Stats

Required backtest CLI sanity: node /app/agent/dist/backtest.js --symbols VLO,MPC,PSX --strategy mean-reversion --lookback20 --entry2 --exit0 --days1460 . Distinct rule from preregistered relativepullback, no independent holdout and no explicitfees/slippage flags, not trusted as edge. Source https://runway.adityaperswal.workers.dev/data/stock-bars . Preserve output rather than combine its return/sample with vectorbt. No action authorized.

Required backtest CLI sanity: node /app/agent/dist/backtest.js --symbols VLO,MPC,PSX --strategy mean-reversion --lookback20 --entry2 --exit0 --days1460 . Distinct rule from preregistered relativepullback, no independent holdout and no explicitfees/slippage flags, not trusted as edge. Source https://runway.adityaperswal.workers.dev/data/stock-bars . Preserve output rather than combine its return/sample with vectorbt. No action authorized.

Premise1: Sep9completedSIPdailyclose VLO388.95/MPC399.44/PSX260.78, SMA50319.4324/327.0682/217.0564; ret3+4.92595/+3.02546/+2.40320%, relative20toXLE+12.8874/+11.8695/+9.6588points. Premise2: fixed rule needs ret3negative; all3fail. ADR209.978725/10.187355/6.57678, recentannualvol24.141/21.965/22.343%; at212.7475positioncap nonefitsonewhole share. Premise3: oneADRreference stop and1.5ADRtarget VLO378.9713/403.9181,MPC389.2526/414.7210,PSX254.2032/270.6452 are researchlevels anchoredtohistoricalclose, NOT entryquotes or proposedorders. Conclusion no setup; do not use fractions merely to sidestep unvalidated afterhours exitcoverage. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; exactfigures20614. Factsendatcompletedbars; tomorrowgap/returnunknown.

Premise1: last120 daily OHLC rows resampled jointly within day, independent days;10000paths/symbol seed91006 after sizing draws; preserves recent bullish drift and gap/high/low/close dependence, omits serial clustering. InitialpriceSep9close, stop1ADR20,target1.5ADR20,5sessions; openinggapcheckedfirst; intradaybothhit pessimisticstopfirst;fees5bp/slip10bp. Premise2: stop/target/time% VLO46.72/48.87/4.41,MPC42.12/52.42/5.46,PSX44.20/48.67/7.13; netmeans .2075/.5774/.3822%; fifthpercentile-4.7916/-3.6494/-3.6301%. Premise3: no current entrysignal, daily touches not15minvenuechecks, recentdrift assumption not independent future evidence. Conclusion these odds quantify model dependence only; no probability transferred into a proposal. Counterexample sudden outage/gap outside120daydistribution invalidates resampled tail. Full model outputs20612, source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all . Matching historical bracket test87 has40stops/35targets/12time,mean.11644%,worst-4.90189%,maxloss2.17138R; stopdistance is not guaranteed loss bound. No automatic trade.

Premise1: investedcash holdout p52.8736%,win3.54074%,loss3.02751% gives quarterKelly .25*(p/loss-(1-p)/win)=1.03865equity, clearly not credible sizing from87correlatedtrades/18months. Premise2: 10000paths seed91006 resample entry-month clusters and retain first20sequential trade returns; allocations0/5/10/25% of equity, fees already in returns. At25% terminal5/50/95percentiles -8.7376/+1.9699/+15.7627%; loss39.35%,maxDD>10%6.15%. At5% terminal-1.8194/+.3835/+2.9977%,loss39.44%; at10%-3.5877/+.8422/+6.0162%,loss38.70%. Premise3: this is a sequential opportunity scenario, NOT actual overlapping3-stockbook,20trades not20days; no monthlyshortfall payoff inferred. Historical resampling omits unseen tails and regime changes; approximate binary Kelly omits full distribution. Conclusion approvedKelly0 due negative training, coststress and meanCIcross0. Venue212.7475positioncap,278.67fundcap,8.5099riskcap cannot be overridden by nominal pointKelly. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; all figures20610. Valid conditional simulation, not sound forecast odds.

