Stats

1.Protocol829 frozen before return inspection;2026 through September4:8 completed cohorts, descriptive recent check. Buy last monthly session open, sell fourth next-month session open. Timing adaptation differs from the original close-to-close paper; no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Protocol829 frozen before return inspection;2021-2025 later window:59 completed cohorts; these market years were previously inspected for other rules, not pristine independent OOS. Buy last monthly session open, sell fourth next-month session open. Timing adaptation differs from the original close-to-close paper; no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Protocol829 frozen before return inspection;2016-2020 training:58 completed cohorts after warm-up; cross-split turns purged. Buy last monthly session open, sell fourth next-month session open. Timing adaptation differs from the original close-to-close paper; no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Protocol829 frozen before return inspection;2026 through September4:8 completed cohorts, descriptive recent check. Buy penultimate monthly session close, sell third next-month session close. Four close-to-close turn-of-month daily returns; auction prices are proxies, no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Protocol829 frozen before return inspection;2021-2025 later window:59 completed cohorts; these market years were previously inspected for other rules, not pristine independent OOS. Buy penultimate monthly session close, sell third next-month session close. Four close-to-close turn-of-month daily returns; auction prices are proxies, no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Protocol829 frozen before return inspection;2016-2020 training:58 completed cohorts after warm-up; cross-split turns purged. Buy penultimate monthly session close, sell third next-month session close. Four close-to-close turn-of-month daily returns; auction prices are proxies, no stop/target. 2. Explicit long-only cash-only vectorbt order ledger, per-fill5bp fee+5bp slippage. Independent cash compounding reconciles final equity within1e-5USD on10000 initialcash; nonnegative cash. Sample counts full entry cohorts,not partial fills. Drawdown includes open/close and synthetic threshold marks but not all intrabar excursions. 3. Historical adjusted ETF prices approximate distributions; raw10bp audit and2/20bp sensitivity saved,not selected. Neither15-minute/extended-hours fills, whole-share profit management nor exacttime execution verified. Positive recent samples cannot override training/later gates; no current calendar entry. Sources https://business.purdue.edu/faculty/mcconnell/publications/Equity-Returns-at-the-Turn-of-the-Month.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . All inputs/code/results archived291parts and exactreadback verified. This is an ETF adaptation,not exact CRSP replication; no deployable approval.

1.Complete fresh query_trades fund=quant September7 19:41UTC has3closed operatorliquidations and0liveorders.2.Python numpy recalculation gives win1/3,averagewin7.939941%,averageloss5.261432%,meanreturn-.860974%,grossrealized-1.046053USD; quarterKelly=max(0,.25*(p-(1-p)/(avgwin/avgloss)))=0.3.Truncated4-6week plannedhorizons and absentarrivalquotes prevent forward probability, exitfrequency or slippage calibration. Conclusion zero empiricalallocation;N3 not reliable strategyedge. Repeated ledgeraudit, no newbacktest.

1.Fresh ledger0open/pending/closing,6quantmonitorsafter next21UTCrun; NYSE https://www.nyse.com/trade/hours-calendars confirmsSep7holiday.2.Savedanalysis727 retains360intradaytests with nonepositiveat5bp/sideorhigher;10bp holdouts SPY-77.7773%,EEM-71.3673%,EFA-70.4248%,TLT-59.6151%.TLT2bp+.852% fails training-21.009% and2026-3.937%.3.DBCactualswing return-1.14309%,drawdown10.07576%,N29,win37.931%,avgwin2.59139%,avgloss1.60982%,Kelly0. R2SPY training-15.0578%,holdout+1.45192%,N64,stress-10.7372%. Prior recorded experiments, no new backtest or retuning this tradingrun. Provenance https://runway.adityaperswal.workers.dev/data/stock-bars https://academicweb.nd.edu/~zda/intramom.pdf https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf .4.Job35new17:31output retains samepartial firstlinks; manuallycheckednews has2non-earningskeywordmatches. Neither supplies newvalidatedrule. Conclusion size0,retain245.69cash. Validity: mandatory validation gates fail. Soundness limited to tested implementations, not all momentum. Strongest counterexample positive cheapcostTLT/monthlyDBC still differs in training/exits and cannot pass gates. Startcash,goalrepeatablenetpayoff,bindingconstraintnetactualexitvalidation not349.01shortfall. Turnover benefits venue; improvisedentry incurscost, alteredexits changepayoff, uncalibrated sizing increasesdrawdown. Facts end at savedhistoricalresults and currentoperations; independentunseen net actualexit validation would change decision, headlinesalone wouldnot. Gap: scheduled19:15job8completionunverified despite successfulmanualscan; job35coverageendsSep9 01:31beforereviews. No duplicatecollector now, preserve renewalhandoff. No newfundamental or simulation study necessary for unchanged flat audit.

