Stats

Primary L20 remains fixed. L10/L40 are sensitivity only; no best-lookback selection after seeing holdout. Split-only excludes dividends and therefore is PRICE-only performance, not shareholder total return; changing adjustment also changes signal dates and exits. Same2ADRstop/3Rtarget/40sessions,1bpfees+5/15bpslip, vectorbt open/execution/close marks, train2021-24/test2025-Sep11. Results (return/DD/win/N; open marks excluded from win): [{"L":10,"adj":"all","start":"2021-01-01","slip":5,"return":173.0235627099403,"DD":36.405626540674774,"win":44.827586206896555,"N":29,"open":0},{"L":10,"adj":"all","start":"2021-01-01","slip":15,"return":157.63865441192326,"DD":36.95610214104493,"win":44.827586206896555,"N":29,"open":0},{"L":10,"adj":"all","start":"2025-01-01","slip":5,"return":13.928076401532453,"DD":23.0937318662049,"win":38.46153846153847,"N":13,"open":1},{"L":10,"adj":"all","start":"2025-01-01","slip":15,"return":10.893281540409827,"DD":23.55378842355995,"win":38.46153846153847,"N":13,"open":1},{"L":40,"adj":"all","start":"2021-01-01","slip":5,"return":241.46079024917464,"DD":29.232458861948885,"win":55.00000000000001,"N":20,"open":0},{"L":40,"adj":"all","start":"2021-01-01","slip":15,"return":228.0719069395432,"DD":29.51496394570704,"win":55.00000000000001,"N":20,"open":0},{"L":40,"adj":"all","start":"2025-01-01","slip":5,"return":-6.465847960939863,"DD":23.915974669682882,"win":22.22222222222222,"N":9,"open":1},{"L":40,"adj":"all","start":"2025-01-01","slip":15,"return":-8.226130469467986,"DD":24.371112535718677,"win":22.22222222222222,"N":9,"open":1},{"L":20,"adj":"split","start":"2021-01-01","slip":5,"return":82.81082072400874,"DD":37.73404618555906,"win":40.74074074074074,"N":27,"open":0},{"L":20,"adj":"split","start":"2021-01-01","slip":15,"return":73.20083083760669,"DD":38.53820424160471,"win":40.74074074074074,"N":27,"open":0},{"L":20,"adj":"split","start":"2025-01-01","slip":5,"return":-6.394336303977532,"DD":25.66153871800738,"win":27.27272727272727,"N":11,"open":1},{"L":20,"adj":"split","start":"2025-01-01","slip":15,"return":-8.522607403030591,"DD":26.84149058960168,"win":27.27272727272727,"N":11,"open":1}]. Source https://runway.adityaperswal.workers.dev/data/stock-bars . All exact means/CIs/exitcounts recorded26226-26265. The test period is shared across variations; variations are not independent evidence.

