},{status:'ORIGINAL_CHECKER_UNREPAIRED',detail:'account_checker.ts requiredToday=baseTarget+static priorCarry; no per-day rollover. September8 remaining target output is wrong. Independent day-by-day arithmetic is in audit; do not trust checker remainingToday until a tested code repair.'}],validationWorkRemaining:['Freeze a small set of clearly defined setups before evaluation; compare trend-continuation, range/reversal, relative-value and no-trade baselines.','Evaluate full active universe including unselected/non-filled candidates, execution delays, spread, fees, funding, Mark triggers and realistic partial fills.','Separate audited-session hypotheses from held-out chronological validation; count all tested variants and account for selection bias.','Evaluate expected net return plus uncertainty, net TP-before-SL probability within intended horizon, MAE/MFE, target near misses and joint portfolio outcomes.','Unattended proposals need validated exits that require no nonexistent monitoring, discretionary cancellation or dynamic stop adjustment.']},priorArchiveKey:archive,allPriorUnfilledPlansObsolete:true,excludedSymbol:'xyz:JPY unless user expressly re-enables',nextStep:'Continue from evidence-backed audit when user returns. Do not emit a new trade recommendation automatically. Do not claim an overnight study, profitable model, daily target achievement, or implemented strategy repair.'};\nawait blob.setJSON(currentKey,updated);await blob.setJSON('project:hyperliquid:audit:current-research-handoff',updated);\nconsole.log(JSON.stringify({saved:true,key:currentKey,archivedPreviousKey:archive,fetchedAt:now,positions:positions.length,openOrders:openOrders.length,usdcTotal:Number(usdc.total),usdcHold:Number(usdc.hold),mode:'AUDIT_ONLY_NO_TRADE_INSTRUCTIONS',strategyDayTotals:realized.summary.dayTotals,originalCarryBugStillUnrepaired:true}));",[210],"f713473c-0fb6-4b17-a986-6d32c82ef6ae","index.tsx",54,["D",1788700751921],18,"http","import { blob } from \"https://esm.town/v/std/blob/main.ts\";\n// Public market information only. Fixed upstream; no wallets, keys or trading.\nconst ORIGIN = \"https://hyperliquid-readonly-proxy.netlify.app\";\nconst VERSION = \"1.2.1\";\nconst COLS = [\"symbol\",\"venue\",\"markPx\",\"oraclePx\",\"change24hPct\",\"volume24hQuote\",\"oiBase\",\"oiNotionalQuote\",\"fundingHourly\",\"markOracleBps\",\"impactSpreadBps\",\"discoveryGate\",\"collateralSymbol\",\"maxLeverage\",\"szDecimals\"];\nconst CURATED_THEMES:Record={\n solana_ecosystem:[\"SOL\",\"JUP\",\"JTO\",\"PENGU\",\"FARTCOIN\",\"kBONK\",\"WIF\",\"PYTH\"],\n};\nfunction reply(data:unknown,status=200){return Response.json(data,{status,headers:{\"Cache-Control\":\"no-store\",\"X-Robots-Tag\":\"noindex, nofollow\"}});}\nfunction err(e:unknown){return e instanceof Error?e.message:String(e);}\nfunction median(xs:number[]){const a=xs.filter(Number.isFinite).toSorted((x,y)=>x-y);if(!a.length)return null;const m=Math.floor(a.length/2);return a.length%2?a[m]:(a[m-1]+a[m])/2;}\nfunction themeDiscovery(active:any[]){\n const eligible=active.filter((r:any)=>r.discoveryGate&&!r.gateUnitWarning&&r.assetClass===\"crypto\");\n const base24=median(eligible.map((r:any)=>r.change24hPct));\n const base15=median(eligible.map((r:any)=>r.deltas?.[\"15m\"]?.pricePct).filter((x:any)=>Number.isFinite(x)));\n const base1h=median(eligible.map((r:any)=>r.deltas?.[\"1h\"]?.pricePct).filter((x:any)=>Number.isFinite(x)));\n const themeMembers:Record>={};\n for(const [theme,symbols] of Object.entries(CURATED_THEMES))themeMembers[theme]=new Set(symbols);\n for(const r of eligible)for(const tag of (r.tags??[])){if(tag===\"unclassified\")continue;(themeMembers[\"tag:\"+tag]??=new Set()).add(r.symbol);}\n const metric=(rows:any[],tf:string)=>rows.map((r:any)=>r.deltas?.[tf]?.pricePct).filter((x:any)=>Number.isFinite(x));\n const signals=Object.entries(themeMembers).map(([theme,set])=>{\n const rows=eligible.filter((r:any)=>set.has(r.symbol));if(rows.length<2)return null;\n const m24=median(rows.map((r:any)=>r.change24hPct)),m15=median(metric(rows,\"15m\")),m1h=median(metric(rows,\"1h\"));\n const r24=m24==null||base24==null?null:m24-base24,r15=m15==null||base15==null?null:m15-base15,r1h=m1h==null||base1h==null?null:m1h-base1h;\n const hasCurrentInterval=m15!=null||m1h!=null; const score=hasCurrentInterval?((r24??0)*0.2+(r15??0)*0.35+(r1h??0)*0.45):null;\n const leaders=rows.toSorted((a:any,b:any)=>((b.deltas?.[\"1h\"]?.pricePct??b.change24hPct/24)-(a.deltas?.[\"1h\"]?.pricePct??a.change24hPct/24))).slice(0,5).map((r:any)=>({symbol:r.symbol,change24hPct:r.change24hPct,change15mPct:r.deltas?.[\"15m\"]?.pricePct??null,change1hPct:r.deltas?.[\"1h\"]?.pricePct??null,oi1hPct:r.deltas?.[\"1h\"]?.oiBasePct??null,volume24h:r.volume24hQuote}));\n return {theme,memberCount:rows.length,positive24h:rows.filter((r:any)=>r.change24hPct>0).length,positive15m:rows.filter((r:any)=>(r.deltas?.[\"15m\"]?.pricePct??-Infinity)>0).length,positive1h:rows.filter((r:any)=>(r.deltas?.[\"1h\"]?.pricePct??-Infinity)>0).length,median24hPct:m24,median15mPct:m15,median1hPct:m1h,relative24hPctPoints:r24,relative15mPctPoints:r15,relative1hPctPoints:r1h,currentSignalValid:hasCurrentInterval,signalStatus:hasCurrentInterval?\"CURRENT_INTERVAL_EVIDENCE\":\"INSUFFICIENT_CURRENT_INTERVAL_DATA\",themeStrengthScore:score,leaders};\n }).filter(Boolean) as any[];\n const ranked=signals.filter((s:any)=>s.currentSignalValid).toSorted((a:any,b:any)=>b.themeStrengthScore-a.themeStrengthScore); const historical24hThemes=signals.toSorted((a:any,b:any)=>(b.relative24hPctPoints??-Infinity)-(a.relative24hPctPoints??-Infinity)).slice(0,8);\n const discoveryLeaders=eligible.toSorted((a:any,b:any)=>((b.deltas?.[\"1h\"]?.pricePct??0)*0.6+(b.deltas?.[\"15m\"]?.pricePct??0)*0.3+b.change24hPct*0.1)-((a.deltas?.[\"1h\"]?.pricePct??0)*0.6+(a.deltas?.[\"15m\"]?.pricePct??0)*0.3+a.change24hPct*0.1)).slice(0,20).map((r:any)=>({symbol:r.symbol,tags:r.tags??[],change24hPct:r.change24hPct,change15mPct:r.deltas?.[\"15m\"]?.pricePct??null,change1hPct:r.deltas?.[\"1h\"]?.pricePct??null,oi1hPct:r.deltas?.[\"1h\"]?.oiBasePct??null,fundingHourly:r.fundingHourly,volume24h:r.volume24hQuote}));\n return {baseline:{eligibleCrypto:eligible.length,median24hPct:base24,median15mPct:base15,median1hPct:base1h},hotThemes:ranked.slice(0,8),coldThemes:ranked.toReversed().slice(0,8),historical24hThemes,discoveryLeaders,note:\"hotThemes require current 15m and/or 1h interval evidence. historical24hThemes are context only and must not be described as currently accelerating. Curated ecosystem taxonomy + upstream tags are detection signals, not proof of capital flow or causal fundamentals; verify promoted themes with fresh news/flow/fundamental evidence.\"};\n}\nfunction marketView(d:any,offset:number,limit:number,summary:boolean){\n const active=d.markets.filter((r:any)=>!r.isDelisted);\n const p=(r:any)=>({symbol:r.symbol,mark:r.markPx,change24hPct:r.change24hPct,volume24h:r.volume24hQuote,oiBase:r.oiBase,oiNotional:r.oiNotionalQuote,fundingHourly:r.fundingHourly,impactSpreadBps:r.impactSpreadBps,markOracleBps:r.markOracleBps});\n const eligible=active.filter((r:any)=>r.discoveryGate&&!r.gateUnitWarning);\n return {fetchedAt:d.fetchedAt,fetchedAtJST:d.fetchedAtJST,scope:d.scope,venueStatus:d.venues??d.venueStatus,history:d.history,clusters:d.clusters,cache:d.cache,\n note:\"All active rows retained. Impact spread is NOT best bid/ask. Collection time is not an exchange timestamp. Missing deltas are unknown.\",\n ...(summary?{themeDiscovery:themeDiscovery(active),liquid:eligible.toSorted((a:any,b:any)=>b.volume24hQuote-a.volume24hQuote).slice(0,15).map(p),gainers:eligible.toSorted((a:any,b:any)=>b.change24hPct-a.change24hPct).slice(0,15).map(p),losers:eligible.toSorted((a:any,b:any)=>a.change24hPct-b.change24hPct).slice(0,12).map(p),negativeFunding:eligible.filter((r:any)=>r.fundingHourly<0).toSorted((a:any,b:any)=>a.fundingHourly-b.fundingHourly).slice(0,12).map(p)}:{}),\n columns:COLS,offset,count:Math.min(limit,Math.max(0,active.length-offset)),total:active.length,hasMore:offset+limitCOLS.map(k=>r[k]??null))};\n}\nexport default async function(req:Request){\n if(req.method!==\"GET\")return reply({error:\"GET_ONLY\"},405);\n try{\n const u=new URL(req.url),mode=u.searchParams.get(\"mode\")??\"health\";\n if(mode===\"health\")return reply({ok:true,version:VERSION,marketDataOnly:true,noMicrofn:true,noSchedules:true,noWallets:true,upstream:ORIGIN,modes:[\"market\",\"technical\",\"read\"],usage:\"market: live all venues; technical: coins=HYPE,BTC; read: data=market or technical, offset/limit, no upstream fetch\"});\n if(![\"market\",\"technical\",\"read\"].includes(mode))return reply({error:\"INVALID_MODE\"},400);\n const offset=Number(u.searchParams.get(\"offset\")??0),limit=Number(u.searchParams.get(\"limit\")??0);\n if(!Number.isInteger(offset)||offset<0||!Number.isInteger(limit)||limit<0||limit>100)return reply({error:\"INVALID_PAGE\"},400);\n let kind=mode===\"read\"?(u.searchParams.get(\"data\")??\"market\"):mode;\n if(![\"market\",\"technical\"].includes(kind))return reply({error:\"INVALID_DATA\"},400);\n const key=kind+\"-latest\";\n let d:any,network=false;\n if(mode===\"read\"){\n d=await blob.getJSON(key);if(!d)return reply({error:\"NO_STORED_SNAPSHOT\"},404);\n }else{\n let path=\"/market?venue=all&kind=all\";\n if(mode===\"technical\"){\n const coins=(u.searchParams.get(\"coins\")??\"HYPE,BTC\").split(\",\");\n if(coins.length<1||coins.length>2||new Set(coins).size!==coins.length||coins.some(c=>!/^(?:[a-z][a-z0-9_-]{0,24}:)?[A-Za-z0-9_.-]{1,32}$/.test(c)))return reply({error:\"INVALID_COINS\"},400);\n path=\"/technical?coins=\"+encodeURIComponent(coins.join(\",\"));\n }\n const now=Date.now(),last:any=await blob.getJSON(key);\n if(last&&last.bridgePath===path&&now-Date.parse(last.fetchedAt)<60000){d=last;}\n else{\n const day=new Date().toISOString().slice(0,10);let budget:any=await blob.getJSON(\"budget\");if(budget?.day!==day)budget={day,count:0,lastMarket:0,lastTechnical:0};\n if(budget.count>=200)return reply({error:\"LOCAL_DAILY_UPSTREAM_CAP_200\",budget},429);\n const k=mode===\"market\"?