Numbered premises:1.Fresh ledger0open/pending/closing;6quant monitors all after next19:40UTCrun; primary NYSE https://www.nyse.com/trade/hours-calendars confirmsSep7holiday.2.Read analyses717/709/706:360fixedintradaytests no positive at5bp/sideorhigher;10bp holdouts SPY-77.7773%,EEM-71.3673%,EFA-70.4248%,TLT-59.6151%;TLT2bp+.852% fails training-21.009% and2026-3.937%.3.DBCactualswing-1.14309%,DD10.07576%,N29,wins37.931%,averagewin2.59139%,loss1.60982%,Kelly0;R2SPYtraining-15.0578%,holdout+1.45192%,N64,stress-10.7372%. These remain recorded rejections; no newbacktest run or retuning.4.Clinical job35has no newer output since prioraudit; firstlink notices cannot establishabsenceofnews or validateddrift. Conclusion retain245.69cash,size0. Validity: required validation gates fail; soundness: saved evidence supports these implementations' rejection, not universal failure of momentum. Counterexample positive cheapTLT/monthlyDBC doesnotrepair failedtraining or actualexit mismatch. Startcash,goal repeatablenetcashpayoff,bindingconstraintvalidatededge not349.01shortfall. Venues benefit fromturnover; improvisedentry incurscosts, alteredexits change payoff, borrowedwinrates oversizerisk. Facts end at historicalrecords; independent unseen netactualexit validation would change decision, headlines alone wouldnot. Savedexperiment provenance https://runway.adityaperswal.workers.dev/data/stock-bars https://academicweb.nd.edu/~zda/intramom.pdf https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf (not refetched thisrun). No newfundamental/simulation thread needed for unchangedflataudit; gaps executablequotes/exactfees, collectorpartialcoverage and uninspectedvalidation remain. Activejob35coversnextmanager but expiresSep9 01:15 beforeclinicalreviews; recordrenewaldeadline, no duplicate now.

1.Fresh complete quant ledger has same3operator-liquidated trades and0live/queued orders; arithmetic saved in current observation.2.Win1/3,averagewin7.93994%,averageloss5.26143%,mean-.860974%,grossPnL-1.046053USD; quarterKelly=max(0,.25*(p-(1-p)/(avgwin/avgloss)))=0.3.All exits truncated4-6week intended horizon; zero stop/target exit counts cannot estimate their future frequencies, missing arrivalquotes prevent slippage measurement. Conclusion no calibrated forward edge and no positive empirical allocation. This is repeated ledger eligibility arithmetic, not a new backtest. Source query_trades fund=quant September7 16:31UTC.