Primary L20 remains fixed. L10/L40 are sensitivity only; no best-lookback selection after seeing holdout. Split-only excludes dividends and therefore is PRICE-only performance, not shareholder total return; changing adjustment also changes signal dates and exits. Same2ADRstop/3Rtarget/40sessions,1bpfees+5/15bpslip, vectorbt open/execution/close marks, train2021-24/test2025-Sep11. Results (return/DD/win/N; open marks excluded from win): [{"L":10,"adj":"all","start":"2021-01-01","slip":5,"return":112.67295913234567,"DD":32.16931880264319,"win":37.5,"N":40,"open":0},{"L":10,"adj":"all","start":"2021-01-01","slip":15,"return":96.32186799489786,"DD":32.575084814074316,"win":37.5,"N":40,"open":0},{"L":10,"adj":"all","start":"2025-01-01","slip":5,"return":160.4555454070435,"DD":22.96278069120994,"win":53.333333333333336,"N":15,"open":1},{"L":10,"adj":"all","start":"2025-01-01","slip":15,"return":152.5055331700635,"DD":23.423620604047237,"win":53.333333333333336,"N":15,"open":1},{"L":40,"adj":"all","start":"2021-01-01","slip":5,"return":36.95938733593715,"DD":30.152624809063354,"win":34.61538461538461,"N":26,"open":0},{"L":40,"adj":"all","start":"2021-01-01","slip":15,"return":30.019492658679205,"DD":30.570454769486478,"win":34.61538461538461,"N":26,"open":0},{"L":40,"adj":"all","start":"2025-01-01","slip":5,"return":-5.8818647506624355,"DD":26.356540897697922,"win":23.076923076923077,"N":13,"open":1},{"L":40,"adj":"all","start":"2025-01-01","slip":15,"return":-8.3889660922706,"DD":26.943338884677125,"win":23.076923076923077,"N":13,"open":1},{"L":20,"adj":"split","start":"2021-01-01","slip":5,"return":31.916768191091947,"DD":31.703362138101866,"win":32.25806451612903,"N":31,"open":0},{"L":20,"adj":"split","start":"2021-01-01","slip":15,"return":23.986304575835238,"DD":32.11191552206013,"win":32.25806451612903,"N":31,"open":0},{"L":20,"adj":"split","start":"2025-01-01","slip":5,"return":85.24123287186694,"DD":24.042364503408532,"win":47.05882352941176,"N":17,"open":1},{"L":20,"adj":"split","start":"2025-01-01","slip":15,"return":78.87010986338649,"DD":24.496746300099115,"win":47.05882352941176,"N":17,"open":1}]. Source https://runway.adityaperswal.workers.dev/data/stock-bars . All exact means/CIs/exitcounts recorded26226-26265. The test period is shared across variations; variations are not independent evidence.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. On a surviving day reaching2R, sell quarter and make stop breakeven next day; test full stop/target first, so intraday ordering and 15minute management not reconstructed. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 1 open mark; N11 and win36.36363636363637% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-4.524399609507263,5.843595649618188]% is uncertainty not proof of zero edge. 1 targets/7 stops/3 timeouts; buyhold74.95342579176152%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=STNG&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Holdout/stress weakness prevents treating the screen as an established stock edge.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. On a surviving day reaching2R, sell quarter and make stop breakeven next day; test full stop/target first, so intraday ordering and 15minute management not reconstructed. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 0 open mark; N27 and win44.44444444444444% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-1.5149595070621042,8.708828524196868]% is uncertainty not proof of zero edge. 6 targets/14 stops/7 timeouts; buyhold377.8375088751394%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=STNG&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Holdout/stress weakness prevents treating the screen as an established stock edge.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. Full fixed exits; no discretionary trailing. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 1 open mark; N11 and win27.27272727272727% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-5.197660030798593,5.848221916807604]% is uncertainty not proof of zero edge. 1 targets/7 stops/3 timeouts; buyhold74.95342579176152%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=STNG&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Holdout/stress weakness prevents treating the screen as an established stock edge.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. Full fixed exits; no discretionary trailing. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 0 open mark; N26 and win46.15384615384615% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-0.6130553619346064,10.616290391886833]% is uncertainty not proof of zero edge. 6 targets/12 stops/8 timeouts; buyhold377.8375088751394%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=STNG&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Holdout/stress weakness prevents treating the screen as an established stock edge.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. On a surviving day reaching2R, sell quarter and make stop breakeven next day; test full stop/target first, so intraday ordering and 15minute management not reconstructed. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 1 open mark; N18 and win44.44444444444444% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-2.240298785942834,8.466477223104828]% is uncertainty not proof of zero edge. 6 targets/11 stops/1 timeouts; buyhold280.5877803557618%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Positive result retains watch; must resolve upcoming raw-stop distribution mechanics.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. On a surviving day reaching2R, sell quarter and make stop breakeven next day; test full stop/target first, so intraday ordering and 15minute management not reconstructed. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 0 open mark; N33 and win42.42424242424242% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-1.7832491318441552,5.371609867705768]% is uncertainty not proof of zero edge. 7 targets/24 stops/2 timeouts; buyhold191.44447020635292%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Positive result retains watch; must resolve upcoming raw-stop distribution mechanics.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. Full fixed exits; no discretionary trailing. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 1 open mark; N17 and win47.05882352941176% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-1.3523231382960088,10.722254655961946]% is uncertainty not proof of zero edge. 7 targets/9 stops/1 timeouts; buyhold280.5877803557618%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Positive result retains watch; must resolve upcoming raw-stop distribution mechanics.

Frozen before this run's results: prior20close breakout AND SMA50, next session open; stop2*priorADR20, target3R, max40sessions including entry. Prior2020warmup. Full fixed exits; no discretionary trailing. Gap exits use open, ambiguous daily bars stopfirst, entryday exits allowed (observed 0). Costs1bpfee+5bpslippage/side, stress15bpslip. vectorbt long-only fractional full-equity portfolio reconciled to manual cash including entryfee, marks ordered open/execution/close. Return includes 0 open mark; N33 and win33.33333333333333% exclude open outcomes. Mean bootstrap10000 IID resamples seed91206, 95%CI[-1.9720576853637237,6.135799986935027]% is uncertainty not proof of zero edge. 9 targets/21 stops/3 timeouts; buyhold191.44447020635292%. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=sip&adjustment=all , retrievedSep12/endSep11T23:40Z. Adjustmentall returns model synthetic dividend-adjusted asset, not cash dividend payment timing or executable raw-price fills. Training and test separated flat; test period previously inspected for other rules, thus chronological holdout for this frozen specification but not globally unseen evidence. Quantifies payoff shape, not whole-share/15min implementation. Positive result retains watch; must resolve upcoming raw-stop distribution mechanics.