\"lastMarket\":\"lastTechnical\",gap=mode===\"market\"?65000:31000;\n if(now-(budget[k]??0)4000000)throw new Error(\"UPSTREAM_TOO_LARGE\");\n if(!r.ok)return reply({error:\"UPSTREAM_HTTP_\"+r.status,detail:text.slice(0,1000),retryAfter:r.headers.get(\"retry-after\")},r.status===429?429:502);\n d=JSON.parse(text);const age=Date.now()-Date.parse(d.fetchedAt??\"\");if(!Number.isFinite(age)||age>90000||age< -5000)throw new Error(\"STALE_OR_INVALID_COLLECTION_TIME\");\n if(mode===\"market\"){\n if(!d.scope?.coverageComplete||!d.scope?.registryVerified||!Array.isArray(d.markets))throw new Error(\"INCOMPLETE_MARKET_COVERAGE\");\n const a=d.markets.filter((r:any)=>!r.isDelisted);\n if(a.length!==d.scope.perpScreenedValid||a.some((r:any)=>!r.quality?.valid))throw new Error(\"INVALID_ACTIVE_ROWS\");\n }\n if(mode==='technical'&&(!d.complete||!Array.isArray(d.symbols)||!d.out||d.symbols.some((s:string)=>['5m','15m','1h','4h','book'].some(t=>!d.out[s]?.[t]?.ok))))throw new Error('INCOMPLETE_TECHNICAL_SNAPSHOT');\n d.bridgePath=path;d.bridgeReceivedAt=new Date().toISOString();await blob.setJSON(key,d);network=true;\n }\n }\n const ageMs=Date.now()-Date.parse(d.fetchedAt??\"\");\n const meta={bridgeVersion:VERSION,networkFetch:network,storedSnapshot:mode===\"read\",snapshotAgeMs:ageMs,liveForExecution:ageMs>=0&&ageMs<=90000,servedAt:new Date().toISOString(),source:ORIGIN+d.bridgePath};\n if(u.searchParams.get(\"view\")===\"raw\")return reply({...meta,data:d});\n if(kind===\"market\"){\n if(u.searchParams.has('symbols')){const wanted=u.searchParams.get('symbols')!.split(',');if(wanted.length>25||wanted.some(s=>!/^(?:[a-z][a-z0-9_-]{0,24}:)?[A-Za-z0-9_.-]{1,32}$/.test(s)))return reply({error:'INVALID_SYMBOL_SELECTION'},400);const selected=d.markets.filter((r:any)=>wanted.includes(r.symbol));return reply({...meta,data:{fetchedAt:d.fetchedAt,scope:d.scope,history:d.history,requestedSymbols:wanted,foundSymbols:selected.map((r:any)=>r.symbol),rows:selected}});}\n return reply({...meta,data:marketView(d,offset,limit,mode!==\"read\"||u.searchParams.get(\"summary\")===\"1\")});\n }\n const compactOut=Object.fromEntries(Object.entries(d.out??{}).map(([coin,sections]:[string,any])=>[coin,Object.fromEntries(Object.entries(sections).map(([name,value]:[string,any])=>{if(name==='book')return [name,value];const {lastClosedBars,pivotsHigh,pivotsLow,...rest}=value;return [name,{...rest,pivotsHigh:pivotsHigh?.slice(-4),pivotsLow:pivotsLow?.slice(-4),lastClosedBars:lastClosedBars?.slice(-3)}];}))]));\n return reply({...meta,data:{...d,out:compactOut}});\n }catch(e){return reply({ok:false,error:err(e),at:new Date().toISOString()},502);}\n}\n",[217],"aea1e4bf-b494-4651-a94f-7034530c95b9","execution_preflight.ts","68c6ffbc-aa58-11f1-b79f-1607ee4eb77e",114,["D",1788751330350],4,"// Read-only order-planning diagnostics. Never submits exchange actions or changes leverage.\nconst user = Deno.env.get(\"HL_ACCOUNT_ADDRESS\");\nif (!user) throw new Error(\"MISSING_HL_ACCOUNT_ADDRESS\");\nconst post = async (body: Record) => {\n const r = await fetch(\"https://api.hyperliquid.xyz/info\", {method:\"POST\", headers:{\"Content-Type\":\"application/json\"}, body:JSON.stringify(body), signal:AbortSignal.timeout(10000)});\n if (!r.ok) throw new Error(`HL_HTTP_${r.status}`);\n return await r.json();\n};\nconst started = new Date().toISOString();\nconst [feeResult, ...assets] = await Promise.allSettled([\n post({type:\"userFees\", user}),\n ...[\"CRV\", \"xyz:MINIMAX\", \"xyz:GOLD\"].map(async coin => ({coin, data:await post({type:\"activeAssetData\", user, coin})}))\n]);\nconst f = feeResult.status === \"fulfilled\" ? feeResult.value : null;\nconsole.log(JSON.stringify({readOnly:true, startedAt:started, finishedAt:new Date().toISOString(), fees:f ? {userCrossRate:f.userCrossRate, userAddRate:f.userAddRate, activeReferralDiscount:f.activeReferralDiscount, feeSchedule:f.feeSchedule} : {error:\"USER_FEES_UNAVAILABLE\"}, assets:assets.map((r,i)=>r.status === \"fulfilled\" ? {coin:r.value.coin, leverage:r.value.data?.leverage, maxTradeSzs:r.value.data?.maxTradeSzs, availableToTradeDiagnosticOnly:r.value.data?.availableToTrade} : {coin:[\"CRV\",\"xyz:MINIMAX\",\"xyz:GOLD\"][i],error:\"ACTIVE_ASSET_DATA_UNAVAILABLE\"})}));\n",[225],"1616297b-244f-404e-9118-b58493383711","closed_trade_path_audit_readonly.ts","859a560e-aad3-11f1-a097-1607ee4eb77e",142,["D",1788796088059],"// Retrospective audit only: public candle /info reads, private audit blob writes.\n// No trading, no signing, no scheduling. Candle extrema are NOT historical Mark or guaranteed fills.\nimport { blob } from 'https://esm.town/v/std/blob/main.ts';\nconst latest:any=await blob.getJSON('project:hyperliquid:audit:latest-realized');if(!latest?.key)throw new Error('MISSING_REALIZED_AUDIT');\nconst raw:any=await blob.getJSON(latest.key);if(!raw?.rawFills)throw new Error('MISSING_RAW_FILLS');\nconst oids=new Set(raw.registeredOids.map((r:any)=>String(r.oid))),tids=new Set(raw.registeredTids.map((r:any)=>String(r.tid)));\nconst selected=['LINK','HYPE','UNI','ZEC'];const result:any[]=[];const rawCandles:Record={};\nconst n=(v:any)=>{const x=Number(v);if(!Number.isFinite(x))throw new Error('BAD_NUMBER');return x;};\nfor(const coin of selected){\n const fs=raw.rawFills.filter((f:any)=>f.coin===coin&&(oids.has(String(f.oid))||tids.has(String(f.tid)))).sort((a:any,b:any)=>n(a.time)-n(b.time));\n let pos=0,batch:any[]=[],cycles:any[][]=[];\n for(const f of fs){if(f.dir!=='Open Long'&&f.dir!=='Close Long')throw new Error('NON_LONG_EPISODE');batch.push(f);pos+=(f.side==='B'?1:-1)*n(f.sz);if(pos < -1e-8)throw new Error('UNSEEDED_EPISODE');if(Math.abs(pos)<1e-8){cycles.push(batch);batch=[];pos=0;}}\n if(pos!==0||!cycles.length)throw new Error('NOT_FLAT');\n const cycle=cycles.at(-1)!;const en=cycle.filter(f=>f.dir==='Open Long'),ex=cycle.filter(f=>f.dir==='Close Long');\n const sum=(a:any[],k:string)=>a.reduce((s,f)=>s+n(f[k]),0);const qty=sum(en,'sz'),entry=en.reduce((s,f)=>s+n(f.px)*n(f.sz),0)/qty;\n const entryStart=n(en[0].time),fullyFilledAt=n(en.at(-1).time),exitAt=n(ex.at(-1).time);\n const resp=await fetch('https://api.hyperliquid.xyz/info',{method:'POST',headers:{'Content-Type':'application/json'},body:JSON.stringify({type:'candleSnapshot',req:{coin,interval:'1m',startTime:entryStart-60000,endTime:exitAt+60000}}),signal:AbortSignal.timeout(15000)});if(!resp.ok)throw new Error('CANDLE_HTTP_'+resp.status);const candles=await resp.json();if(!Array.isArray(candles)||candles.length>=5000)throw new Error('CANDLE_INCOMPLETE');\n const inner=candles.filter(c=>n(c.t)>=fullyFilledAt&&n(c.T)n(c.t)<=exitAt&&n(c.T)>=fullyFilledAt);\n const ext=(a:any[])=>a.length?{low:Math.min(...a.map(c=>n(c.l))),high:Math.max(...a.map(c=>n(c.h)))}:null;const lo=ext(inner),hi=ext(boundary);\n const fee=sum(cycle,'fee'),gross=sum(ex,'closedPnl');const funding=raw.rawFunding.filter((f:any)=>f.delta?.coin===coin&&n(f.time)>=entryStart&&n(f.time)<=exitAt).reduce((s:number,f:any)=>s+n(f.delta.usdc),0);\n const exitGrouped=new Map();for(const f of ex){const k=String(f.oid);const a=exitGrouped.get(k)||{oid:k,qty:0,notional:0,at:new Date(n(f.time)).toISOString()};a.qty+=n(f.sz);a.notional+=n(f.sz)*n(f.px);exitGrouped.set(k,a);}\n const report={coin,entryStart:new Date(entryStart).toISOString(),fullyFilledAt:new Date(fullyFilledAt).toISOString(),exitAt:new Date(exitAt).toISOString(),qty,entryVwap:entry,exitVwap:ex.reduce((s,f)=>s+n(f.px)*n(f.sz),0)/sum(ex,'sz'),gross,fees:fee,funding,net:gross-fee+funding,innerFullMinuteBars:inner.length,innerPriceRange:lo,boundaryInclusivePriceRange:hi,mfeGrossInner:lo?Math.max(0,lo.high-entry)*qty:null,mfeGrossBoundaryUpper:hi?Math.max(0,hi.high-entry)*qty:null,maeGrossInner:lo?Math.max(0,entry-lo.low)*qty:null,exitOrders:[...exitGrouped.values()].map(a=>({...a,vwap:a.notional/a.qty})),caveat:'MFE/MAE use trade-price candle extrema after all entry tranches fill. Inner minute bars omit boundary seconds. Inclusive bars may include pre-entry or post-exit prices. Neither proves Mark-trigger reach or executable profit.'};\n result.push(report);rawCandles[coin]=candles;\n}\nconst summary={schemaVersion:'closed-trade-path-audit-v1',asOf:new Date().toISOString(),sourceRealizedKey:latest.key,trades:result,notBacktest:true,noProfitabilityClaim:true};\nconst key='project:hyperliquid:audit:'+summary.asOf+':paths';await blob.setJSON(key,{summary,rawCandles});await blob.setJSON('project:hyperliquid:audit:latest-paths',{key,summary});console.log(JSON.stringify({privateAuditKey:key,...summary}));",[233],"ef6405d3-1656-4eba-a21c-9a70c41c2715","account_checker.ts","960fd2b8-aa09-11f1-a49e-1607ee4eb77e",105,["D",1788742421944],3,"import { sqlite } from \"https://esm.town/v/std/sqlite/main.ts\";\n\nconst INFO = \"https://api.hyperliquid.xyz/info\";\nconst JST_MS = 9 * 60 * 60 * 1000;\nconst DAY_MS = 86_400_000;\nconst DEFAULT_EFFECTIVE_JST = \"2026-09-07\";\nconst DEFAULT_BASE_TARGET = 200;\n\nconst num = (v: unknown) => {\n const n = typeof v === \"number\" ? v : Number(String(v ?? \"0\").trim());\n return Number.isFinite(n) ? n : 0;\n};\nconst round = (v: number, dp = 6) => {\n const p = 10 ** dp;\n return Math.round((v + Number.EPSILON) * p) / p;\n};\nconst jstDateKey = (ms: number) => new Date(ms + JST_MS).toISOString().slice(0, 10);\nconst jstMidnightMs = (d: string) => Date.parse(`${d}T00:00:00+09:00`);\nconst fillKey = (f: any) => String(f?.tid ?? `${f?.time}|${f?.coin}|${f?.oid}|${f?.side}|${f?.px}|${f?.sz}`);\nconst fundingKey = (r: any) => `${String(r?.hash ?? \"\")}|${String(r?.time ?? \"\")}|${String(r?.delta?.coin ?? \"\")}|${String(r?.delta?.usdc ?? \"\")}`;\nconst projectDelta = (f: any) => {\n const d = String(f?.dir ?? \"\").toLowerCase();\n const s = Math.abs(num(f?.sz));\n if (d.includes(\"open long\")) return s;\n if (d.includes(\"close long\")) return -s;\n if (d.includes(\"open short\")) return -s;\n if (d.includes(\"close short\")) return s;\n return 0;\n};\nconst uniqueBy = (rows: any[], key: (r: any) => string) => {\n const m = new Map();\n for (const r of rows) m.set(key(r), r);\n return [...m.values()];\n};\n\nasync function post(body: Record) {\n const r = await fetch(INFO, {\n method: \"POST\",\n headers: { \"Content-Type\": \"application/json\" },\n body: JSON.stringify(body),\n });\n if (!r.ok) throw new Error(`HL_HTTP_${r.status}:${(await r.text()).slice(0, 200)}`);\n return await r.json();\n}\n\nasync function fetchRange(type: \"userFillsByTime\" | \"userFunding\", user: string, startTime: number, endTime: number, depth = 0): Promise {\n const body: any = { type, user, startTime, endTime };\n if (type === \"userFillsByTime\") body.aggregateByTime = false;\n const rows = await post(body);\n if (!Array.isArray(rows)) throw new Error(`${type}_NOT_ARRAY`);\n if (rows.length < 2000) return rows;\n if (depth >= 24 || endTime - startTime <= 1) throw new Error(`${type}_WINDOW_TOO_DENSE`);\n const mid = Math.floor((startTime + endTime) / 2);\n return [...await fetchRange(type, user, startTime, mid, depth + 1), ...await fetchRange(type, user, mid + 1, endTime, depth + 1)];\n}\n\nawait sqlite.execute(`CREATE TABLE IF NOT EXISTS strategy_oids (\n oid TEXT PRIMARY KEY, coin TEXT, side TEXT, venue TEXT, order_type TEXT,\n limit_px REAL, trigger_px REAL, orig_sz REAL, parent_oid TEXT,\n plan_id TEXT, registered_at TEXT NOT NULL\n)`);\nawait sqlite.execute(`CREATE TABLE IF NOT EXISTS strategy_tids (\n tid TEXT PRIMARY KEY, plan_id TEXT, registered_at TEXT NOT NULL\n)`);\nawait sqlite.execute(`CREATE TABLE IF NOT EXISTS strategy_seeds (\n coin TEXT PRIMARY KEY, signed_size REAL NOT NULL, at_time INTEGER NOT NULL,\n note TEXT, registered_at TEXT NOT NULL\n)`);\nawait sqlite.execute(`CREATE TABLE IF NOT EXISTS strategy_meta (\n key TEXT PRIMARY KEY, value TEXT NOT NULL\n)`);\n\nconst user = Deno.env.get(\"HL_ACCOUNT_ADDRESS\");\nif (!user) throw new Error(\"MISSING_HL_ACCOUNT_ADDRESS\");\nconst now = Date.now();\nconst todayJST = jstDateKey(now);\nconst metaRows = (await sqlite.execute(\"SELECT key,value FROM strategy_meta\")).rows as any[];\nconst meta = Object.fromEntries(metaRows.map((r: any) => [String(r.key), String(r.value)]));\nconst effectiveJST = meta.effective_from_jst || DEFAULT_EFFECTIVE_JST;\nconst effectiveStart = jstMidnightMs(effectiveJST);\nconst todayStart = jstMidnightMs(todayJST);\nconst baseTarget = num(meta.base_daily_target_usd) || DEFAULT_BASE_TARGET;\nconst priorCarry = num(meta.prior_carry_in_usd);\n\nconst [mainState, mainOrders, xyzState, xyzOrders, spotState, userAbstraction, fillsRaw, fundingRaw, oidRows, tidRows, seedRows] = await Promise.all([\n post({ type: \"clearinghouseState\", user }),\n post({ type: \"frontendOpenOrders\", user }),\n post({ type: \"clearinghouseState\", user, dex: \"xyz\" }),\n post({ type: \"frontendOpenOrders\", user, dex: \"xyz\" }),\n post({ type: \"spotClearinghouseState\", user }),\n post({ type: \"userAbstraction\", user }),\n fetchRange(\"userFillsByTime\", user, effectiveStart, now),\n fetchRange(\"userFunding\", user, effectiveStart, now),\n sqlite.execute(\"SELECT * FROM strategy_oids\"),\n sqlite.execute(\"SELECT * FROM strategy_tids\"),\n sqlite.execute(\"SELECT * FROM strategy_seeds\"),\n]);\n\nconst fills = uniqueBy(fillsRaw, fillKey).filter((f: any) => {\n const c = String(f?.coin ?? \"\");\n return c && !c.startsWith(\"@\") && /(Long|Short)/i.test(String(f?.dir ?? \"\"));\n}).sort((a: any, b: any) => num(a.time) - num(b.time));\nconst funding = uniqueBy(fundingRaw, fundingKey).sort((a: any, b: any) => num(a.time) - num(b.time));\nconst oidSet = new Set((oidRows.rows as any[]).map((r: any) => String(r.oid)));\nconst tidSet = new Set((tidRows.rows as any[]).map((r: any) => String(r.tid)));\nconst seeds = Object.fromEntries((seedRows.rows as any[]).map((r: any) => [String(r.coin), { signedSize: num(r.signed_size), atTime: num(r.at_time) }]));\nconst tagged = fills.filter((f: any) => oidSet.has(String(f?.oid ?? \"\")) || tidSet.has(String(f?.tid ?? \"\")));\n\nconst byCoin = new Map();\nfor (const f of tagged) {\n const c = String(f.coin);\n if (!byCoin.has(c)) byCoin.set(c, []);\n byCoin.get(c)!.push(f);\n}\nconst activeAt = (coin: string, t: number) => {\n const seed = seeds[coin];\n let pos = seed && seed.atTime <= t ? seed.signedSize : 0;\n for (const f of byCoin.get(coin) || []) {\n const ft = num(f.time);\n if (ft > t) break;\n if (seed && ft < seed.atTime) continue;\n pos += projectDelta(f);\n }\n return Math.abs(pos) > 1e-12;\n};\n\nlet accountGross = 0, accountFees = 0, accountFunding = 0;\nlet strategyGross = 0, strategyFees = 0, strategyFunding = 0, strategyFundingCount = 0;\nfor (const f of fills) {\n const t = num(f.time);\n if (t < todayStart || t > now) continue;\n accountGross += num(f.closedPnl);\n if (String(f.feeToken ?? \"USDC\").toUpperCase() === \"USDC\") accountFees += num(f.fee);\n}\nfor (const f of tagged) {\n const t = num(f.time);\n if (t < todayStart || t > now) continue;\n strategyGross += num(f.closedPnl);\n if (String(f.feeToken ?? \"USDC\").toUpperCase() === \"USDC\") strategyFees += num(f.fee);\n}\nfor (const r of funding) {\n const t = num(r.time);\n if (t < todayStart || t > now || String(r?.delta?.type ?? \"\") !== \"funding\") continue;\n const v = num(r?.delta?.usdc);\n accountFunding += v;\n const coin = String(r?.delta?.coin ?? \"\");\n if (activeAt(coin, t)) {\n strategyFunding += v;\n strategyFundingCount++;\n }\n}\n\nconst accountNet = accountGross - accountFees + accountFunding;\nconst strategyNet = strategyGross - strategyFees + strategyFunding;\nconst requiredToday = baseTarget + priorCarry;\nconst remainingToday = Math.max(0, requiredToday - strategyNet);\nconst positionRows = [\n ...(Array.isArray(mainState?.assetPositions) ? mainState.assetPositions : []),\n ...(Array.isArray(xyzState?.assetPositions) ? xyzState.assetPositions : []),\n].map((x: any) => x.position).filter((p: any) => Math.abs(num(p?.szi)) > 1e-12);\nconst openOrders = [\n ...(Array.isArray(mainOrders) ? mainOrders : []),\n ...(Array.isArray(xyzOrders) ? xyzOrders : []),\n];\nconst usdcBalance = (Array.isArray(spotState?.balances) ? spotState.balances : []).find((b: any) => String(b?.coin ?? \"\").toUpperCase() === \"USDC\");\nconst usdcTotal = usdcBalance ? num(usdcBalance.total) : null;\nconst usdcHold = usdcBalance ? num(usdcBalance.hold) : null;\nconst automatedAvailableBalance = userAbstraction === \"unifiedAccount\" && usdcTotal !== null && usdcHold !== null\n ? Math.max(0, usdcTotal - usdcHold)\n : null;\n\nconsole.log(JSON.stringify({\n checkerVersion: \"val-sqlite-v1.0.0\",\n authority: \"VAL_TOWN_SQLITE\",\n fetchedAt: new Date(now).toISOString(),\n jst: { date: todayJST, dayStartAt: new Date(todayStart).toISOString(), timezone: \"Asia/Tokyo\" },\n headroom: {\n accountAbstraction: String(userAbstraction ?? \"unknown\"),\n authoritativeMethod: automatedAvailableBalance !== null ? \"UNIFIED_ACCOUNT_USDC_TOTAL_MINUS_HOLD\" : \"UI_AVAILABLE_BALANCE_REQUIRED\",\n availableBalanceUSDC: automatedAvailableBalance !== null ? round(automatedAvailableBalance, 8) : null,\n usdcTotal: usdcTotal !== null ? round(usdcTotal, 8) : null,\n usdcHold: usdcHold !== null ? round(usdcHold, 8) : null,\n },\n positions: positionRows,\n openOrders,\n targetTracking: {\n authoritativeMethod: \"ACCOUNT_WIDE_JST_STRICT_REALIZED_PERP_NET\",\n todayStrictNetUSD: round(accountNet),\n grossClosedPnlUSD: round(accountGross),\n tradingFeesUSD: round(accountFees),\n fundingUSD: round(accountFunding),\n },\n strategyTracking: {\n authoritativeMethod: \"VAL_TOWN_SQLITE_TAGGED_HYPERLIQUID_FILLS_AND_FUNDING\",\n effectiveFromJST: effectiveJST,\n priorCarryInUSD: round(priorCarry),\n baseDailyTargetUSD: round(baseTarget),\n requiredTodayUSD: round(requiredToday),\n todayStrategyStrictNetUSD: round(strategyNet),\n remainingTodayUSD: round(remainingToday),\n todayGoalMet: remainingToday <= 0,\n grossClosedPnlUSD: round(strategyGross),\n tradingFeesUSD: round(strategyFees),\n fundingUSD: round(strategyFunding),\n fundingAttributedCount: strategyFundingCount,\n registeredOidCount: oidSet.size,\n registeredTidCount: tidSet.size,\n seededPositionCount: Object.keys(seeds).length,\n },\n recentPerpFills: fills.slice(-25).map((f: any) => ({\n time: num(f.time), at: new Date(num(f.time)).toISOString(), tid: String(f.tid ?? \"\"), oid: String(f.oid ?? \"\"),\n coin: String(f.coin ?? \"\"), dir: String(f.dir ?? \"\"), px: num(f.px), sz: num(f.sz),\n closedPnl: num(f.closedPnl), fee: num(f.fee), feeToken: String(f.feeToken ?? \"USDC\"),\n tagged: oidSet.has(String(f.oid ?? \"\")) || tidSet.has(String(f.tid ?? \"\")),\n })),\n}));\n",[240],"028a7759-744a-4039-97a6-0d0677324b5d","README.md","01a06ff1-37fe-726c-835e-88db01f1a3f6",9,["D",1788585922073],1,"# Hyperliquid market bridge — verified on-demand retrieval\n\nThis service is public MARKET DATA ONLY. It contains no wallet address, positions, credentials, order signing or trading functions. It has no scheduled jobs. It was verified with actual Hyperliquid responses on 2026-09-05, not only mocked tests.\n\n## Stable identity\n- Val: `kamibaba/hyperliquid-market-bridge`\n- HTTP file: `index.tsx`; branch: `main`\n- Bridge version: `1.1.1`\n- Fixed upstream: `https://hyperliquid-readonly-proxy.netlify.app`\n- Netlify project: `hyperliquid-readonly-proxy`, site ID `378ad326-88f1-4aef-aac5-1a3734fe8627`\n- No custom domain is needed. The unrelated sound-notes.net project must not be edited.\n\n## ChatGPT workflow — no Mac operation or uploaded JSON required\nUse the connected `Val_Town.fetch_val_endpoint` tool. Always pass the actual val identity and HTTP file; do not construct or guess val endpoint URLs.\n\n1. Fetch all venues: `{ \"val\": \"kamibaba/hyperliquid-market-bridge\", \"path\": \"index.tsx\", \"search\": \"mode=market&limit=0\" }`.\n2. Validate HTTP status, `data.scope.coverageComplete`, `registryVerified`, active/valid counts, venue errors and `data.fetchedAt`. The universe is dynamic, not a fixed 315 or 370.\n3. Read every active row without refetching upstream: `mode=read&data=market&offset=0&limit=100`; then 100, 200, 300 until `hasMore=false`. If tool output truncates, reduce limit. Check fetchedAt is unchanged across pages.\n4. Inspect shortlisted symbols including interval deltas: `mode=read&data=market&symbols=HYPE,ZEC,NEAR,DASH,LIT`. Keep selection small to avoid output truncation. This is explicitly a subset of the already-fetched full market.\n5. Fetch closed candles and book for at most two symbols per call: `mode=technical&coins=HYPE,BTC` or `mode=technical&coins=xyz:MU,xyz:AVGO`. Default response is compact; raw is available using `view=raw`.