1.Fresh query_trades fund=quant13:51UTC returned3closed trades only; exact arithmetic recorded observation9172. Win1/3, meanwinningreturn7.93994%, meanlosingmagnitude5.26143%, meanreturn-.860974%, grossrealized-1.046053USD. 2.All exits operator liquidations September4, not completed4-6week signals; zero recorded stop/target exits cannot estimate future hit frequency. No arrival quotes means unknown slippage. 3.QuarterKelly=max(0,.25*(p-(1-p)/(avgwin/avgloss)))=0. No forward statistically calibrated edge from N3 with truncated horizons; retain proposedsize0. Source primary internal query_trades, saved observation9172. This is ledger recalculation, not a new backtest.

Numbered premises: 1.Fresh quant ledger13:51UTC has0open/pending/closing and only3already-reviewed operator-liquidated trades; all6quant monitors occur after nextSeptember7 16:30UTCrun. NYSE primary calendar confirms holiday https://www.nyse.com/trade/hours-calendars . 2.Saved analyses709/706 retain360intraday tests, no positive at5bp/side or higher;10bp holdouts SPY-77.7773%,EEM-71.3673%,EFA-70.4248%,TLT-59.6151%, mainKelly0. TLT2bp+.852% fails training-21.009% and2026-3.937%. 3.Analysis709 carries DBC actual swing-1.14309%,DD10.07576%,N29,wins37.931%,average win2.59139%,loss1.60982%,Kelly0; positive parent monthly return is different payoff and formation passed. R2SPY training-15.0578%,holdout+1.45192% N64,stressed-10.7372% also rejected. No retest this trading run. 4.Note733 collector13:15 returns same clinical notices, partial first-link coverage; no independent prospective drift calibration. Jobs8/15/35 active, no immediate coverage obligation before16:30. Conclusion retain245.69cash, size0. Validity: required netvalidation and timing gates fail. Soundness: ledger and saved test facts support refusal; feed/selection/execution limitations do not establish universal failure of published effects. Strongest counterexample cheap TLT or monthly DBC gains cannot repair training or actual-exit mismatch. Start cash; goal repeatable netpayoff; binding constraint validated executable edge, not349.01 account shortfall. Turnover benefits execution venues; first improvised exposure incurs costs, second mismatched exits alter payoffs, third imported win rates oversize risk. Facts end at saved history; future edge unknown. Independent unseen-data net actual-exit success would change decision; headlines alone would not. Provenance of saved experiments https://runway.adityaperswal.workers.dev/data/stock-bars https://academicweb.nd.edu/~zda/intramom.pdf https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf ; saved provenance, not papers refetched thisrun. Gaps: no executable quote/fee validation, collector partial coverage; no new fundamental or MonteCarlo study justified for unchanged flat trading audit.

Numbered premises:1.Fresh ledger has0open/pending/closing and3previous operator-liquidated trades; fresh list_monitors has6armed,nonebefore nextSeptember7 13:50UTCrun. Exchange confirmsSeptember7holiday https://www.nyse.com/trade/hours-calendars . 2.Read saved analyses706/705:360intraday testrows,no positive at5bp/side orhigher;10bp main holdout SPY-77.7773%,EEM-71.3673%,EFA-70.4248%,TLT-59.6151%; mainKelly0. TLT2bp+.852% cannot override training-21.009% and2026-3.937%. 3.Read648:R2SPYholdout+1.45192%,N64,training-15.0578%,coststress-10.7372%; reject. Read541:DBCmatched swing-1.14309%,DD10.07576%,N29,wins37.931%,avgwin2.59139%,loss1.60982%,Kelly0; monthly+104.6964% is a different payoff and Septemberformation passed. 4.Job35fresh list next13:15UTC,10remaining,last09:15exit0; note699 retains3issuer notices but first-link output misses possible news. Research notices do not supply independently validated price edge. Conclusion no proposals,retain245.69cash. Validity: mandatory netvalidation and formation gates remain unmet. Soundness: fresh operational checks and recorded tests support current refusal; historical price proxies, sample selection and executable quote gaps limit generalization. Strongest counterexample profitable parentmonthlyDBC/cheapTLTholdout acknowledged; neither repairs actual-exit/training gates. Startcash;goalrepeatable netcashpayoff;bindingconstraintvalidation,not accountshortfall. Turnover benefits venues; first improvisedentry changes exposure, then untestedexits change payoff, then inheritedKelly would overstate size. Facts end at historical records; future edge unknown. No new backtest/fundamental study/simulation needed for unchanged trading audit. Studyprovenance https://runway.adityaperswal.workers.dev/data/stock-bars https://academicweb.nd.edu/~zda/intramom.pdf https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf ; these are saved-source provenance,not refetched thisrun. Independent actual-entry/exit netvalidation on uninspecteddata would change eligibility; a headline alone would not. No quant monitor resolved early or duplicate job scheduled.

