{"assessments":[],"deployments":[],"fuzz":[],"identity":{"adapter":"0xde152afb7db5373f34876e1499fbd893a82dd336","chainId":1,"collection":"0x0000ec93127baa929e58e97dd0095a2bfb38ec1d","registry":"0x8004a169fb4a3325136eb29fa0ceb6d2e539a432"},"interpretation":"Records acceptance and evidence. Neither completion nor an AI assessment establishes correctness, safety, or independent review.","jobId":"0433e898-1166-471c-b899-48e3a639cbc2","kind":"skill:research-report","nodes":[{"acceptedSubmissionHash":"d65c15c7d72806cf4bee3d03a8f7ded9c8d53631c8f095cf5d1c4015edd63ad8","dependsOn":[],"execution":{"network":true,"profile":"none","requires":["network"],"skillHash":"3ddca93330036359dd721585e58e67820336a0398b7927b3c89369d6134f30f6","skillId":"research-report","tools":[]},"key":"research_report","kind":"code","role":"implement","skillHash":"3ddca93330036359dd721585e58e67820336a0398b7927b3c89369d6134f30f6","skillId":"research-report","state":"accepted"}],"objective":"Tune the ERL-100 indicator (Excursion Restitution Ledger) to work its best for ETH perps at exactly 100x leverage on Papertrade (papertrade.xyz), on the 1m, 3m, 5m and 15m timeframes, and deliver drop-in settings for our site.\n\nERL-100 is fully specified in section 4 (and the live liquidation-map veto in section 5) of the attached papertradeResearch input, public at https://api.imd.fun/artifacts/6e8652275deb7261fcc223d0cbcb13342ab70f0f84d1a3063dff85857a9af858 . Keep its core construction (surviving wick ledger erased by later bodies, 24-bar recency-weighted cohort, sweep-and-reclaim of the 12-bar closing frontier, support coverage J, upper-corridor obstacle A, payout-adjusted netRR, ERL = J*(1-A)*netRR, no lookahead, no repaint, bar-close entries at the next bar open). You may tune every parameter and may add exit-management rules if they help, but say which are additions.\n\nBaseline we measured with the reference implementation at default parameters on OKX ETH-USDT-SWAP candles (ETH 1m: 30 days; 5m: 90 days), fees and Papertrade's winning-payout curve applied, R = net PnL / risked stop: ETH 1m 204 trades, 43.1% win, -0.126 R/trade, PF 0.61, exits TIME 110, FRONTIER_FAIL 55, STOP 36, TARGET 3. ETH 5m 267 trades, 39.7% win, -0.174 R, PF 0.55, exits TIME 138, STOP 79, TARGET 6, FRONTIER_FAIL 44. Zero busts. Diagnosis: the 6-bar time stop closes most trades before the 3R target can be reached. An exploratory in-sample check of maxHoldBars on ETH 1m gave 12 bars -0.045 R, 24 bars -0.094 R, 48 bars -0.024 R (PF 1.01); on 5m longer holds were worse. Treat those as leads only.\n\nWork:\n1. Data: at least 120 days of public 1m ETH perp candles (Hyperliquid candleSnapshot if reachable, else OKX ETH-USDT-SWAP history-candles or Binance ETHUSDT USD-M klines), aggregated to 3m, 5m and 15m. Use BTC only as an optional confirmation filter.\n2. Economics: exactly 100x; Papertrade fees, asymmetric impact, winning-payout holdback curve (use the worse of the standard and off-hours curves), full losses, bust distance, plus a 1 bp adverse-execution sensitivity per side.\n3. Search per timeframe: maxHoldBars, cooldownBars, targetRiskMultiple, targetMin/targetMax, riskMin/riskMax, stopPad, sweepMin/sweepMax, reclaimBuffer, confirmMin, rangeMax, supportMin, obstacleMax, netRRMin, budgetMin, ledgerBars, frontierBars; exit variants (break-even move, trailing after +1R, partial at 1-1.5R, frontier-fail rule on/off); filters (higher-timeframe trend, ATR percentile, hour of day UTC, long-only or short-only if justified on training data only).\n4. Robustness: chronological split with an untouched final holdout of at least 30 days, walk-forward with at least three folds, minimum 60 holdout trades per timeframe, parameter-neighbourhood stability (prefer plateaus), and a multiple-testing caveat. Report holdout numbers as the headline.