\n6. If a response is truncated or interrupted, `mode=read&data=technical` reads the saved result without an upstream call. Verify `data.symbols` matches the intended pair before using it. A 429 may coexist with a recently completed request: inspect the stored snapshot rather than blindly retrying.\n7. Refresh market or selected technical data near the final trading decision. `mode=read` is historical snapshot access, never a promise of current market data.\n\nThe connector's built-in `Val_Town.web_fetch` also successfully fetched the Netlify /health endpoint, but health alone does not verify market coverage.\n\n## Quotas and cost\nThis val is on the Free tier. The upstream Netlify site remains on its existing Free plan; shared account usage still applies. No paid plan, auto recharge, inference model, cron or trading agent is configured. Additional tool calls may use ChatGPT limits separately.\n\n- Market upstream cooldown: 65 seconds; technical upstream cooldown: 31 seconds.\n- Reuse same-query snapshots younger than 60 seconds.\n- An application-level budget limits upstream attempts to 200/day (UTC). This is a best-effort application counter, NOT an atomic billing cap or a promise of unlimited free usage. Netlify also enforces its own request limits and Free plan cap.\n- Read pages use the stored snapshot, not more Hyperliquid/Netlify collection calls.\n- Stored snapshots are overwritten; there is no unbounded full-snapshot archive here.\n\n## Interpretation safeguards\n- All-perp scope includes Core, XYZ and every HIP-3 venue returned by the official registry. `kind=all` also fetches spot metadata/contexts, but this bridge's default rows are active perps; do not claim deep analysis of all spot pairs.\n- `fetchedAt` is collection time. Asset contexts have no independent exchange event timestamp. The order book has its own source timestamp.\n- 24-hour returns are rolling returns, NOT the return since midnight JST.\n- Impact spread is NOT best bid/ask spread; use the book for execution analysis.\n- OI base and OI notional are distinct. OI increase does not identify which directional side initiated trading.\n- Missing interval history remains unknown; no inferred liquidation maps or spot CVD.\n- Curated tags are incomplete and may overlap. ZEN is a Base token; grouping it with historical privacy names is a thematic comparison, not a same-chain assertion. NEAR also has privacy-related products despite its basic layer1 label.\n- All technical indicators use closed contiguous bars. ATR is neither a guaranteed range nor an automatic minimum stop distance.\n- Reject partial, stale or wrong-symbol detailed responses. `liveForExecution` only checks snapshot age, not profitability or fill guarantees.\n- Use market snapshots for setup screening, then check specific execution prices and expiry conditions. Do not enlarge risk merely to target a daily dollar profit.\n\n## Actual verification\n2026-09-05T05:15:51.880Z and 2026-09-05T05:21:23.682Z: all 11 discovered venues returned HTTP 200; 512 listed perp records, 315 active/valid, 326 spot pairs. All 315 active rows were read from the first snapshot in four pages. The second snapshot refreshed the full universe and ranking outputs.\n\nTechnical requests succeeded for HYPE, BTC, NEAR, ZEC, DASH, LIT, CASHCAT, ZRO, SKR and ZORA. HYPE/ZEC were refreshed at 2026-09-05T05:23:28.897Z. These are historical verification times, never current trading prices in a future chat.\n\nMarket collection does not use microfn. microfn remains a separate account-checking route only when the user reports position/order changes. Do not use this public val to store the user's account data.\n",[248],"66a54470-729c-4769-97d9-65136dbc2676","PROJECT_RULES.md","8c82135a-a905-11f1-bc44-1607ee4eb77e",140,["D",1788795680414],48,"# Hyperliquid / XYZ Project Rules — Persistent Canonical Rules\n\nUpdated: 2026-09-07 JST\n\nThis file contains only reusable trading/workflow rules. It intentionally contains no wallet address, account balances, positions, or private account state.\n\n## Startup procedure for every new chat\n1. Fix current JST time.\n2. If the user reports a position/order change or this is a handoff boundary, run the Val Town script `account_checker.ts` in `kamibaba/hyperliquid-market-bridge` to retrieve live positions/orders and Strategy tracking. Do not use that route for market screening. MicroFn account functions are non-authoritative fallbacks only; do not depend on them for ledger registration or PnL authority.\n3. Retrieve the full current Hyperliquid/HIP-3 universe through `kamibaba/hyperliquid-market-bridge`.\n4. Validate `coverageComplete=true`, `registryVerified=true`, no venue failures, fresh `fetchedAt`, and active/valid counts.\n5. Read all active rows using stored-snapshot pagination; do not claim full-universe screening without reading the full active set.\n6. Screen Long and Short candidates independently, then decide whether the best portfolio should be one-sided or multi-leg. Long+Short is not mandatory.\n7. For finalists, fetch exact-symbol 5m/15m/1h/4h closed candles and L2 book (max two symbols per detailed request).\n8. Check current macro/news/fundamentals where material. Execution levels must use Hyperliquid/XYZ prices, not external cash prices.\n9. Define structural SL and structural TP first, then risk, notional, leverage and margin mode.\n10. Present exact prices, sizes, leverage, margin mode, SL/TP, expiry, cancellation conditions and chase cutoff.\n\n## Market data transport\nVal: `kamibaba/hyperliquid-market-bridge`\nHTTP file: `index.tsx`\nBridge version baseline: `1.2.1` (verify current version if behavior changes)\nUpstream: deployed read-only Netlify Hyperliquid market service.\nNo paid third-party liquidation API is required or configured. Use liquidation data only when it is genuinely available at no added cost or is supplied directly by the user; otherwise proceed without it rather than weakening the setup with an uncertain or paid dependency.\nNo schedules / no automated trading / no wallet data in the market bridge.\n\nLiquidity veto: as of 2026-09-07 the user has explicitly rejected `xyz:JPY` for actionable trading because practical liquidity is inadequate. Exclude `xyz:JPY` from trade instructions unless the user explicitly re-enables it after fresh evidence of adequate executable liquidity. More generally, do not infer usable liquidity from mark/oracle proximity or displayed impact spread alone; verify actual L2 depth, executable VWAP, 24h turnover, and observed fill practicality before sizing.\n\nTypical calls:\n- `mode=market&limit=0` — refresh all venues.\n- `mode=read&data=market&offset=0&limit=100`, then 100/200/300... until `hasMore=false` — read the same saved snapshot without another upstream collection.\n- `mode=read&data=market&symbols=...` — shortlisted exact rows/deltas.\n- `mode=technical&coins=AAA,BBB` — exact symbols; closed 5m/15m/1h/4h bars + L2 book, max 2 symbols.\n\nNever assume the universe is a fixed count. Discover venues and active markets dynamically each time.\n\n## Account-data route\nThe authoritative account checker is the Val Town script `account_checker.ts` in `kamibaba/hyperliquid-market-bridge`. It is for live position/order verification after reported account changes and handoff boundaries, JST account-wide realized-PnL auditing, and the project-scoped Strategy ledger. Do not use it for broad market screening. The account address is stored in the val environment as `HL_ACCOUNT_ADDRESS`, not in the public market HTTP file.\n\nThe authoritative Strategy ledger is the val-scoped SQLite database in `kamibaba/hyperliquid-market-bridge`, using tables `strategy_oids`, `strategy_tids`, `strategy_seeds`, and `strategy_meta`. `account_checker.ts` returns live `positions`, `openOrders`, account-wide `targetTracking`, Strategy-scoped `strategyTracking`, and recent perp fills with a `tagged` flag. `targetTracking` can include independent discretionary trades. `strategyTracking` is the project authority and includes only fills whose OID/TID is registered in SQLite plus reconstructed funding while a project position is open.\n\nMicroFn account functions are retained only as non-authoritative recovery/audit fallbacks. On 2026-09-07, all three MicroFn execution routes returned generic `Function execution failed` even with a one-line diagnostic function, so ledger/PnL operation MUST NOT depend on MicroFn availability.\n\nWhen the user says an instructed order was placed/modified, immediately run `account_checker.ts`, identify the actual parent/child OIDs and any newly executed TIDs, and write them with parameterized `INSERT OR REPLACE` statements through Val Town `sqlite_batch`. Future fills on registered OIDs are then attributed automatically. For positions carried across a ledger boundary, add one signed row in `strategy_seeds` at the boundary so subsequent funding can be reconstructed. Independent discretionary orders must never be tagged as project orders.\n\nVal Town Strategy authority starts at 2026-09-07 JST. The 2026-09-06 final strict Strategy Net baseline is fixed at `-$2.965045`, producing a 2026-09-07 prior carry-in of `$202.965045`. The carried PUMP short was seeded at the 2026-09-07 00:00 JST boundary. The legacy MicroFn ledger remains untouched for later audit/reconciliation if its execution service recovers.\n\nFor usable new-position headroom, first query the user's account abstraction through official Hyperliquid `/info` and the current `spotClearinghouseState`. When `userAbstraction=\"unifiedAccount\"`, `account_checker.ts` automatically computes the Hyperliquid Balances-table USDC Available Balance as `USDC total - USDC hold` and returns it under `headroom.availableBalanceUSDC`; this is the primary headroom authority. Do NOT substitute API `withdrawable`, `accountValue`, `marginSummary.accountValue`, or `crossMarginSummary.accountValue`, because those are different quantities. If the abstraction mode is not unifiedAccount, the required balance fields are unavailable, or a user-visible UI value demonstrably disagrees with the computed value, fall back to the Hyperliquid UI `Balances → USDC → Available Balance` and investigate the discrepancy before sizing further margin-consuming orders. Always re-run `account_checker.ts` after any margin-changing event (entry/add fill, position-size change, newly accepted margin-reserving order, leverage/margin adjustment, or other event that changes hold) so the headroom value is refreshed; never reuse a stale prior balance or infer fresh headroom merely by subtracting theoretical `notional/leverage`. Position sizing must still be limited by structural stop risk rather than by maximum buying power.