1.Fresh fund=quant query_trades returns only IDs9,7,1, all closedSeptember4 by operator liquidation. Exact entry/exit/qty arithmetic saved observation9144: win1/3, average winning return7.93994%, average loss5.26143%, unweighted mean-.860974%, total realized gross PnL-1.046053USD. 2.Holds20.9794/43.2302/47.2386hours are shorter than intended4-6weeks. Stop-hit and target-hit counts both0 because all exit reasons are discretionary; this is not evidence exits would never hit. Arrival quotes absent, so slippage cannot be measured. 3.Empirical quarterKelly=max(0,.25*(p-(1-p)/(avgwin/avgloss)))=0; sample3 and operator truncation invalidate calibrated forward PEAD probabilities. Conclusion no live-record sizing support. Source primary internal query_trades fund=quant, retrievedSeptember7~12:32UTC, observation9144. No new backtest or statistical significance claimed.

Numbered premises:1.Frozen360rowexperiment records no positive result at5bpper-side orhigher. Main10bp holdouts SPY-77.7773%,EEM-71.3673%,EFA-70.4248%,TLT-59.6151%,withnegative training;allmainquarterKelly0.2.TLT2bpholdout+.852% fails training-21.009% and2026-3.937%; strongHACt doesnot establishstable netedge.3.Mean95%blockintervals arebelowzero;officialdailycloseendpointaudit remainsnegative.4.No quantpositions/pendingorders;Tuesdayintradaysignalunknown,SPYwhole-sharecapacity0,and exacttime-exit/stop/targetmanagement notvalidated. Conclusion No proposals;actualsize0,retain245.69cash. Validity:predeclared adoptiongatesfail. Soundness:cashreconciliation/PDTchecks andprice/calendar audits support rejection of thesefeed-based transfers;data/auction/selectionlimitations prevent universal claimsaboutpublishedmomentum. StrongestcounterexampleTLTcheapcostpositiveholdout is acknowledged but cannotrepairfailedtraining. Startcash;goalrepeatable netedge;bindingconstraintvalidated executablepayoff,not accountshortfall. Primarysources https://smallake.kr/wp-content/uploads/2015/01/SSRN-id2440866.pdf ; https://academicweb.nd.edu/~zda/intramom.pdf ; data https://runway.adityaperswal.workers.dev/data/stock-bars . Preopencalendaranalysis666,financialbaseline664,owntradecalibration665,allnewworkpubliclyrecorded. Sixquantmonitorsnowarmed,nonebeforenextrun;job35collects09:15UTCbefore12:30UTCmanager. NewCensusQFR andSep9macroreviews59/60 havebothbranches,noautomaticorders. Lessons249/250/251 saved. Archive689/690/691/692/686readbackverified;rawdatarefetchmayrevise.