Edge test: ten issuer-dated scheduled AMC earnings March2024–June2026, daily pre-release close to next-session close, median absolute4.700378%, signed mean-3.027709%. At Sep11 S2.53, Oct16 K2.5 call .21(Sep11catalog) /put .16(Sep10catalog, stale) plus assumed .05/share roundtrip costs require +9.090909%/-9.486166%; both1/10=10%, below40%. Old frozenN9 remains1/9,median5.494505%; preserve it separately. No documented3x expectedoptionpayoff. Source option_contracts and https://runway.adityaperswal.workers.dev/data/stock-bars ; /data/option-bars. Issuer Aug25 https://hightideinc.com/high-tide-to-announce-third-fiscal-quarter-2026-financial-results/ says 'after financial markets close on Monday, September 14, 2026'. Oct16expiry32daysafter; defaultSep11RTHexit already elapsed. Hurdles exceednewmedian4.390531/4.785788pp, noholdexception. Closes are not executable asks; intrinsic is not actualoptionPnL. 1. vectorbt926SIPallbarsJan2023–Sep11 2026. BuyT-2/3/5USsessionopen,exitT-1close beforeAMCdate;1000initial,fees10bp/slippage10bpeachside,stress30bpslippage. AddedSep2025quartertotrainingonly; holdout2026N3alreadyexaminedpreviousruns, retrospectiveauditnotfreshindependentOOS. Priortrain-selectedT5preserved, nochoosingrulefromholdout. 2. T5trainN7return+3.867305%,DD11.915644%,win42.857143%;holdoutN3return-2.631698%,DD5.363985%,win66.666667%;stressholdout-3.793140%,win33.333333%. Smallwinsdonotcompensateloss. CorrectedtrainT3nowhighest+4.438587%butholdout-4.762379%/stress-5.898406%; unstablemodelselection. T2train+3.842905%,holdout+2.134481%/stress+.916186%; knownholdoutpositivitycannotjustifyposthocselection. Full12rowssavedobservations. 3. RequiredCLI momentum20/days1000 diagnostic685bars,N27return26.210635%,DD44.737554%,win29.629630%,buyhold55.246914%; differentrule,noheldoutCLIwindow,notoptionevidence. Nohistoricaloptionreturn/DD/winrateavailable; nooptionsorderfromstockproxy. Costsassumedratherthanmeasured; dailycloseexecutionproxydoesnotvalidate15minuteoptionstops.

Required CLI: node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback20 --days980. BTC981bars: return46.460751%,DD14.692428%,23/67wins(34.328358%);ETH981bars40.808012%,DD24.943256%,17/55wins(30.909091%);SOL748bars72.151727%,DD18.795394%,19/36wins(52.777778%). Buy-and-hold BTC75.661593%,ETH12.116248%,SOL-35.313196%. Premise1 inspection of /app/agent/src/backtest/strategies.ts andengine showsprior20CLOSE screen/sameclose policy; no explicitfees orheldoutwindow inthisdiagnostic. Premise2 itincludescurrentpartialdailybarandSOLshorterhistory; itisnotHIGHbreakout/pullbacknext-open net3R20dayrule. Conclusion recordCLIasdiagnostic,notvalidationorpermissiontochooseSOLbecausegrossreturnhighest. Independentlycosted200cellsandchronologicalholdoutrecordedinC91204analyses addressactualproposedpolicy. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Eightpredefinedsensitivitycells reconstructETHdailyOHLCfromall24calendarhours; incompletehourdaysbecomeNA,strict20dayATR/HIGHrequiresfullhistory. Retainoriginalfilesand15minmanagedexecution. Premise1 historicaldailyversushourlydifferencesexist:2025Dec2ETHopen3020.1375rawversus2798.84houraggregate(-7.3274%);2026all253completeETHdaysOHLCmatch exactly. Premise2 coherentdailyinputchangessomeavailabilityaroundmissingMarchday; thisisdata/controlchange,notalpha. Hourlybaselineunchanged:train42.312262%,DD37.521903%,7/13wins,oneopen;holdout19.206747%,DD14.542745%,3/3wins,oneopen;stressholdout17.549943%. Dailyholdoutchangesfrom-.292829%N3to-1.925730%N2,stress-2.707198%;dailytrainingunchanged-27.126920%. Conclusion aggregationissuesdonotexplainawaypositivehourlycandidate; dailynegative/cellselection/smallNandnonpristineholdoutstilllimitdeployment. All8cellsrecorded,bringingpolicycomparators200 plusCLI3diagnostics; not200independentstrategies/events. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204aggregate_control.py.