\n\n## Portfolio selection\nThe goal is the highest-expectancy portfolio, not preservation of the last favorite symbol.\nConsider:\n- long only\n- short only\n- multiple longs\n- multiple shorts\n- long + short\n- relative-value pairs\n- current-position maintenance/reduction/exit\n\nAlways inspect both Long and Short opportunity sets, but do NOT force a hedge. First assess broad market regime, majors, sector/category/chain rotation and correlation. Independently of any user hint, perform an automatic narrative/theme-discovery pass on every meaningful crypto decision: start from cross-sectional leaders/laggards and short-interval relative performance, cluster symbols by shared tokenomics/chain/sector/catalyst characteristics, then actively verify likely common causes with fresh fundamental/news/flow evidence. Explicitly test themes such as buybacks/burns/value-accrual, unlock/supply overhang, launchpad/memecoin rotation, chain migration/ecosystem rotation, AI/agent, DeFi/RWA, privacy, staking/emissions, and any newly emerging theme suggested by the live leaders; this list is illustrative, not exhaustive. A theme should be promoted when enough leaders/benchmarks show current relative strength plus fresh causal evidence; the trade candidate itself does NOT need to have already accelerated. After a theme is promoted, split the basket into (a) leaders/continuation names and (b) catch-up/laggard candidates. Explicitly compare buying the leader continuation versus buying an unextended laggard that is genuinely exposed to the same catalyst/flow, has no idiosyncratic negative catalyst, retains acceptable 1h/4h structure, has adequate liquidity, and is less crowded on funding/OI. Do not reject a laggard merely because it is flat while leaders are moving; that flatness can be the opportunity. Conversely, never buy a laggard solely because it has not moved—distinguish true catch-up candidates from fundamentally weak non-participants. If the user separately identifies a live theme, treat that as an additional high-priority hypothesis to test, not as the only source of theme discovery. A theme leader that is already extended must not be dismissed solely for being up a lot; evaluate breakout/retest, small-core + pullback-add, and continuation structures so non-fill opportunity cost is modeled. The max-two-symbol technical endpoint is a per-call constraint, not a reason to skip serial detailed checks on additional high-priority theme finalists.\n\nFor crypto, explicitly classify the broad regime BEFORE selecting directional trades using at minimum BTC/ETH/SOL/BNB relative strength, 5m/15m/1h/4h trend context, breadth across major Layer-1/DeFi/meme clusters, funding/OI where available, and current macro/flow/fundamental context. A sharp short-term decline in one or two majors must not by itself flip the broad regime to bearish when market breadth and higher-timeframe structure remain risk-on. If most majors/major-alt clusters are positive and higher-timeframe structure is still up, do not build a multi-leg net-short portfolio unless it is explicitly a relative-value/hedged construction with controlled net beta.\n\nFor any trade whose thesis is causally derived from another symbol, theme leader, catalyst or flow source, explicitly record that dependency and keep re-checking it while the trade is open or while adds remain pending. The dependent trade's own technical trend is not sufficient to preserve the thesis if the causal driver has materially changed. Example: if PUMP weakness is hypothesized to result from attention/trading-volume rotation toward PONS or other competing launchpad/chain tokens, then those competitors stalling or declining is a mandatory re-check of whether the proposed rotation still exists; it is NOT evidence that their selling mechanically causes PUMP selling. Distinguish competing-flow (potential inverse relationship) hypotheses from shared-beta (positive co-movement) hypotheses. Price co-movement alone does not prove either causal flow. Broad-market risk-on strengthening also requires re-checking a counter-regime short. When the causal driver stops confirming, first cancel pending adds; then reassess whether to reduce/exit the dependent position even if its own 15m/1h/4h trend still points the original way. Never keep adding merely because the dependent symbol remains technically down/up after the original causal impulse has faded.\n\nIf Long+Short is used, size by structural-stop risk contribution and correlation/beta, not equal dollar notional. Highly correlated instruments that express the same underlying thesis (for example WTI + Brent, BTC + ETH beta longs, or multiple names driven by the same single catalyst) must be treated as one risk bucket for portfolio risk and daily-capacity stress tests. Do not count their profit capacities as fully independent when applying the single-largest-independent-setup-failure test, and cap aggregate stop/gap risk at the bucket level rather than per symbol.\n\nDirectional concentration is not a defect by itself. If the broad market regime and the best independent setups all point the same way, a one-sided portfolio (all LONG or all SHORT) may be optimal. Never add an opposite-direction position merely to create visual balance, reduce net exposure mechanically, or satisfy a hedge quota; in a persistent trend this creates predictable drag and can guarantee that one side loses. Regime-aligned trades use the normal entry standard: in a clearly bullish regime, LONGs do not require a special idiosyncratic catalyst beyond a valid setup; in a clearly bearish regime, SHORTs do not require one. Counter-regime trades require materially stronger independent evidence: e.g. clear relative weakness/strength plus a specific fundamental, flow, supply-demand, catalyst, or structural reason that explains why that asset should diverge from the broad market. Apply this symmetrically: bullish regime -> SHORT needs extra justification; bearish regime -> LONG needs extra justification. Same-direction additions are restricted only when they are duplicative/highly correlated without incremental expectancy or push aggregate structural-stop risk above the chosen risk budget. Portfolio direction must follow regime + expectancy, not forced symmetry.\n\nUser clarification on 2026-09-08: the primary failure mode to avoid is NOT simply too many positions, too few positions, or large/small lot size. The decisive quality gate is whether the selected symbol, direction, entry, structural invalidation, take-profit path and combined portfolio are themselves high-expectancy. Diversification, smaller sizing, opposite-direction exposure or more symbols must never be used as a substitute for a weak setup. Every proposed leg must first pass a standalone setup-quality test: current regime fit; verified relative strength/weakness; fresh causal/fundamental/flow support where relevant; entry location that is not merely chasing recent performance; structural SL outside normal noise; realistic TP(s) supported by observed structure or explicitly labeled extension; and net RR/expectancy after fees/slippage. Only then may legs be combined. Portfolio construction must then test whether correlations, shared catalysts and stop/target geometry improve or degrade the joint distribution. A portfolio of individually weak trades is invalid even if nominally diversified.\n\nIf one leg's thesis breaks, close/reduce that leg; do not keep a broken long merely because a short hedge exists.\n\n## Crypto-specific behavior\nDo not apply stock logic mechanically to crypto. Include:\n- weekend alt rotations / “alt hour” behavior\n- chain/category/theme capital rotation\n- short/long squeeze risk\n- funding crowding\n- OI changes\n- Mark/Oracle premium\n- liquidity / spread / order book\n- liquidation data only when actually available. When a verified liquidation heatmap / liquidation-level source is available (including a user-supplied CoinGlass screenshot or a fetched API/source), use it as a secondary liquidity-map input: distinguish overhead short-liquidation clusters from downside long-liquidation clusters where possible, compare distance/size with current trend, funding/OI, book and ATR, and use nearby clusters to refine continuation probability, TP/scale-out and squeeze risk. Do not describe estimated heatmap bands as guaranteed actual liquidations or proof of deliberate \"liquidation hunting\"; they are modeled potential concentration zones, not deterministic targets. In a trend-aligned setup, nearby liquidation clusters in the direction of travel can increase continuation expectancy, but never use them as the sole entry thesis.\n- BTC/ETH relative strength and alt divergence\n\nFunding sign alone is never an entry signal. OI changes alone do not identify directional initiators.\n\n## Entry execution\nExecution choice is part of expectancy. Compare:\n- marketable core\n- single limit\n- manual limit ladder\n- core + ladder\n- breakout core + retest add\n- scale-out on exits\n\nDo not default to one “perfect” limit. Model non-fill opportunity cost.\n\nUse a small Core when momentum/hedge urgency makes waiting costly. Do not chase if already directionally exposed and pullback RR is better.\n\nThe default actionable entry must be placeable immediately without requiring the user to watch for a later candle close, reclaim, or other manual confirmation trigger. Prefer resting Limit / ladder / core+ladder orders with hard attached protection. If a setup is only valid after later confirmation, either choose another immediately placeable symbol from the full universe or label it watch-only rather than making it the primary instruction. Do not require continuous screen-watching.