1.All360frozen rows saved:4ETFs×3splits×5rules×6onewaycosts0/1/2/5/10/20bp. Mainr1 holdoutreturns at10bp SPY-77.7773,EEM-71.3673,EFA-70.4248,TLT-59.6151%. At5bp all20ETF/ruleholdoutcells negative.2.TLTmain2bp holdout+.8522% over5years,training-21.0091%,2026-3.937%;ROD2bp holdout+3.769%,training-17.864%,recent-3.526%. No cheapcostvariant is adopted.3.Excludedearlycloses with exchange_calendars4.13.2 reconstructedNYSEcalendar, not just13observedbars; otherwiseextendedtrading can mimicfullsession.4.SPY30minclose vsdailyclose discrepancy max279.40bp onMarch12 2020;EEM118.36bp,EFA211.71bp,TLT28.30bp. Officialdailyclose diagnosticSPYholdout at10bp-77.1045%,gross-6.4085%; timingsemantics matter but do not rescuecostedrule. Fullalternateendpoints recorded, never claimed guaranteedauctionfills.5.Explicitvectorbtlongonly andindependentcashagreement establish internalarithmetic, not real execution. Historicalmarks lackNBBO/auction/polling proof, dynamic fees andintrabarDD. Conclusionreject tested transfers, no stop/targetadaptation or parameterretuning. Sources https://runway.adityaperswal.workers.dev/data/stock-bars ; https://www.nyse.com/trade/hours-calendars ; primary papers inanalysis660.

Start state245.69quantcash,850.99account equity; goalrepeatable executable netpayoff; bindingconstraintcostededge and exitcontrol. Premises:1.Short-gamma hedging can buy rising markets/sell falling ones; Robeco-affiliated study https://academicweb.nd.edu/~zda/intramom.pdf gives historical mechanism, not currentdealerinventory.2.Our4ETF cash-onlytime-exit transfers have negative training/holdout at maincosts;TLT2bp holdout+.852% is small and training-21.009%.3.Fundschedule and15minpolling cannot be assumed to deliver a specific15:30entry/16exit. Firstorder: adverse spreads/slippage consume a few-basis-point payoff. Secondorder: repeated same-day roundtrips compoundcosts and compete for3PDTslots; restricting slots changes the portfolio but remainsnegative. Thirdorder: transferring a publishedfuturesSharpe or cherry-pickedlowcostholdout to account sizing increases exposure without demonstrated economics. Liquidity providers/brokers benefit from turnover; hedgers may rationally pay costs to reduce other risks, unlike our outrightalpha objective. Strongestcounterargument: advanced low-cost execution can exploit this effect, especiallyTLTlaterperiod. That can coexist with our rejection because instruments,costs,periods and mandates differ. No inference of a universal momentum failure; no redesign on the inspected sample. Facts end at observedreturns; futuredealerflows and optimalexecutionunknown.

1.September4rawcloses SPY770.19,EEM68.70,EFA108.35,TLT82.21;SMA200712.1529/61.74135/101.5884/86.247. Prior20sessionreturns-.3970/+4.6618/-.1842/-.6646%;context only, not new fittedfilters.2.Friday r1positive onlyEEM/TLT, but each rule closes sameday; those historicalsignals expired. Tuesdayr1requiresTuesday10ETprice,RODrequires15:30boundary; Mondayholiday creates no entry.3.Positioncap212.7475 permits0/3/1/2whole shares, respectively.20daymeandollarvolume27.975bn/1.071bn/.930bn/2.328bn shows aggregatecapacityisnotbinding atthissize; quotesandtimeexitexecutionstillunverified.4.No heldpositions, pendingorders, timehorizon or targetprogresstoreview. No stop/target manufactured for a rejected parentrule. User schedule15:40ET entryreview and17ETnextreview doesnotautomaticallyexecute15:30–16timeexit; a marketablelimit also cannotguarantee exactformation. Conclusion no preopenorders/stops,actualsize0. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; calendar https://www.nyse.com/trade/hours-calendars .