\n5. Pick ONE best setting per timeframe (1m, 3m, 5m, 15m) even if still negative, and name the single best timeframe overall for ETH.\n\nDeliver report.md with, per timeframe: exact parameters, exact long and short rules including any added exits, holdout trades, win rate, mean R, profit factor, max drawdown, bust rate, exit-reason counts, fold-by-fold results, and an honest statement of whether any edge survives. Deliver erl-eth-settings.json shaped exactly as {\"version\":1,\"symbol\":\"ETH\",\"leverage\":100,\"best\":\"<tf>\",\"timeframes\":{\"60\":{...},\"180\":{...},\"300\":{...},\"900\":{...}}} where keys are timeframe seconds and each value has \"params\" (an object using ONLY these field names: ledgerBars, frontierBars, sweepMin, sweepMax, reclaimBuffer, confirmMin, stopPad, riskMin, riskMax, rangeMax, targetRiskMultiple, targetMin, targetMax, supportMin, obstacleMax, netRRMin, budgetMin, latencyReserve, maxHoldBars, cooldownBars; fractions of price, not percent), \"exits\" (precise description and numeric fields of any added exit rule, or null), \"filters\" (precise numeric description, or null), and \"holdout\" {trades, winRate, meanR, profitFactor, maxDrawdownR, bustRate, from, to}.","parentJobId":null,"planHash":"fa70068fadb374fc6e62cfd42ab425baf20c0b110224b23b27cff20f09eadf47","previousHash":"396a04bb1b0d76842cd15805e133d0c4ad386ac6d7c21aa21b96112542019154","projectId":"0433e898-1166-471c-b899-48e3a639cbc2","publication":{"commit":"10d4ab2eae3c072f60138bfec57e07214fd79d32","deliveredAt":"2026-11-04T05:18:15.090Z","repoUrl":"https://github.com/Identity-md/research/blob/main/jobs/0433e898-1166-471c-b899-48e3a639cbc2/_identitymd/README.md"},"receiptIdentity":{"adapter":"0xde152afb7db5373f34876e1499fbd893a82dd336","chainId":1,"collection":"0x0000ec93127baa929e58e97dd0095a2bfb38ec1d","registry":"0x8004a169fb4a3325136eb29fa0ceb6d2e539a432"},"registry":"0xb6d0a187b050fa5bb0b87033a203f37becf4a775","research":[],"schema":"identitymd-work-v1","signals":[],"site":null,"snapshotHash":"3e6bbdcfd47ec4a2cc4af4a064d9c827c39baeafa59694762bbbb05ec3ee34dd","state":"completed","submissions":[{"artifacts":[{"bytes":33361,"hash":"a70a06e418d688e36c5de65c8a402fef4dec7d47d055e5dec132c256fd44fe37","mediaType":"text/markdown","name":"report","path":"artifacts/report.md"},{"bytes":16132,"hash":"7311aa30dd79ab70e1da19a514a9bc03af3390b2eaa8e9ac321648c98ee5360f","mediaType":"application/json","name":"erlEthSettings","path":"artifacts/erl-eth-settings.json"}],"attempt":1,"bundleHash":"0d9135e8052287df27c6fbbee2d8cf9864a83328912634cc537697defe1707c6","device":"0b9c1289a6def0a1","findings":[],"hash":"d65c15c7d72806cf4bee3d03a8f7ded9c8d53631c8f095cf5d1c4015edd63ad8","nodeId":"51525903-55cc-4329-8e7a-4ddfda55b72f","outcome":"completed","summary":"Repaired the bundle: losslessly compressed candles, removed redundant ZIPs, and preserved offline reproduction.\n\n- Compressed source bundle: **7,027,669 bytes**, below 8 MiB.\n- Offline verification: **all checks passed**.\n- Deliverables: [report](artifacts/report.md), [settings](artifacts/erl-eth-settings.json).\n\nResults remain unchanged: **3m is best, but no timeframe demonstrated a surviving edge.**","treeHash":"7c100cc4b411cb99eb1d15f6560e3743b06b9d7c","usage":{"cachedInputTokens":3093504,"inputTokens":98358,"model":null,"outputTokens":54578,"runtime":"codex","turns":12,"wallClockMs":1330574}}],"verification":[{"checks":[],"detail":"paths and tree verified; no suite was run for this kind of work","evaluation":"structural","profile":"none","status":"accepted","submissionHash":"d65c15c7d72806cf4bee3d03a8f7ded9c8d53631c8f095cf5d1c4015edd63ad8","verifiedTreeHash":"7c100cc4b411cb99eb1d15f6560e3743b06b9d7c","verifierVersion":"0.1.0+df310a90"}]}