\n\nFor catalyst-driven crypto that has already made a sharp extension, do not treat the catalyst alone as permission to passively buy a pullback. If lower-timeframe rollover makes a passive Limit a falling-knife entry, choose a different symbol/setup that supports a structurally protected resting order rather than requiring the user to wait for a manual confirmation trigger. A scheduled buyback/reserve purchase with undisclosed size is qualitative support, not a quantitatively guaranteed bid. For scheduled tokenomics events such as unlocks, claims, emissions changes or burns, re-verify the event's current status and timestamp before demoting or replacing an otherwise stronger setup; do not treat a stale \"today/upcoming\" page state as proof that the event is still ahead. If the event status is genuinely uncertain but the asset remains a regime-aligned leader with strong multi-timeframe structure and positive value-accrual/flow support, manage the uncertainty through smaller Core size, wider structural risk allowance or delayed adds rather than automatically replacing it with a weaker asset. When momentum is strong and passive pullback entries risk non-fill, explicitly compare a small marketable Core + deeper adds against an all-limit ladder.\n\nBefore placing a structural stop around a support/resistance level, compare the buffer beyond that level with current 5m/15m ATR, spread, and plausible stop-market slippage. If the invalidation requires a wider stop, reduce size rather than moving the stop inside normal noise. When a first tranche is losing momentum into support, do not automatically add solely because price reaches the planned level. If fast downside momentum makes passive averaging materially worse, cancel/omit the add and choose a different immediately placeable setup rather than requiring the user to wait for a manual confirmation trigger.\n\n## Native Scale orders\nUser's Hyperliquid UI cannot attach SL/TP while creating a native Scale entry.\nTherefore:\n- do not use native Scale as the default way to open a fresh unprotected position;\n- prefer normal Limit ladder entries with protection;\n- native Scale is primarily for Reduce-Only scale-out / partial profit taking;\n- native Scale add is acceptable only if an already-active Position SL demonstrably covers the additional filled size.\n\nFixed-size reduce-only TPs do not auto-resize after position additions. Recalculate after fills. Position TP/SL can manage the remaining position dynamically.\n\nFor an unattended multi-entry ladder that belongs to one thesis, do not place any pending re-entry beyond another tranche's hard SL. LONG: every planned buy Entry must remain above the shared structural SL. SHORT: every planned sell Entry must remain below the shared structural SL. Otherwise the sequence can become stop-out followed by automatic re-entry, which is contradictory for a single thesis. If different tranches genuinely use different invalidations, treat and label them as separate independent setups; do not present them as one ladder. Prefer a common structural SL for a single unattended ladder, with size adjusted so the shared stop remains acceptable.\n\nWhen the user says they are going to sleep, going offline, or otherwise cannot monitor the position, switch to **sleep/unattended mode**. Default behavior is: cancel all pending entry/add orders and leave only already-open positions with verified hard SL/TP protection. Do not allow a position to increase automatically while the user is asleep merely because a preplanned adverse-price add is reached. An overnight pending entry/add is exceptional and requires explicit justification that (a) the setup remains valid without future discretionary confirmation, (b) worst-case all-fill structural/gap risk is acceptable as one unattended portfolio, (c) correlated risk buckets are already aggregated, and (d) no stale Available Balance is being reused. If those conditions are not proven, no overnight adds.\n\nHyperliquid TP/SL trigger orders are triggered by **Mark Price**, not by the last trade / TradingView candle price. Therefore a candle wick can pass through a TP trigger level without the TP firing if Mark Price never crosses the trigger. When the goal is to capture a trade-price touch/wick at a profit target, prefer a resting Reduce-Only Limit (or Reduce-Only Scale exit) rather than a TP trigger order. Keep trigger-market orders for protection when Mark-based activation is desired.\n\nWhen an intermediate TP executes, cancel stale unfilled add orders if they could rebuild the position after profit taking.\n\n## Margin and leverage\nDecision order:\n1. thesis\n2. entry\n3. structural SL\n4. structural TP\n5. max acceptable loss\n6. notional\n7. leverage + Cross/Isolated\n\nFor fixed notional and SL, increasing leverage does not itself increase stop-loss dollars; it changes margin efficiency and liquidation distance.\n\nFor existing Isolated positions, increasing the leverage setting may NOT automatically release previously allocated isolated margin. Check/remove isolated margin separately when appropriate.\n\nChoose leverage so liquidation remains safely beyond structural SL after allowing for slippage, gaps, Mark moves and volatility. Do not use unnecessarily low leverage if it only locks capital without lowering the defined risk.\n\nCross can be efficient for portfolio-level offsets but shares equity risk. Isolated contains idiosyncratic failures. Neither mode changes the requirement for a structural hard stop.\n\n## SL / TP / sizing\nStructural invalidation first. Do not place stops merely “close” to make risk small.\nUse 4h/1h/15m/5m structure and ATR context; 1m is execution detail, not thesis invalidation.\n\nTPs come from market structure, not the daily dollar goal. Consider resistance/support, prior highs/lows, breakouts, volume/liquidation levels when actually available, and scale-out to reduce near-miss risk.\n\nSize from max loss divided by stop distance. Daily target does not determine notional.\n\n### Volatility-normalized sizing, concentration and rule consent\nUser clarification on 2026-09-07: maintain the minimum Net +$200 daily objective and the existing carry accounting; the objection after the Brent loss was excessive single-name concentration and insufficient full-universe portfolio selection, not permission to abandon the objective or automatically halt the day.\n\nThe assistant's post-loss 0.50%/0.75% per-name, 1.50% aggregate, -2% halving and -3% no-new-risk thresholds were NOT user-approved mandates. They are superseded as binding project rules. Do not retroactively enforce them on existing positions or claim that a numerical reduction follows from user consent. There is no newly approved permanent numerical risk cap; proposed risk budgets must be labeled proposals and disclosed in dollars and equity percentages. Do not substitute guaranteed-return claims or unlimited risk for the removed thresholds.\n\nKeep risk-first sizing: structural invalidation and market-appropriate volatility allowance first, explicit per-setup and portfolio risk budget second, quantity third; margin feasibility is a separate check. Compute planned net stop loss including fees and plausible slippage, plus a separately labeled adverse-volatility/gap scenario. ATR multipliers are scenario assumptions, NOT statistical loss bounds, guaranteed protection, or proof of a profitable setup. Report 15m/1h/4h ATR, structural stop distance, scenario distance, size, notional, modeled stop/scenario loss and equity percentage.\n\nFor each meaningful portfolio decision, compare Long and Short candidates across the complete current active universe, including viable HIP-3 venues. Compare one-name concentration with multiple independent setups; do not impose a fixed number of symbols or force an opposite-direction trade. Aggregate correlated names and common catalysts as one risk bucket; offsetting notional is not proof of a hedge. Compare all-fill/all-stop, single-largest-setup failure, sector-wide failure, whipsaw across long/short legs, partial/non-fill and funding costs. All-target payoff is not expectancy or a guaranteed daily profit. Explain any shortfall versus both today's base target and the carry-inclusive objective without declaring either objective abandoned.\n\nFor adverse-price ladders, evaluate all tranches together before placement. A larger add into deteriorating momentum needs fresh positive justification; otherwise omit/cancel pending adds. Avoid escalating exposure to a recently failed thesis merely to recover its loss. Do not reduce or retain an existing position solely to obey a newly invented percentage: re-evaluate its actual thesis, total portfolio risk and current market structure. Permanent changes to the objective or numerical risk rules require explicit user agreement.\n\n## Fees\nPrefer the user's actual observed fee tier. Known historical observed rates were roughly taker 0.0081% and maker 0.0027%; re-check if needed.\nEvaluate Net PnL after fees, funding and slippage.\n\n## Daily target\nUser target: at least Net +$200 per JST day effective 2026-09-06 JST. The prior base target remains $100 for 2026-09-04 and 2026-09-05, but the user explicitly reset carry at the start of 2026-09-06: the 2026-09-06 required target is a clean $200 with no prior carry-in. From 2026-09-06 onward, new shortfalls roll fully into the next day; profit above the required amount clears carry but does not reduce the next day's applicable base target.\n\n$200 is a minimum realized Strategy Net threshold, not the profit-capacity planning ceiling. A set presented as capable of completing the day must include headroom for non-fills, partial fills, fees, funding, slippage and targets that narrowly miss. As a default completion-capacity gate, require prospective net profit capacity of at least `max(remaining strategy target × 1.50, remaining strategy target + $50)` from the still-open/planned opportunities. In addition, stress-test whether the day can still plausibly reach $200 if the single largest independent setup fails while the others behave as planned. If either the 50% headroom gate or the single-largest-setup-failure test cannot be supported by high-EV setups, label the set as an intermediate cycle and re-screen/redeploy sequentially; never mechanically enlarge risk solely to satisfy the buffer arithmetic.