Start: crypto fund $235.86 cash, total equity $850.99, no holdings or pending orders. Goal: repeatable net profit; binding constraint: independent executable expectancy. Premises: 1. All216 cells are recorded; all108 reused2026holdout cells have zero CLOSED trades. Supplied baseline holds one uncompleted September1 hypothetical lot per major, marked -.7073%/-.2961%/-1.8358%; strict reconstructed/matched baseline is flat. 2. Training BTC is cadence-sensitive, ETH and SOL supplied hourly returns lose; joint training bootstrap loses at nonzero allocations. 3. Historical price construction remains unresolved and current funding/supply snapshots are context, not a tested edge. Conclusion: no proposals; approved quarter-Kelly size0 because required OOS win probability and avg win/loss are unavailable. BTC training-only quarter-Kelly1.7815% of fund is ~$4.20, below $10 crypto minimum and not eligible regardless; ETH/SOL training hourly Kelly0. Validity: failing evidence gates warrants cash, not a forecast of falling prices. Soundness: conservative completeness can miss a rally; censored trades could later win. Neither changes what is known now. Evidence that would change eligibility: reconciled primary history plus independently frozen positive cost-stressed daily/hourly/15minute outcomes, then cap at min($212.75 position,$235.86 fund,$8.5099 nominal stop risk,quarter-Kelly) with gap/combined-risk checks. The $349.01 shortfall never enters probability or size. Sources https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://api.exchange.coinbase.com/products/BTC-USD/candles ; https://docs.alpaca.markets/us/docs/crypto-fees .

Premises: 1. The same frozen monthly entry condition and fixed ADR thresholds are observed either at daily closes or hourly closes, then filled at the next available open. 2. Supplied-history 2025 training daily/hourly returns are BTC+11.0372/+1.2896%, ETH-22.9129/-14.5782%, SOL-.5283/-2.0384%; counts7/6/4 unchanged. 3. BTC hourly cash mean+.2771% has bootstrapCI[-3.0491,+3.2910]%; ETH and SOL hourly means are negative. Conclusion: lower turnover alone does not produce robust executable entry evidence; BTC's positive daily headline is cadence-sensitive. Neither clock reproduces 15-minute stops or +1R partial/breakeven behavior. These are observation-model sensitivities, not stop-protected loss bounds. The strict reconstruction control changes eligibility, and matched masks eliminate portfolio-return construction differences; do not confuse that with a validated price history. No order, stop or target is recommended. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Premises: 1. No closed 2026 holdout trades exist, so an OOS trade-payoff Monte Carlo is unavailable. 2. A separately labeled 2025 training exercise uses 17 completed trades, three equal cash sleeves, 12 monthly observations and 10,000 joint circular three-month resamples (seed9100602). 3. Actual training portfolio return is -4.8278%; monthly mean confidence interval [-1.0938%,+.2693%]. At 25%/50%/100% of fund capital, simulated median annual return is -1.0327%/-2.1304%/-4.5639%; loss probability 85.21%/86.55%/88.38%, 95th percentile monthly-observed drawdown 3.9160%/7.7296%/15.1195%. Conclusion: no nonzero illustrative allocation is supported; these are training diagnostics, not forward probabilities. Full-fund median dollar loss is $10.76 on $235.86, not an estimate of this month's result. Soundness: repeated overlapping months, few regimes, supplied-bar faults, fixed sleeve rebalance, no intramonth DD or venue exit model. Approved0. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Premises1 NumPyseed20260910061 resamples6stressedgreenexitwholeeventreturnsinto50000ten-eventpaths,starting49.14. Theseareconditionalempiricalscenarios,not10tradesbySeptemberend orindependentfutureodds. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=DRI&timeframe=15Min&feed=sip&adjustment=all ; DARDEN06audit1832.2 Fullreinvestmentnoinjectedgap terminal5/50/95=45.68044/53.10953/59.98258,lossfrequency19%,95thmaxDD12.76551%. InjectASSUMEDindependent1%chanceof-20%returnperevent:41.78709/52.58812/59.76240,loss26.59%,DD22.16006%. NeitherhazardnorgapcalibratedtoDRI; sensitivitynotforecast.3 Quarterbookexposureterminal5/50/95=48.30106/50.14046/51.67571withoutshock,47.37669/50.00579/51.63430withshock;zeroexposure49.14throughoutbeforeoperatingcosts. All8size/shockcellsrecorded.4 Observedtarget4/6,partial5/6,stop2/6;Wilson95win/target[.29999,.90323],partial[.43650,.96995]. Bootstrapcannotgenerateunseentails.5 GeneralempiricallogKellyunconstrainedcapital8.81465x,quarter2.20366x,cashcap1x,butuncertainty/executiongateapproved0. Binaryaveragepayoffapproxquarterrisk.081274 convertscapital3.38028x; distinguishriskfromnotionalandfromgeneraldistributionKelly. No margin contemplated. 20%hypotheticalgapfullfundloss9.828 exceeds8.5099accountrisk;assumedgapriskcap42.5495notworstcasebound. Conclusionzeroallocation,positivepointKellynotpermission. Strongcountercasefutureeventsresembleobservedwins; cannotknowthatfromsixselectedreturns. Monthly349.01shortfallabsentfrommodel.