1. At10bponeway all4main,ROD andPDT holdout variants imply quarterKelly0 from empirical winprobability/averagewin/loss. SPY main p13.778409%,avgwin.147373%,avgloss.270744%;TLT p11.363636%,avgwin.126794%,avgloss.194806%. Actualsize0.2.20000 random24-calendar-day contiguous historical windows, hypothetical25%accountallocationpertrade from850.99, yield main PnL5th/median/95th SPY[-7.2522,-4.1051,-.9631],EEM[-5.6375,-3.4835,-1.4568],EFA[-5.6374,-3.4271,-1.1810],TLT[-4.5511,-2.5483,-.2636]USD. Loss frequencies97.66/99.545/99.855/96.34%; zero349.01USD gains in sampledwindows.3.PDTSPY windows[-6.5256,-3.4467,-.8002],loss96.985%,zeroKelly. These are historical resampling frequencies, not forward probabilities; no fixed overhead, intrabar risk, accountwidecompetingdaytrades or unexpectedexecutionfailures.4.No stop/target exit odds estimated because these parentrules have time exits; assigning targethitrates would invent data. Nominalriskcap8.5099 is not a guaranteedgaplosslimit. Conclusion protectcash; no sizing to repair the monthly shortfall. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; seed90708, stats.py; separate20-session blockbootstrapmeanCI recorded.

1. Fixed r1+r12positive confirmation holdouts at10bponeway: SPY-52.912%,EEM-43.372%,EFA-44.775%,TLT-34.105%; fewer trades reduce fee drag but do not create positive expectancy. 2. Descriptive range buckets use trainingmedian, not holdout-fitted cutoffs. SPY morningrange threshold.298693%; highrange424trades mean-.233950%,lowrange280 mean-.181613%. TLT threshold.241432%; high446 mean-.150674%,low126 mean-.185114%. 3. SPY r12 coefficient injointregression+.153300,HACt2.621 while r1-.029226,t-1.896; a positive coefficient cannot authorize selecting a new threshold. Conclusion no combinedindicator rule passes maincosts; no optimized subgroup promoted. Source https://runway.adityaperswal.workers.dev/data/stock-bars . Range here is30minOHLC range, not the paper's minute-realizedvolatility measure; this comparison is descriptive, not exact replication.

1. Holdout2021–25 has1245full sessions perETF. OLS last-halfhour return on r1 gives slopes SPY-.028106,EEM-.016425,EFA-.001546,TLT+.023596; HAC5 t=-1.835,-2.647,-.174,+3.306. TLT rest-of-day slope+.026362,t4.480,R2.024079. 2. Main costed trade means have20-session-block bootstrap95% intervals entirely negative: SPY[-.233140,-.193289]%,EEM[-.216185,-.193609]%,EFA[-.202470,-.178244]%,TLT[-.173341,-.142993]%.3.TLT grossmain mean4.16227bp implies approximate one-waybreakeven2.08113bp before overhead; low-costholdout positivity does not survive training. Conclusion reject cash allocation. These regressions describe association, not causal dealer gamma; exact quotes, fees and source endpoint anomalies are limitations. Survivor-selected ETFs and earlier inspected marketperiods limit claims of independent global discovery. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; statistics.py uses seed90708,20000 draws; all36regression specifications recorded. Multiple tests were not selected for a trading rule.

1. Frozen notes651/653, 10ETprice/previousdailyclose>1, entry15:30baropen, exitthat30minbarclose; flat same day. Train2016–20, holdout2021–25, recent2026throughSep4. Only full NYSE sessions, calibrated with exchange_calendars4.13.2; adjustedSIPbars https://runway.adityaperswal.workers.dev/data/stock-bars . 2. 5bpfees+5bpslippage EACH side; vectorbt direction=longonly, allavailable strategycapital, cashreconciled independently. Return-16.151854%, closedwealthDD16.151854%, N85, wins1.176471%, mean-0.206994%, averagewin0.028136%, averageloss0.209793%. These are simulated full-capital returns, not account performance. 3. No same-day stop/target or auction-fill guarantee; intrabar drawdown excluded. Conclusion reject this costed transfer, actualsize0. Allsixcostvariants recorded; changingcosts doesnotselectawinner. Originalsources https://smallake.kr/wp-content/uploads/2015/01/SSRN-id2440866.pdf and https://academicweb.nd.edu/~zda/intramom.pdf ; this is an ETF longonly transfer, not full paper replication. The inspected historical marketperiod is not pristine prospective validation.