\n\nFor audit, Val Town `account_checker.ts` computes ACCOUNT-WIDE JST 00:00→now strict realized perp Net PnL as `sum(closedPnl) - sum(USDC trading fee) + sum(userFunding delta.usdc)`. It excludes spot fills and does not count unrealized PnL. This is useful for reconciliation but may include the user's independent discretionary trades.\n\nThe project's minimum daily goal must use Val Town SQLite-backed `strategyTracking`, a separate STRATEGY-SCOPED net PnL that includes only fills/funding/fees attributable to registered project OIDs/TIDs plus seeded carried positions. Discretionary trades such as the user's 2026-09-04 SOFTBANK trade must not count toward project performance. Maintain account-wide `targetTracking` and project-wide `strategyTracking` separately; never silently substitute one for the other.\n\nFrom 2026-09-07 JST onward, `account_checker.ts` + val-scoped SQLite is the authoritative remaining-target transport. Its 2026-09-07 starting carry is `$202.965045`, derived from the fixed 2026-09-06 final strict Strategy Net baseline `-$2.965045`. Never distort SL, TP, leverage or position size merely to hit the target, but do check whether planned trades have realistic profit capacity relative to the remaining target. If a proposed order set cannot realistically cover the remaining target, explicitly label it as an intermediate cycle rather than presenting it as a complete daily-target plan.\n\n## Time budget\nBefore committing to a trade, estimate whether its target is a scalp, 1–6 hour trade, or potentially beyond the current day. Do not rely on a single slow trade if its realistic maximum profit is far below the remaining daily target; continue screening independent high-quality opportunities in parallel, without forcing trades.\n\n## Final answer format\nFor each actionable leg include exact:\n- Symbol / direction\n- Core entry and notional/quantity\n- Add entries\n- Max total notional\n- Leverage\n- Cross/Isolated\n- Hard SL and order type\n- TP1/TP2/final TP + allocations\n- Reduce Only\n- order expiry\n- cancellation condition\n- chase cutoff\n- partial-fill handling\n- max stop risk / net RR estimate\n\nDo not leave percentages for the user to convert into actual prices.\n\n## Change management\nUpdate this canonical file only for persistent rule/process changes, tool/transport changes, or newly learned platform mechanics. Do not persist live positions here; live account state is always re-fetched because stored positions become stale.\n\n\n## Persistent private handoff state\nThe latest account-specific handoff is stored in this val's **project-scoped blob storage**, not in a public file.\n\n- Blob key: `project:hyperliquid:authoritative-handoff:current`\n- Read it with the connected Val Town `readBlob` tool using storage `{type:\"val\", val:\"kamibaba/hyperliquid-market-bridge\"}`.\n- It contains the latest handoff snapshot, resolved/closed trade context, the public account address needed by the read-only checker, and the startup checklist.\n- It is **not** exposed by the public market HTTP endpoint.\n- Stored positions/prices are a handoff snapshot only. At the start of a new chat, immediately re-fetch live positions/orders and live markets; live values override the blob.\n- Update the blob at major handoff boundaries or when important project state/rules change. Ordinary fills can be recovered from the live account checker and do not require rewriting the public rules file every time.\n",[256],"b6b94529-0982-458e-87c7-bf368b242635",["D",1788830916590],[28],"59057ef7-5f95-4053-939e-c506715a032f","PIVOT216_RESEARCH_PROTOCOL.md","3dd00504-ab1d-11f1-9b37-1607ee4eb77e",145,["D",1788827750427],"# Pivot v2.16 NVDA research protocol — frozen before outcome inspection\nDate: 2026-09-08 JST. Research only; no exchange execution or schedules.\n\n## Scope\nPrimary instrument: xyz:NVDA trade-price candles from official Hyperliquid candleSnapshot. Compare prior completed cash NVDA RTH daily OHLC (Yahoo independent feed, not TradingView-certified parity) and xyz:NVDA UTC 24-hour OHLC as pivot sources. Cash NVDA 5-minute RTH candles are an additional transferability check, NOT a substitute for XYZ backtesting.\nThe original uploaded Pine is an indicator, not a strategy. This study tests its Traditional pivot formulas and price reactions, not a claim to replicate its full RSI/divergence/alert state machine or TradingView runtime. Cash and XYZ prices are never silently interchanged as execution prices.\n\n## Frozen primary rules\n- D/W/M Traditional formulas exactly as v2.16; intraday default hierarchy excludes Y.\n- Visible-default level set: D P/R1/R2/S1/S2; W P/R1/S1; M closest above/below previous completed close, selected before the next candle. R short, S long, P direction from approach. R/S role reversal is not in this first test.\n- Use only a previous completed pivot period; first partial W/M source period discarded. Cash daily OHLC dates are mapped to the execution chart UTC trading-day key; outside cash RTH, this is an explicit assumed XYZ chart boundary, not certified TradingView bar-by-bar parity.\n- ATR14 Wilder RMA, 50 execution bars warm-up. A new touch requires previous close outside 0.25 current closed-bar ATR on the valid approach side and a signal candle whose price range intersects level +/-0.25 ATR. Floor buffer: 2 ticks, NVDA test tick=0.01.\n- Raw touch variant accepts that event. Rejection variant additionally requires the signal close back on the approach side beyond max(0.10 ATR,2 ticks). These are study definitions, not native entry instructions from Pine.\n- If multiple candidates qualify in a candle, take nearest level to the preceding close; one simulated position at a time in each independently tested variant. No reentry on the same exit candle. This suppresses repeated-contact inflation.\n- Enter next candle open. TP=1 ATR, SL=1 ATR frozen at signal close. 5m horizon=12 bars (1h),15m=16 (4h),1h=6 (6h). Cash-only simulations force exit by regular session close and never enter across a session gap. XYZ permits around-the-clock holding. Reject non-contiguous XYZ paths if price gaps are missing data, not fabricate fills.\n- Both TP and SL within one OHLC bar => stop-first. Gap beyond SL => open-price loss; favorable gaps exit at target, no beneficial fill improvement. Time expiry exits at close.\n- Base round-trip all-in trading friction=6bps (0.06% of entry notional), plus historical funding when available. Sensitivities:0,12,20bps. These are assumptions, not verified historical spreads/fees. Historical Mark, L2 queue, latency, partial fill, large-size capacity and exact real exchange stop triggers are NOT recreated.\n- Report count, TP/SL/timeouts, net-positive percentage, average net R, average net bps, profit factor, max cumulative-R drawdown, break-even friction and ambiguous bars. Never present target payoff as expectancy.\n- Split chronological execution coverage 70/30 by timestamp; exclude trades spanning split. No tuning to the last segment. Weekly-cluster bootstrap intervals (1000 draws) are exploratory; few weeks and retrospective NVDA selection limit inference.\n- Compare 20 deterministic shifted pivot grids per scenario. Each D/W/M period grid shifts by a seeded random +/-0.2..0.8 of its previous-period range. Same level count, formulas, entry rules and execution assumptions. This is a descriptive placebo benchmark, NOT a formal significance test or perfectly exchangeable control.\n- Break down D/W/M, long/short, cash RTH versus other times, early/late period. These are diagnostics, not independently validated strategies. All tested variants must remain in results.\n\n## Persistence\nImmutable raw/normalized data + acquisition SHA256 under project-scoped blob prefix research:pivot216:NVDA:20260908. Keep research results separate from live Strategy ledger, project goals and authoritative trading handoff. Future symbols are configured by exact execution symbol, pivot source, tick, trading-day/session/calendar and fees; never inherit NVDA's session conventions for FX, metals, Korea equities etc.\n",[267],"82c64020-9ca2-49fa-b4f0-4a9879eb8ce7","PIVOT216_RESEARCH_HANDOFF.md","2aacf692-ab1f-11f1-88bd-1607ee4eb77e",151,["D",1788828577335],"# NVDA Pivot v2.16 — completed first research handoff\nDate: 2026-09-08 JST. Research only; no new trade instructions and no exchange order writes. Do not change the separate live trading handoff or Strategy ledger from these results.\n\n## Requested scope\nUser supplied RSI Trigger + D/W/M/Y Pivot v2.16 and requested an NVDA-first backtest of pivot reliability on tradeXYZ, reusable for other symbols. The supplied code is an indicator, not a complete strategy. This research tests pivot-only experimental entries/exits, NOT the full RSI/YENZOU/divergence/star state machine. Original attachment SHA256: 695571c2f7a4107419b7e8dd82c95d437d52d1ba6f2f633ee3135f0ead109fcd; 106709 bytes. The original source was included in the user's local download package, not published in this val.\n\n## Authoritative research records\nVal: kamibaba/hyperliquid-market-bridge; storage type: val (project-scoped).\nBlob prefix: research:pivot216:NVDA:20260908\n- :manifest = acquisition record, asOf 2026-09-08T00:32:12.227Z.\n- :raw: = source payload, request, fetch timestamp and SHA256.\n- :bars: = normalized closed OHLC bars.\n- :summary:5m / :summary:15m / :summary:1h = compact completed scenario results.\n- :result: = full individual simulated trades, statistics, chronological split, cost sensitivities, D/W/M, side and session attributions, 20 shifted-grid controls.\nDataset names: xyz_NVDA_5m (5038), xyz_NVDA_15m (5012), xyz_NVDA_1h (5002), xyz_NVDA_1d (300), cash_NVDA_5m (4681 raw normalized, 4680 clean), cash_NVDA_1d (502).\nScenario IDs: XYZ_{5m,15m,1h}_{CASH_RTH,XYZ_UTC}_{TOUCH,REJECT}; plus CASH_5m_CASH_RTH_{TOUCH,REJECT}. Total14 primary scenarios, 280 controls.\nFiles: PIVOT216_RESEARCH_PROTOCOL.md (frozen before outcome inspection), pivot216_data_collect.ts, pivot216_research_engine.ts, pivot216_research_run.ts. Runner's current saved config is 1h. Do NOT rerun the collector into the same prefix: it uses rolling dates and can overwrite archived inputs. Preserve this baseline and use a new prefix for new dates/symbols.