Premises1 use onlypreentryindicators;greenexitstresspayoffsfromDARDEN06audit1832, source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=DRI&timeframe=15Min&feed=sip&adjustment=all withSIPdaily.2 Positive20daymomentumtrainingN7mean-.56733%,testN4+1.53465%; trendSMA50positive trainN8-.18623%,testN3+1.43563%. Combinedtrend+momentumtrainN7-.56733%,testN3+1.43563%,latterCI[-.35844,3.07784]%.3 SPYabove200SMAtrainN7-.68158%,testN5+1.80542%; SPYbelowtrainN5+1.57002%,testN1-4.45028%. FavorabletestCIabove0inselectedN4/5groupsdoesnotestablishstablefilter:trainingdirectionreversesandmultiplecomparisonsunadjusted.4 ATR/cuttrainingmedian1.702676%:highvoltrainN6+.80897%,testN5+.98705%,testCI[-1.77498,2.75288]%;lowvoltestN1-.35844%notadistribution.5 CurrentSep9close209.42below20SMA219.0445and50SMA209.453,20day-4.33074%;notentrywindowSep17. Conclusionnoadoptedindicatorcombination,noholdoutrefit. ValiditysigninstabilityandsmallNblockcalibration; soundnessaffectedbyselectedevents/interdependence/currentvsfuturefundamentalregimes. Countercasetrueconditionaledgemayexist; requireprespecifiedprospectiveevents,notreweightingthese18.

Premises1 fixed audit1832 reuses original1376 dates:2022–24 training12,2025–Jun2026 chronological test6. Daily holdout was already known when execution variants were specified; this is a post-selection execution audit, NOT a fresh independent holdout. 18SECearningsdatesreverified;6testadvancenoticespreviouslyverified1299,only1/12trainadvancevintagesnowverified. CurrentAug27issuer confirmsSep24BMO https://investor.darden.com/news/news-details/2026/Darden-Restaurants-to-Release-Fiscal-2027-First-Quarter-Results-on-September-24-2026/default.aspx .2 BuyT5 09:30,forceT1 15:45open;1prior20ADRstop/1.5ADRtarget,at+1ADR sellhalf/movestopentry. FractionalRTHonly15minopenpriceproxy; gaps/overshoots fillobservedprice,notbarrier. 10bpfeesand10bpslipbaseline/25bpstressoneachfill. 2340RTHbars,90sessions,0missing;1427DRI+1427SPYdaily. https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=DRI&timeframe=15Min&feed=sip&adjustment=all .3 Vectorbtlongonlycashreconcilesmanualledgerincludingpartialfees andall18events. Trainbase+6.64856%,stress+2.87749%,DDstress8.30497%;testbase+6.35126%,stress+4.45405%,DDstress4.84462%,4/6wins,W2.34638%,L2.40436%,mean.76280%,bootstrap50kCI[-1.49818,2.52709]%. 5partials,4targetfinalexits,2stopfinalexits. Time-onlytestbase6.91174/stress5.00453%,barrierwithoutpartials7.70904/5.78762%;nonechosenafterseeingtest. Sixtrainingonlystop.75/1/1.25,target1.5/2ADRstressvariantsrange-7.82789to+4.74945%,allmeanCIsinclude0. Conclusionresearchwatch,operationalallocation0. Validitypositivepointpayoffdoesnotclearuncertaintygate. SoundnesslimitsN6/selectedissuer,reusedtest,11missingtrainnotices,tradebarsnotexecutablequotes,gridphaseassumed,noextendedhoursfractionalexit. Strongcountercasepositivecostedtrainingandtestmeritfuturepaperobservations,butnotpromotingnewfiltersorclaimingindependentreplication.