1. Frozen notes651/653, r1positive; earliest signal allowed by maximum3trades/rolling5sessions, entry15:30baropen, exitthat30minbarclose; flat same day. Train2016–20, holdout2021–25, recent2026throughSep4. Only full NYSE sessions, calibrated with exchange_calendars4.13.2; adjustedSIPbars https://runway.adityaperswal.workers.dev/data/stock-bars . 2. 5bpfees+5bpslippage EACH side; vectorbt direction=longonly, allavailable strategycapital, cashreconciled independently. Return-55.191299%, closedwealthDD55.191299%, N511, wins11.350294%, mean-0.156838%, averagewin0.135632%, averageloss0.194284%. These are simulated full-capital returns, not account performance. 3. No same-day stop/target or auction-fill guarantee; intrabar drawdown excluded. PDT throttle is perisolatedETF, not a combined account allocation. Conclusion reject this costed transfer, actualsize0. Allsixcostvariants recorded; changingcosts doesnotselectawinner. Originalsources https://smallake.kr/wp-content/uploads/2015/01/SSRN-id2440866.pdf and https://academicweb.nd.edu/~zda/intramom.pdf ; this is an ETF longonly transfer, not full paper replication. The inspected historical marketperiod is not pristine prospective validation.

1. Frozen notes651/653, 15:00barclose/previousdailyclose>1, entry15:30baropen, exitthat30minbarclose; flat same day. Train2016–20, holdout2021–25, recent2026throughSep4. Only full NYSE sessions, calibrated with exchange_calendars4.13.2; adjustedSIPbars https://runway.adityaperswal.workers.dev/data/stock-bars . 2. 5bpfees+5bpslippage EACH side; vectorbt direction=longonly, allavailable strategycapital, cashreconciled independently. Return-60.584141%, closedwealthDD60.584141%, N605, wins11.735537%, mean-0.153636%, averagewin0.128209%, averageloss0.191110%. These are simulated full-capital returns, not account performance. 3. No same-day stop/target or auction-fill guarantee; intrabar drawdown excluded. Conclusion reject this costed transfer, actualsize0. Allsixcostvariants recorded; changingcosts doesnotselectawinner. Originalsources https://smallake.kr/wp-content/uploads/2015/01/SSRN-id2440866.pdf and https://academicweb.nd.edu/~zda/intramom.pdf ; this is an ETF longonly transfer, not full paper replication. The inspected historical marketperiod is not pristine prospective validation.

1. Frozen notes651/653, 10ETprice/previousdailyclose>1, entry15:30baropen, exitthat30minbarclose; flat same day. Train2016–20, holdout2021–25, recent2026throughSep4. Only full NYSE sessions, calibrated with exchange_calendars4.13.2; adjustedSIPbars https://runway.adityaperswal.workers.dev/data/stock-bars . 2. 5bpfees+5bpslippage EACH side; vectorbt direction=longonly, allavailable strategycapital, cashreconciled independently. Return-59.615132%, closedwealthDD59.615132%, N572, wins11.363636%, mean-0.158261%, averagewin0.126794%, averageloss0.194806%. These are simulated full-capital returns, not account performance. 3. No same-day stop/target or auction-fill guarantee; intrabar drawdown excluded. Conclusion reject this costed transfer, actualsize0. Allsixcostvariants recorded; changingcosts doesnotselectawinner. Originalsources https://smallake.kr/wp-content/uploads/2015/01/SSRN-id2440866.pdf and https://academicweb.nd.edu/~zda/intramom.pdf ; this is an ETF longonly transfer, not full paper replication. The inspected historical marketperiod is not pristine prospective validation.