\n\n## Main results\nAll14 scenarios had negative average R and negative average basis-point returns after ASSUMED 6bps (0.06%) round-trip aggregate trading friction. Funding EXCLUDED. This is not a measured all-in actual trading Net and not proof that pivots or the full original indicator cannot work.\nTOUCH primary results, mean cost-adjusted R:\n- Cash-RTH-derived levels on XYZ5m: n193, win37.8238%, meanR -0.689562.\n- Cash-RTH-derived levels on XYZ15m: n353, win47.3088%, meanR -0.324435.\n- Cash-RTH-derived levels on XYZ1h: n714, win47.3389%, meanR -0.154107.\n- XYZ-UTC-derived levels on XYZ5m: n191, win35.6021%, meanR -0.797978.\n- XYZ-UTC-derived levels on XYZ15m: n389, win43.7018%, meanR -0.428866.\n- XYZ-UTC-derived levels on XYZ1h: n747, win46.0509%, meanR -0.194299.\nCash-based REJECT on XYZ5m/15m/1h: n125/271/584, meanR -0.635427/-0.309774/-0.177379. Full14 rows in summary blobs.\nThe14 early70% and late30% segments were also negative at6bps. This is a retrospective chronological check, not genuinely prospective holdout evidence. Different timeframes use different dates AND holding horizons; do not rank them as equivalent period tests.\nXYZ15m CASH_RTH TOUCH by selected level: D n253 win44.2688% meanR-0.381314; W n58 win53.4483% meanR-0.185102; M n42 win57.1429% meanR-0.174219. M equal-notional meanbps is +0.520010, unlike equal-risk R; do not hide the weighting distinction. Small M sample, not validated edge.\nSame scenario by signal-time session: RTH n120 meanR-0.138854; outside_RTH n180 meanR-0.385059; cash_closed n53 meanR-0.538728. Closed days had TP33/53=62.3% but cost-positive23/53=43.4%, illustrating tiny target profits versus assumed friction.\nCash15m TOUCH costs0/6/12/20bps produce meanR +0.009076/-0.324435/-0.657946/-1.102628. 'Cash15m' here means cash-derived pivot levels executed on XYZ15m, NOT cash-price fills.\n20 shifted-grid comparisons: actual XYZ15m CASH_RTH TOUCH at85th percentile; XYZ1h CASH_RTH TOUCH at50th. Descriptive, not a valid formal p-value.\n\n## Completed and incomplete validation\nSuccessful checks in completed runs: Traditional formulas, Wilder ATR, year-crossing Monday week keys, no same-period OHLC selection; execution candle OHLC/order/interval validation had bad/duplicate/gap=0 (normal cash overnight gaps excluded).\npivot216_funding_collect.ts creation was blocked: no funding data acquired. pivot216_research_diagnostics.ts was created, but its run was blocked: do NOT claim its extra alignment tests or execution fixture completed, and do not expect :diagnostics or :quality_checks blobs.\n\n## Material limits\nNo TradingView per-bar parity certificate or exact user's source/settings. Source period mapping UTC assumption; instrument-specific session/time_tradingday semantics need confirmation. Yahoo cash feed is not TradingView feed identity. Visible D/W/M tested, not full hidden R/S depths or yearly levels. R/S role reversals untested. Monthly nearest selection uses prior close to avoid hindsight selection. Study's 1ATR SL/TP and next-open entry are not original indicator rules. No historical funding, Mark triggers, L2 queues, partial fills, measured slippage, latency, large-size capacity or tick rounding. Stop-first within-bar ambiguity is conservative, not proof of real ordering. Weekly-bootstrap intervals are exploratory; multiple comparisons and prior NVDA selection remain.\n\n## Continuation\nFirst reconcile actual TradingView pivot exports with this study's point-in-time levels. Then separately define/freeze original star/RSI confirmation, trend regime and session filters and evaluate on new untouched periods. Preserve all tested variants, including failures. Do not repeatedly optimize SL/TP on this same sample and call the best variant validated.\nFor other symbols reuse the mathematical engine, but parameterize exact execution symbol, pivot-source instrument/price multiplier, source and execution sessions/calendar/holidays/DST, tick and costs. Current NY session diagnostic and the collector are NVDA-specific; do not claim all XYZ assets are plug-and-play.\n",[274],"2e50ecae-483a-4d27-ac23-d7f563d4c130","01a06ff1-37fe-726c-835e-844c9ed57932",["D",1788584474524],"directory",[49],"pulls","branches","forks","envVars",{"_292":252,"_294":62},"connections",{"_292":62,"_293":62},"count","unfilledCount","emptyCount",{"_42":280,"_32":49,"_30":281,"_151":-5,"_40":41,"_39":252,"_153":58,"_37":385,"_13":62,"_33":283,"_34":-5,"_24":25,"_157":-5,"_317":386,"_27":387},{"_42":378,"_32":379,"_30":380,"_151":281,"_40":41,"_39":252,"_153":58,"_37":381,"_13":62,"_33":155,"_34":382,"_24":25,"_157":-5,"_317":383,"_27":384},{"_42":371,"_32":372,"_30":373,"_151":281,"_40":41,"_39":252,"_153":58,"_37":374,"_13":62,"_33":20,"_34":375,"_24":25,"_157":-5,"_317":376,"_27":377},{"_42":273,"_32":274,"_30":275,"_151":-5,"_40":41,"_39":276,"_153":58,"_37":368,"_13":62,"_33":20,"_34":278,"_24":25,"_157":-5,"_317":369,"_27":370},{"_42":266,"_32":267,"_30":268,"_151":-5,"_40":41,"_39":269,"_153":58,"_37":365,"_13":62,"_33":20,"_34":271,"_24":25,"_157":-5,"_317":366,"_27":367},{"_42":263,"_32":28,"_30":31,"_151":-5,"_40":41,"_39":14,"_153":25,"_37":362,"_13":252,"_33":20,"_34":36,"_24":25,"_157":-5,"_317":363,"_27":364},{"_42":255,"_32":256,"_30":257,"_151":-5,"_40":41,"_39":258,"_153":58,"_37":359,"_13":260,"_33":20,"_34":261,"_24":25,"_157":-5,"_317":360,"_27":361},{"_42":247,"_32":248,"_30":249,"_151":-5,"_40":41,"_39":250,"_153":58,"_37":356,"_13":252,"_33":20,"_34":253,"_24":25,"_157":-5,"_317":357,"_27":358},{"_42":239,"_32":240,"_30":241,"_151":-5,"_40":41,"_39":242,"_153":58,"_37":353,"_13":244,"_33":155,"_34":245,"_24":25,"_157":-5,"_317":354,"_27":355},{"_42":232,"_32":233,"_30":234,"_151":-5,"_40":41,"_39":235,"_153":58,"_37":350,"_13":62,"_33":155,"_34":237,"_24":25,"_157":-5,"_317":351,"_27":352},{"_42":224,"_32":225,"_30":226,"_151":-5,"_40":41,"_39":227,"_153":58,"_37":347,"_13":229,"_33":155,"_34":230,"_24":25,"_157":-5,"_317":348,"_27":349},{"_42":216,"_32":217,"_30":56,"_151":-5,"_40":41,"_39":218,"_153":58,"_37":344,"_13":220,"_33":221,"_34":222,"_24":25,"_157":-5,"_317":345,"_27":346},{"_42":209,"_32":210,"_30":211,"_151":-5,"_40":41,"_39":212,"_153":58,"_37":341,"_13":62,"_33":155,"_34":214,"_24":25,"_157":-5,"_317":342,"_27":343},{"_42":202,"_32":203,"_30":204,"_151":-5,"_40":41,"_39":205,"_153":58,"_37":338,"_13":62,"_33":155,"_34":207,"_24":25,"_157":-5,"_317":339,"_27":340},{"_42":195,"_32":196,"_30":197,"_151":-5,"_40":41,"_39":198,"_153":58,"_37":335,"_13":62,"_33":155,"_34":200,"_24":25,"_157":-5,"_317":336,"_27":337},{"_42":188,"_32":189,"_30":190,"_151":-5,"_40":41,"_39":191,"_153":58,"_37":332,"_13":62,"_33":155,"_34":193,"_24":25,"_157":-5,"_317":333,"_27":334},{"_42":180,"_32":181,"_30":182,"_151":-5,"_40":41,"_39":183,"_153":58,"_37":329,"_13":185,"_33":155,"_34":186,"_24":25,"_157":-5,"_317":330,"_27":331},{"_42":173,"_32":174,"_30":175,"_151":-5,"_40":41,"_39":176,"_153":58,"_37":326,"_13":62,"_33":155,"_34":178,"_24":25,"_157":-5,"_317":327,"_27":328},{"_42":166,"_32":167,"_30":168,"_151":-5,"_40":41,"_39":169,"_153":58,"_37":323,"_13":62,"_33":155,"_34":171,"_24":25,"_157":-5,"_317":324,"_27":325},{"_42":159,"_32":160,"_30":161,"_151":-5,"_40":41,"_39":162,"_153":58,"_37":320,"_13":62,"_33":155,"_34":164,"_24":25,"_157":-5,"_317":321,"_27":322},{"_42":148,"_32":149,"_30":150,"_151":-5,"_40":41,"_39":152,"_153":58,"_37":316,"_13":62,"_33":155,"_34":156,"_24":25,"_157":-5,"_317":318,"_27":319},["D",1788706808115],"children",[],[149],["D",1788751088273],[],[160],["D",1788796021112],[],[167],["D",1788830898322],[],[174],["D",1788827982423],[],[181],["D",1788827871383],[],[189],["D",1788828080514],[],[196],["D",1788827527840],[],[203],["D",1788796224033],[],[210],["D",1788700751921],[],[217],["D",1788751330350],[],[225],["D",1788796088059],[],[233],["D",1788742421944],[],[240],["D",1788585922073],[],[248],["D",1788795680414],[],[256],["D",1788830916590],[],[28],["D",1788827750427],[],[267],["D",1788828577335],[],[274],"eecdc355-933e-4c99-99e5-481180e6832a","favicon.svg","01a06ff1-37fe-726c-835e-801b891f2122",["D",1788584474524],"\n",[],[49,372],"94aaa291-56be-425b-8726-a9d3e3c0f5eb","counter.ts","01a06ff1-37fe-726c-835e-7c219a514093",["D",1788584474524],"const count = document.getElementById(\"count\")!;\nconst button = document.getElementById(\"increment\")!;\n\nbutton.addEventListener(\"click\", async () => {\n const res = await fetch(\"/api/increment\", { method: \"POST\" });\n const data = await res.json();\n count.textContent = String(data.count);\n});\n",[],[49,379],["D",1788584474524],[296,297],[49],["D",1788584474524],["D",1789491713377],{"_99":98,"_106":102},{"_99":98,"_100":98,"_101":102,"_18":54,"_103":104,"_105":25},{"_13":88,"_32":89,"_42":41,"_90":43,"_91":-5,"_92":-5,"_93":-5,"_94":-5,"_95":-5,"_96":44,"_97":98},"ctx",{"_81":-5,"_126":426,"_427":428,"_429":430,"_431":25,"_432":25,"_433":25,"_434":58,"_435":-5,"_436":-7,"_437":-7,"_438":62,"_439":25,"_440":25,"_441":-5,"_442":25,"_443":25,"_444":25,"_445":-5},"ENV",{"_397":398,"_399":400,"_401":89,"_402":403,"_404":405,"_406":407,"_408":409,"_410":411,"_412":413,"_414":415,"_416":417,"_418":419,"_420":421,"_422":423,"_424":425},"REACT_APP_SENTRY_DSN","https://586020d86f05489db63a554690acf1e9@o1398359.ingest.sentry.io/6724499","RENDER_GIT_COMMIT","e87eb04e707cf64e7e532c8e8ec395f209e17f15","RENDER_GIT_BRANCH","POSTHOG_API_TOKEN","phc_nmoyuAkStEZuUp69rYlg1DZgrIt19Gx0Y4Hwf8tGgJP","POSTHOG_API_HOST","https://wart.val.town","WEB_API_TEMPLATE","https://{val}.web.val.run","APP_URL","https://www.val.town","MODULE_URL","https://esm.town","VANITY_WEB_API_TEMPLATE","https://{label}.val.run/","DENOLS_URL","https://lsp.val.town","NODE_ENV","production","CLOUDFLARE_TURNSTILE_SITE_KEY","0x4AAAAAACtaIZJH7s9rBMZ6","GOOGLE_CLIENT_ID","875989696899-jtsqkhlbdv5jrlsk8631p7ipjg3c6o62.apps.googleusercontent.com","STRIPE_PUBLISHABLE_KEY","pk_live_51N2zvWEpZTTLipIkVEwXrUG9AT0yeOOEYxszarZT1jpsZV58Ly6wlg3SltZpHN26CUalOBmUSY7adQRQiSkCZz2500qSXNkf8Y","API_URL","https://api.val.town",[],"preferencesKeybindings","default","preferencesTheme","light","preferencesFixers","preferencesCopilot","preferencesSelectionAi","preferencesCloseBrackets","preferencesChatPanelState","preferencesDprint","preferencesDenofmt","notificationCount","flagDenols","isBusinessOrgMember","businessOrgHandle","isBusinessOrgAdmin","isHigherTownieTier","navCollapsed","currentOrgId"]

Diff · v156