ThreeETHhourlyhalfATRmanagedholdouttrades March5–25,April9–29,July7–27 have6/3/9missing15mincheckpoints(total18); thusliteralcompletecoverageclosedN0. Refinedcounterexample toblanketquarantine: for14missingpoints primaryhourhigh/low boundsexclude allinitialstop,net3Rtargetandgross+2Rtouches; missingMarch12 17hour'sfourpoints usecompleteddayrange2018.698–2094.2115 versusMarchtradestop1791.83424/+2R2603.77278/target2924.123397,alsono possibletrigger. Noimputedpricesorzeroreturns. Conditionalonbroadertradebar extrema coveringmissingintervals, all18missingchecks cannotchangethese3tradepaths; modeledholdout+19.206747%,DD14.542745%,3/3winners,1openisretained. Strongcounterargument:broaderbarsaretradebounds,nothistoricalexecutablequotes; theydonotprovefills,checkavailabilityorinfinite liquidity. TrainingmeanCIstillcrosses0. Validity: boundproofsupportsno modeledthresholdtouch,notglobaldataqualityortradingedge. Conclusion do notciteN0literalcoveragealoneasrejection; keepcandidateprospectiveandseparatefreshentry gate. Fullboundsrecordedobservations,C91204gapbounds.py. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

CLI run node /app/agent/dist/backtest.js --symbols FTK --strategy breakout --lookback20 --days2080 --stop-pct10 --target-pct30 (actual invocation uses spaces between flag/value) returned+13.7774%,DD49.9074%,N66/win33.3333%. It usesIEX/noexplicitadjustment orcosts andisnotindependentholdout; cannotoverride negativeSIPresults. https://runway.adityaperswal.workers.dev/data/stock-bars . Fixed10/20/40sensitivity usesSIPadjustmentall,2020warmup,train2021-24/holdout2025-Sep11'26,fees1bp/slip5or15; all12negative. Correctopen/closemarks supersede earlierDDonly. L10trainingDD86.7031/87.1476,holdout46.9088/47.8559; L40training59.3906/60.6103,holdout31.7072/32.8585. L10sample24train/15holdout each1unfinished; L40N16/11flat; unchangedreturns in2847-50/2855-58. No favorablelookbackselection orstrategyadoption. CLI sample66unadjustedtradehistory andSIPcostedsamples answerdifferentquestions.

Supersedes drawdown only in primary analyses2851-2854; retains frozen2531 rule, samples, fills and returns. Original two-slot marks used closing prices in both slots, incorrectly marking exit-day close before modeled intradayexit. Corrected vectorbt price marks alternate actual open/close, while entry occurs slot0 and exit slot1; open/close marked drawdown is not full intraday worstcase. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FTK&timeframe=1Day&feed=sip&adjustment=all . L20closebreakout+SMA50, nextopen,2ADRstop/3Rtarget,40sessions,1bpfees and15bpslip/side,2020warmup,2025-01-01 to2026-09-11. 14 completed/0 open; stopfirst ambiguousbar, gapopenexits, entrydayexits retained; compoundedcashreturn reconciliationpassed. Still rejects trading; 15min and+2R management absent. All12sensitivityrevisedDDtable in latest observation; no parameterretuning.

Supersedes drawdown only in primary analyses2851-2854; retains frozen2531 rule, samples, fills and returns. Original two-slot marks used closing prices in both slots, incorrectly marking exit-day close before modeled intradayexit. Corrected vectorbt price marks alternate actual open/close, while entry occurs slot0 and exit slot1; open/close marked drawdown is not full intraday worstcase. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FTK&timeframe=1Day&feed=sip&adjustment=all . L20closebreakout+SMA50, nextopen,2ADRstop/3Rtarget,40sessions,1bpfees and5bpslip/side,2020warmup,2025-01-01 to2026-09-11. 14 completed/0 open; stopfirst ambiguousbar, gapopenexits, entrydayexits retained; compoundedcashreturn reconciliationpassed. Still rejects trading; 15min and+2R management absent. All12sensitivityrevisedDDtable in latest observation; no parameterretuning.