Premises: 1. September 9 closes BTC 78,283.41, ETH 2,467.70 and SOL 101.5535 exceed supplied SMA200 69,921.47/2,046.77/82.7251; 90-day returns are +23.1589%/+47.6390%/+51.9190%. 2. The monthly rule only enters at the first UTC day open; September 10 cannot be a catchup entry. 3. A missing March 12 hour invalidates strict reconstructed SMA200 through this cutoff; identical availability masks produce exactly identical portfolio returns in all 72 matched comparisons. Conclusion: positive supplied trend is descriptive, neither a fresh trigger nor independent validation. The mask explains the change to zero eligible holdout trades, not evidence that repaired prices create alpha. Full audit: 569/567/329 days differ by >.01bp on any OHLC, with maxima 620.64/790.68/1128.48bp; not every discrepancy is large. Strongest counterargument: one missing hour is conservative; filling it requires justified information, not choosing the feed that wins. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Premise1: relative20filter improves holdout DD but training remainsnegative; fixed SMA50 and3daypullback. Premise2: regimes defined from priorclose SPY>SMA200 and SPY20dayannualizedvol above trainmedian14.357995%; no OOS threshold fitting. Uptrend/highvol mean train-2.7941%N17 -> test+1.1096%N20; uptrend/lowvol train+.4847%N52 ->test+.1660%N61. Bear/highvol train+.1792%N32 ->test-2.2831%N4; bear/lowvol trainN0, testN2+7.78% meaninglesssmall sample. Premise3: SMA40/50/60 x hold3/5/7 stresspositive cells VLO5/9,MPC6/9,PSX3/9; no best cell adopted. Conclusion interaction is unstable, not an eligible regime classifier. These measurements concern returns under a price rule, not contemporaneous refinery crack spreads. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; complete figure arrays recorded20602/20616/20618.

Premise1: fixed OOS exits, 87 trades/18 entry-month clusters, same-sector co-movement. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all . Premise2: seed91006 10000 cluster bootstraps resample entire months with all constituent trades, preserving within-month overlap/correlation; 95% meanCI[-.87379,1.91693]%, 26.25% bootstrap means<=0. Premise3: pooled p=.528736,avgwin3.54074%,avgloss3.02751%,mean .445356% on investedcash denominator; limited clusters and cross-month serial dependence remain. Conclusion no defensible positive lower-bound edge, sizing0. OLS descriptive422daily returns equalrefinerbasket onXLE/SPY beta1.11797/.13699,intercept12.9571bp/day,R2.65442; paircorrelations.8565-.8852. Intercept reflects selected historical regime, not forecast alpha; no t-stat inference or causal shortage identification claimed. Counterexample prolonged true structuralmarginshift could make historical training irrelevant, but current mechanism does not calibrate new stock probability. Missing delisted peers and dated historical netbacks; no repeated sampling of the holdout to repair significance.

Frozen1830beforetesting:priorclose>SMA200,positive90dayreturn,firstUTCdaymonthlynextopenonly;prior20ADRstop1/target1.5,nextopenafterclosecross,28daytimeout. 2024warmup,2025train,2026Jan1-Sep9reusedholdout;quote-basedbars,fees25bpsideonreceivedasset,slip10bpbase20stress;vectorbt. Trainingdaily {"symbol":"SOL","construction":"supplied","look":90,"cadence":"daily","period":"train","slippage_bp":10,"return_pct":-0.5282629201018244,"drawdown_pct":12.360600236169683,"n":4,"open_trades":0,"win_rate_pct":50,"mean_pct":0.45636344825701075,"avg_win_pct":11.228498844271286,"avg_loss_pct":-10.315771947757264,"ci_low_pct":-10.315771947757264,"ci_high_pct":11.228498844271286,"quarter_kelly":0.010160829479219383,"trade_return_basis":"cash including entry fee"};traininghourly {"symbol":"SOL","construction":"supplied","look":90,"cadence":"hourly","period":"train","slippage_bp":10,"return_pct":-2.038394447248011,"drawdown_pct":11.822691073053582,"n":4,"open_trades":0,"win_rate_pct":50,"mean_pct":-0.07619643410124367,"avg_win_pct":9.063540836596294,"avg_loss_pct":-9.215933704798784,"ci_low_pct":-9.215933704798784,"ci_high_pct":9.063540836596296,"quarter_kelly":0,"trade_return_basis":"cash including entry fee"}. Holdouthourly {"symbol":"SOL","construction":"supplied","look":90,"cadence":"hourly","period":"holdout","slippage_bp":10,"return_pct":-1.835836462767984,"drawdown_pct":6.3790855312988555,"n":0,"open_trades":1,"win_rate_pct":null,"mean_pct":null,"avg_win_pct":null,"avg_loss_pct":null,"ci_low_pct":null,"ci_high_pct":null,"quarter_kelly":null,"trade_return_basis":"cash including entry fee"}. Baseline2026onlySep1hypotheticallotstillopen;Nclosed0 meanswinprobabilityandavgwin/lossUNAVAILABLE,not0%. Portfolio markincludesentrycostnotunpaidexitcost;notrealizedperformance. All216predeclaredcellsrecorded20543–20579 (36groups),nonechosenafterholdout. Reconstructed/matched24hourcompletenesscontrolsremoveholdouteligibilitythrough200daywindow;do notreplaceunknownwithprofitableflatreturn. Conclusion:rejecttradequalification,approvednotional0;noSep10monthlycatchup. Missingvendorreconciliation,unseenvalidation,15minand+1Rpartialexecution. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen1830beforetesting:priorclose>SMA200,positive90dayreturn,firstUTCdaymonthlynextopenonly;prior20ADRstop1/target1.5,nextopenafterclosecross,28daytimeout. 2024warmup,2025train,2026Jan1-Sep9reusedholdout;quote-basedbars,fees25bpsideonreceivedasset,slip10bpbase20stress;vectorbt. Trainingdaily {"symbol":"ETH","construction":"supplied","look":90,"cadence":"daily","period":"train","slippage_bp":10,"return_pct":-22.912931943804026,"drawdown_pct":29.600472527637358,"n":6,"open_trades":0,"win_rate_pct":33.33333333333333,"mean_pct":-3.835557626745903,"avg_win_pct":8.307509580031027,"avg_loss_pct":-9.907091230134368,"ci_low_pct":-10.827850683642813,"ci_high_pct":3.2764394017755647,"quarter_kelly":0,"trade_return_basis":"cash including entry fee"};traininghourly {"symbol":"ETH","construction":"supplied","look":90,"cadence":"hourly","period":"train","slippage_bp":10,"return_pct":-14.57820278230409,"drawdown_pct":21.20298251575542,"n":6,"open_trades":0,"win_rate_pct":33.33333333333333,"mean_pct":-2.3849502955813273,"avg_win_pct":6.633046344172845,"avg_loss_pct":-6.893948615458413,"ci_low_pct":-7.055236740973971,"ci_high_pct":2.706662288679817,"quarter_kelly":0,"trade_return_basis":"cash including entry fee"}. Holdouthourly {"symbol":"ETH","construction":"supplied","look":90,"cadence":"hourly","period":"holdout","slippage_bp":10,"return_pct":-0.2961156222521458,"drawdown_pct":4.516462231099483,"n":0,"open_trades":1,"win_rate_pct":null,"mean_pct":null,"avg_win_pct":null,"avg_loss_pct":null,"ci_low_pct":null,"ci_high_pct":null,"quarter_kelly":null,"trade_return_basis":"cash including entry fee"}. Baseline2026onlySep1hypotheticallotstillopen;Nclosed0 meanswinprobabilityandavgwin/lossUNAVAILABLE,not0%. Portfolio markincludesentrycostnotunpaidexitcost;notrealizedperformance. All216predeclaredcellsrecorded20543–20579 (36groups),nonechosenafterholdout. Reconstructed/matched24hourcompletenesscontrolsremoveholdouteligibilitythrough200daywindow;do notreplaceunknownwithprofitableflatreturn. Conclusion:rejecttradequalification,approvednotional0;noSep10monthlycatchup. Missingvendorreconciliation,unseenvalidation,15minand+1Rpartialexecution. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Frozen1830beforetesting:priorclose>SMA200,positive90dayreturn,firstUTCdaymonthlynextopenonly;prior20ADRstop1/target1.5,nextopenafterclosecross,28daytimeout. 2024warmup,2025train,2026Jan1-Sep9reusedholdout;quote-basedbars,fees25bpsideonreceivedasset,slip10bpbase20stress;vectorbt. Trainingdaily {"symbol":"BTC","construction":"supplied","look":90,"cadence":"daily","period":"train","slippage_bp":10,"return_pct":11.037206557667341,"drawdown_pct":10.687911952758999,"n":7,"open_trades":0,"win_rate_pct":57.14285714285714,"mean_pct":1.6479645501167965,"avg_win_pct":6.025415263369898,"avg_loss_pct":-4.188636400887339,"ci_low_pct":-2.3736410337387053,"ci_high_pct":5.473745685965708,"quarter_kelly":0.06837555911437386,"trade_return_basis":"cash including entry fee"};traininghourly {"symbol":"BTC","construction":"supplied","look":90,"cadence":"hourly","period":"train","slippage_bp":10,"return_pct":1.2896311169101864,"drawdown_pct":12.208071728720215,"n":7,"open_trades":0,"win_rate_pct":57.14285714285714,"mean_pct":0.2770930124034394,"avg_win_pct":3.8884019764055537,"avg_loss_pct":-4.537985606266045,"ci_low_pct":-3.049089503251003,"ci_high_pct":3.2909516321710583,"quarter_kelly":0.017815352816196284,"trade_return_basis":"cash including entry fee"}. Holdouthourly {"symbol":"BTC","construction":"supplied","look":90,"cadence":"hourly","period":"holdout","slippage_bp":10,"return_pct":-0.7072560827173948,"drawdown_pct":4.728457474137726,"n":0,"open_trades":1,"win_rate_pct":null,"mean_pct":null,"avg_win_pct":null,"avg_loss_pct":null,"ci_low_pct":null,"ci_high_pct":null,"quarter_kelly":null,"trade_return_basis":"cash including entry fee"}. Baseline2026onlySep1hypotheticallotstillopen;Nclosed0 meanswinprobabilityandavgwin/lossUNAVAILABLE,not0%. Portfolio markincludesentrycostnotunpaidexitcost;notrealizedperformance. All216predeclaredcellsrecorded20543–20579 (36groups),nonechosenafterholdout. Reconstructed/matched24hourcompletenesscontrolsremoveholdouteligibilitythrough200daywindow;do notreplaceunknownwithprofitableflatreturn. Conclusion:rejecttradequalification,approvednotional0;noSep10monthlycatchup. Missingvendorreconciliation,unseenvalidation,15minand+1Rpartialexecution. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

Premise1: fixed protocol note1827; three independently funded sleeves averaged, no cross-sleeve leverage or rebalancing, https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all . Premise2: refiners withrelative20filter train-9.6135%, holdout+11.6746% DD12.734%, stress-.5566%; producers same rule holdout-20.3984%,stress-25.9968%. Premise3: withoutrelativefilter refinertrain-6.9819%,holdout+9.4489%,stress-4.8838%. Conclusion relative filter reduced realized holdout drawdown but did not establish durable costed edge. Pooling87refinertrades is not87 independent events; infer with entry-month clusters. These are normalized full-capital sleeves, not fund-sized trades. Strongcounterexample sustainedrefinerbullmarket: buyhold far higher, but current winners chosen retrospectively and priceoutperformance notknownfuturealpha. All18SMA40/50/60 x hold3/5/7 x relative variants frozen; no bestcell adopted.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=EOG . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train -25.8821%, test -31.6540%, stress -35.8911%. Valid rule simulation; forward soundness unproven.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=CVX . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train -14.4896%, test -20.7078%, stress -25.9204%. Valid rule simulation; forward soundness unproven.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=XOM . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train 26.0290%, test -8.8336%, stress -16.1788%. Valid rule simulation; forward soundness unproven.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=PSX . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train -9.4516%, test -2.3151%, stress -13.0141%. Valid rule simulation; forward soundness unproven.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=MPC . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train -17.6185%, test 20.1209%, stress 6.9647%. Valid rule simulation; forward soundness unproven.

Premise1: preregistered note1827 before results; 2021 warmup,2022-24train,2025-Sep9 2026holdout; SIP adjustmentall https://runway.adityaperswal.workers.dev/data/stock-bars?timeframe=1Day&feed=sip&adjustment=all&symbols=VLO . Premise2: signal previous close>SMA50,3dayreturn<0,20dayreturn>XLE20day; nextopen entry, fifthsubsequentopen exit; nonoverlap; fixed no holdout tuning. Vectorbt fees5bp/slip10bp per side; stress30bp. Trade return denominator entrynotional; divide by1.0005 for investedcash; all independent reconciliation assertions pass. Premise3: no historical shortage labels, survivorship-selected incumbent universe; not causal bottleneck test. Conclusion no approval from isolated favorable holdout; full sensitivity recorded observations REF91006 rows. Cash-only separate fully funded research sleeve, fractional normalized sizing not venue whole-share execution; adjusted prices assume total-return basis rather than literal cash-dividend timing. No greenpartial/15min exits in this test. Train -1.7706%, test 17.2180%, stress 4.3797%. Valid rule simulation; forward soundness unproven.

Frozen notes1829/1831 corrected1833. SIPalladjusted2016warmup;train2017–19,holdout2020–25,2026separate;excludeera-crossingexits. Mode swing;published=penultimateclose..thirdnextmonthclose;calendar=lastdayopen..thirdnextmonthclose;control=sessions10–13; swing=calendarwithpriorADR20stopk/target1.5k,half+1R,BEfollowingday,dailystopfirst/gapopen. Fees1bp+slippage15bp/side; k=1.25. Eightindependentcashsleevesinitialequalweights,nointerETFrebalance;dailycloseDD,notintradayDD. Vectorbt/custommaxerror1.67e-15. Columns[symbol,return%,DD%,N,win%,avgW%,avgL%,mean%,quarterKelly,buyhold%]. [["SPY",1.8053200266264335,-1.8536736522437036,8,50,1.2133451441335708,0.7538798778409628,0.229732633146304,0.047334559802922394,11.646604882170042],["IWM",-3.763088785457258,-5.119634333009426,8,25,2.4852126969253874,1.4423611831344239,-0.46046771311947093,0,17.71753492916417],["EFA",-0.7458320406549923,-4.025830078150827,8,50,1.4923253796591485,1.64776109862341,-0.07771785948213067,0,11.423999871433942],["EEM",2.0365177595741413,-7.099427417387161,8,50,2.228103661547476,1.6776678523611366,0.27521790459316964,0.03088028503148993,23.017757690709306],["EWJ",1.3816224740166305,-4.989152328147783,8,50,2.4163618232601882,2.0078484023922676,0.20425671043396015,0.02113267024703841,19.50619193057992],["EWU",-3.007569655850373,-3.5290736857124916,8,37.5,1.1299451509787506,1.2732238658623896,-0.372035484546962,0,9.069950723629749],["EWC",0.6771191771856477,-3.8220557824172996,8,62.5,1.2624027356300926,1.843966661448652,0.0975142117255634,0.01931123265446899,13.144234558818946],["EWA",-5.619934191853504,-5.619934191853504,8,25,1.677965852969432,1.5058622069153391,-0.7099051919441463,0,13.26369568764103]]. Current-survivorETFs/adjustedprices;notrawdividend,whole-share,3entrycaporovernight15minvalidated. No promoteholdoutneighbors. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all .