Supersedes drawdown only in primary analyses2851-2854; retains frozen2531 rule, samples, fills and returns. Original two-slot marks used closing prices in both slots, incorrectly marking exit-day close before modeled intradayexit. Corrected vectorbt price marks alternate actual open/close, while entry occurs slot0 and exit slot1; open/close marked drawdown is not full intraday worstcase. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FTK&timeframe=1Day&feed=sip&adjustment=all . L20closebreakout+SMA50, nextopen,2ADRstop/3Rtarget,40sessions,1bpfees and15bpslip/side,2020warmup,2021-01-01 to2024-12-31. 26 completed/0 open; stopfirst ambiguousbar, gapopenexits, entrydayexits retained; compoundedcashreturn reconciliationpassed. Still rejects trading; 15min and+2R management absent. All12sensitivityrevisedDDtable in latest observation; no parameterretuning.

Supersedes drawdown only in primary analyses2851-2854; retains frozen2531 rule, samples, fills and returns. Original two-slot marks used closing prices in both slots, incorrectly marking exit-day close before modeled intradayexit. Corrected vectorbt price marks alternate actual open/close, while entry occurs slot0 and exit slot1; open/close marked drawdown is not full intraday worstcase. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FTK&timeframe=1Day&feed=sip&adjustment=all . L20closebreakout+SMA50, nextopen,2ADRstop/3Rtarget,40sessions,1bpfees and5bpslip/side,2020warmup,2021-01-01 to2024-12-31. 26 completed/0 open; stopfirst ambiguousbar, gapopenexits, entrydayexits retained; compoundedcashreturn reconciliationpassed. Still rejects trading; 15min and+2R management absent. All12sensitivityrevisedDDtable in latest observation; no parameterretuning.

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return10.5767%,DD35.0750%,win57.14285714285714%,closedN7,open0. HoldoutJan1-Sep11 2026 return-4.6600%,DD21.7915%,win50%,closedN2,open0; stress-5.4197%. Immediatebreakout control samecost/exit holdout4.0652%,N6. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. SOL history startsAug2024, shortenedtraining. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return27.9720%,DD19.8678%,win75%,closedN4,open0. HoldoutJan1-Sep11 2026 return-12.1890%,DD18.3921%,win33.33333333333333%,closedN3,open0; stress-13.0386%. Immediatebreakout control samecost/exit holdout-23.7294%,N7. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. SOL history startsAug2024, shortenedtraining. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return42.3123%,DD37.5219%,win53.84615384615385%,closedN13,open1. HoldoutJan1-Sep11 2026 return19.2067%,DD14.5427%,win100%,closedN3,open1; stress17.5499%. Immediatebreakout control samecost/exit holdout14.8754%,N6. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return-27.1269%,DD37.0512%,win37.5%,closedN8,open1. HoldoutJan1-Sep11 2026 return-0.2928%,DD21.3949%,win66.66666666666666%,closedN3,open0; stress-1.4822%. Immediatebreakout control samecost/exit holdout-12.7035%,N5. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return-9.9610%,DD33.2952%,win43.75%,closedN16,open0. HoldoutJan1-Sep11 2026 return1.1932%,DD10.1154%,win66.66666666666666%,closedN3,open1; stress0.0072%. Immediatebreakout control samecost/exit holdout4.2262%,N6. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen notes2535/2536. 15-minute bar-open point-check proxy; take25% at gross+2R and set remainingstop toentry, retain net3R target; checks before newentry. HalfATR pullback within3days of first prior20dayHIGH breakout; stop2 prior-completed-day ATR20,20daytimeout. Buyfee25bp credited incoin/sellfee25bp inUSD;10bpslip baseline/30stress eachside. Training2024-25 return-14.7836%,DD26.5712%,win20%,closedN5,open0. HoldoutJan1-Sep11 2026 return6.8464%,DD8.7633%,win66.66666666666666%,closedN3,open0; stress5.5719%. Immediatebreakout control samecost/exit holdout-8.3834%,N6. Premise1 timeframes/policies differ; no substitution for older frozenETHmodel. Premise2 smallsamples and known2026hypothesisdiscovery do not establishprospectiveedge. Premise3 missing observations are skipped/entriescancelled, notzero returns; fullcoverage audit separate. vectorbt.from_orders cash reconciled with explicit creditedfee conversion; end openpositions marked without inventedclose. Source15minopen is a tradebar proxy, not historical executablequotes/actualvenuefills; gaps can delay exits. Conclusion research comparator, no automaticentry; all192cells retained in observations. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .