Stats

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.

Preregisterednote652; numpyOLS dailyROIVreturn=a+bSPYreturn,all-adjustedIEXbars2022-Jan3 throughSep4 2026 https://runway.adityaperswal.workers.dev/data/stock-bars . Chronological2022–24train752pairedreturns beta1.09843,correlation.32355,R2.10468,intercept.04914%/day;2025–Sep4 2026holdout420 beta.91825,corr.39021,R2.15226,intercept.21894%/day. ROIVmean/std daily .09205/3.73385% train,.28739/2.46271%holdout. Premises: lowR2 andlargeidiosyncraticdispersion; clinicaleventisticker-specific. Conclusion macroprintsdescribecontext,notPHocuslikelihood orforecastalpha. Intercepts arein-sampledescriptivefitswithin eachpartition,notout-of-sampleforecasts orcausal abnormalreturnestimates. Countercase broaderisk-oncouldmatter; beta near1supports someexposure butonly10–15% varianceexplained. No statisticalsignificanceorregimestabilityclaimed. Survivorselectedsingleissuer,newscontamination,partialIEXandadjustmentrevisionslimitgeneralization. Requireprospectiveeventdata rather thanrepeatedqueriescreatingfalseindependence; allocation0.

1. Exact fill returns are -5.392364%,+7.939941%,-5.130499%; p=33.333%,averagewin7.939941%,averageloss5.261432%,mean-0.860974%,realized-$1.046053. Wilson95% win interval[6.149,79.234]% makes precision low. 2. All three exits were operator liquidations, not the planned4–6week stop/target or a common rule; stop-hit and target-hit probabilities are unmeasured. All entry orders were market, with no contemporaneous reference quotes, so slippage-by-session comparisons cannot be identified. Conclusion raw quarterKelly0 and actualsize0; do not transfer these returns to intraday momentum or claim a PEAD win probability. The strongest counterargument is DELL's7.94% gain, but one selected winner cannot identify the intended payoff. Source firstparty query_trades(fund=quant) September7; exact arithmetic saved research/ledger.json.

Premises:1 frozen calendar/aggregation sensitivity backtests596/597/598 (06UTC) supply BTC p=29/115,w=.9911125444%,l=1.180520989%; ETH p=12/118,w=4.177281755%,l=1.259869640%; SOL p=12/107,w=5.200698903%,l=1.302715925%. 2 Python recomputation E=p*w-(1-p)*l gives -.63289/-.70694/-.57336% per completed trade; quarter Kelly .25*(p-(1-p)/(w/l)) gives negative estimates, therefore zero eligible allocation. Costed holdout returns-52.1851/-58.5744/-49.2381%,drawdowns52.2339/66.9159/60.9755%,winrates25.2174/10.1695/11.2150%. These are archived experiments reviewed now, not rerun or new independent evidence. 3 Own ledger trade10 is one BTC operator liquidation after47.93min with-.08099%price return; intended3-4week strategy censored, no wins or full strategy exits, fee/slippage attribution unavailable. Conclusion no reliable positive realized or tested edge to size. Strong counterargument daily alternatives have positive1-2trade sensitivity results; no meaningful loss population and hourlyfailure prevent adopting them. Prior simulation605/606 showed drift-removed doubled-cost means approximately-1.2%; alreadyinspected simulations are scenarios, notindependent edge. Soundness limited by olderbarprovenance and sparse daily samples, so no universalclaim all crypto rulesfail. No retuning ornewCLItest in this refresh because rule state stays rejected; prior CLI and vectorbt experiments retained rather than doublecounted. Sources internalresearch596-598,605-606; https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees ; query_trades fundcrypto.

1. Recalculate all supply trades from live query_trades in Python/numpy: FRO3.805162958%,STNG0.752256770%,meanwin2.278709864%,N2,zero observed losses. Both exited by operator liquidation onSep4 afterSep2 entries, not strategy stop/target outcomes. Source venue ledger https://runway.adityaperswal.workers.dev . No open or pending rows. 2. Illustrative IID Wilson95% lower win-probability bound34.2380% for2/2 already exposes uncertainty; shared tanker exposure and discretionary exits violate the IID/setup premise. Do not adopt this bound as calibrated probability. Average loss, execution slippage versus synchronous quotes, and stop/target-hit frequencies are unidentifiable. 3. Freeze existing empirical evidence instead of treating repeated readings as new data: FRO analysis456 N41 meanCI[-0.874362,1.958541]%; aluminum615/616 actual-ADR exits negative. https://runway.adityaperswal.workers.dev/data/stock-bars . These are distinct cohorts, not pooled experiments. No fresh backtest claimed. Conclusion valid under required numeric-edge gate: approvedsize0; soundness limited by finite selected histories and unmodeled15minfills. Counterexample profitable scarcityrepricing ahead of data may occur; it cannot supply missing probability/loss estimates.

1. Frozen main swing rule has holdout samples SPY64,IWM72,EFA35,EEM23,TLT27 in2021–25; earlier market history was inspected under other rules. 2. NumPy20,000 circular three-calendar-month blocks retain inactive months. Mean-trade95% intervals (%): SPY[-0.22752,0.33007],IWM[-0.53203,0.28883],EFA[-0.73200,-0.09902],EEM[-0.66541,0.20921],TLT[-0.68025,-0.04489]. SPY mean0.02903%; positive resample fraction0.58935 is not posterior probability of edge. 3. SPY daily baseline mean0.71629%, interval[0.16584,1.17769], but training loses17.56%; a positive later baseline does not validate changed exits. Matched39 SPY entries: baseline31.3672% versus swing1.2266%; aggregate swing includes25 additional entries. Conclusion reject deployment, not every possible reversal rule. Validity follows frozen train/holdout/cost gates; soundness limited by clustered small sample, assumed stationary blocks, selected historical period and quote/poll omissions. No post-result bucket or threshold becomes a new strategy. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; statistics.py/diagnostics.py and all255 recorded metric rows.

23scheduledpre-earnings windows fromissuerdateobservations8020–8043 andIEX https://runway.adityaperswal.workers.dev/data/stock-bars . Holdout10events mean+.109305% but20,000calendar-yearcluster bootstrap95%meanCI[-1.408228,+.925035]%; only3yearclusters,notpreciseconfidence. Train13mean+.708895%,CI[-.237380,+1.300316]%,4clusters. HoldoutnetGIS minusgrosssamewindowSPYmean-.221269%,versusXLP-.099777%;conservativecost asymmetry,nothedgebacktest. Holdoutnetreturn percentiles0/5/50/95/100: -7.583938/-5.146350/.169236/3.731859/4.767482%. OLS trained on12finiteevents: return= -.10556970 -.02600990*prior20dmomentum +6.67488562*priorATR/open. Holdout10 R2=-2.732571 versusholdoutmeanbenchmark; momentumcorrelationtrain-.051905/holdout+.419999. Coefficientsunstable,notcausal/usableforecast. TrendSMAundefinedin3trainobservations,regressionATR/momentumundefinedin1;donotconflatesamples. Priorpositiveeventedgeunsupported; noweighedapprovalprobability. Wouldrequireindependentlargerprospectiveeventswithnetmeanlowerbound>0andrealisticexecutiontochangesizing. Statisticalresamplingdoesnotcreatenewindependentevents.

1. The frozen 2025–September 4, 2026 holdout has only 22 AA and 26 CENX trades. Net mean returns are 0.8684% and 0.4291%; 10,000 circular four-trade block resamples yield 95% mean intervals [-1.7738,4.2172]% and [-2.7594,3.7542]%. Both include zero. Two operator-liquidation wins in our own book provide no loss distribution or independent calibration. 2. On matched five-session windows, AA and CENX underperform uncosted XME by 1.3465 and 0.3836 percentage points per trade. This compares net strategy to gross benchmark, not clean alpha. Daily AA/CENX correlation is 0.6652. Equal initial strategy sleeves returned 10.3778% with 17.6426% close-mark drawdown and 0.3842 strategy-return correlation, but this is virtual full deployment, not authorized sizing. 3. NumPy OLS with Newey-West Bartlett lag5 covariance on 420 holdout daily returns: AA XME beta1.0600, daily intercept -0.08397%, asymptotic HAC p=.5051, R²=.4705; CENX beta1.1907, intercept+.07418%, p=.6039, R²=.3901. Descriptive exposures, no predictive stock alpha established. statsmodels unavailable; direct matrix implementation used. 4. Train-median volatility splits were determined without holdout data: AA threshold55.95% annualized, CENX69.08%. High-vol holdout means -.4450% N7 and -1.2889% N6; low-vol +1.4813% N15 and +.9446% N20. Small selected subgroups cannot justify a new low-vol rule. Two current surviving candidates, corporate changes, adjusted histories, concurrent metal exposure and no historical first-news labels limit inference. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; observations8047–8078 and individual trade observations8430 onward. Conclusion: retain zero approved size; neither diversification nor a positive average repairs unresolved uncertainty.

Premises:1 Price from underlying forecast independently of feedIV.2 Value must cover ask, exitspread andfees.3 Uncertain favorable recalibration does not validate a trade. Sep7 collection of Sep4 indicative chains: EEM140,TLT110,GLD502 =752contracts, no nextpage tokens, Sep18expiry only. Riskaffordable ask*100+.20<=8.5099 gives29EEM,44TLT,114GLD=187; positivebid7/13/57=77. All187 priced from adjustedIEXSep4close S68.72/82.22/406.78 and frozenannualvol .178580666/.094090586/.193561450. AmericanCRR300steps,r4%assumption,q0,T14/365 fromSep4close toSep18close. Dailyvariance forecast5sessions extended flat to optionhorizon is UNVALIDATED; clock sensitivity uses9futuretrading sessions/252 (Sep8–18), multiplied additionally by.8lowvol. Other scenarios1.2vol andcontinuousyield6%; distributions notverified, continuousyield is NOT discrete-dividend solution. Standard100multiplier/deliverables assumed, actualOCC not verified. Diagnostic margin=100*(fair-ask-halfdisplayedspread)-.20assumedroundtripfee; this is an untruncated VALUE HURDLE, not realized P/L. Where modeled liquidation fair-halfspread<0, actual hypothetical bid floors0 and loss is premium+fees; this does not change zero passing baselinecount. Entryask anddisplayedsizes/quote timestamps,IV/Greeks and every model row are in observations. Modelmargin best positivebidTLT P79.5 -$2.083775, fair.016162 versusask.03/bid.02; EEM C73.5 -$6.306129,fair.028939 versusask.07/bid.03. No187 baselinepass or robustpass. Posthoc sensitivity sigma*sqrt(upper95holdoutvariance-ratio),600steps: EEMC73.5 becomes+$2.185501 at23.5647%vol; TLTbest remains-$1.398324;GLDbest-$2.662994 untruncatedmargin. EEM positive ONLY in favorable post-test forecastshift, not independently heldout and not executable. Numericalcheck600/1200: EEMC73.5 .028979/.028978,EuropeanBS.02900024; TLTP79.5 .016205/.016211,European.01607982; GLDP334 1.80469e-7/1.88509e-7,European1.99832e-7. Stable rejection acrossmesh does not remove forecast/dividend/feed uncertainty. Quotes are modified per provider https://docs.alpaca.markets/us/docs/historical-option-data ; chains https://runway.adityaperswal.workers.dev/data/option-chain/EEM?feed=indicative and/TLT and/GLD . No actualquotehistorybacktest or executionclaim.

Premises:1.query_tradesreturns3closedquantpositions,alloperatorliquidations.2.Recomputedwin1/3,avgwin7.93994%,avgloss5.26143%,mean-0.860974%,PnL-1.046053USD;quarterKellymax(0,[p-(1-p)/(win/loss)]/4)=0.3.PositionswerePEADthesesandoperatorcensored,notR2exits. Conclusion do notborrowtheirpayoffsforreversal sizing;actualallocation0pendingindependenttest. Noquoteatdecisionorfullfillpath,soobservedentrypricescannotidentifystrategyexecutionalphaorslippage. Strongcounterexample:DELLwon7.94%;onewinnerdoesnotestablishpositivepooledmeanornewruletransfer. Source query_trades fund=quant IDs1,7,9 retrievedSep7 2026;observationownrecordaudit.

Premises:1 Future realized variance must be forecast with data available at forecast date.2 Forecast loss improvement is not option-return expectancy.3 Cross-market and serial dependence must enter uncertainty. Fixed SPY,IWM,GLD,XLE,EEM,TLT,IBIT universe; adjusted IEX daily2022-01-01 through2026-09-04,1173bars each exceptIBIT665 from2024-01-11. No duplicate timestamps or missing dates relativeSPY after ownfirstdate; this does not verify complete consolidated coverage. r=log(Ct/Ct-1), target y=mean(next5 r²). Candidate dailyvariance RV20,RV60,EWMA alpha.06/.03,logHAR using log squared-return means1/5/22 sessions plus intercept; OLS logtarget, logfloor1e-10, in-sample exp-residual smearing. Minimum60session warmup. Fit/select minimum training mean(log f+y/f) at dates<2024-12-20, so all target outcomes precede2025 holdout. Training686rows per establishedETF,178IBIT; train selection biased in-sample by design, heldout untouched. HAR selected IBIT/IWM/SPY/EEM/GLD,EWMA.97 TLT/XLE. Holdout2025-01-01 onward sampled every5sessions gives83 disjoint forward windows perETF (581rows,83 independent calendar windows at most, NOT581 independent assets). Manual target-index checks passed all7. Annualized current vol sqrt(252f): IBIT55.00%,IWM17.48%,SPY10.47%,TLT9.409%,XLE21.93%,EEM17.86%,GLD19.36%. Lossdifference selected-RV20 negative better: SPY-.115592,IWM-.067709,TLT-.051919,XLE-.037129,EEM-.071902,IBIT+.018637,GLD+.134709. 50000 circular moving-block bootstrap draws,4 five-session observations perblock, common indices acrossassets preserve dependence; Bonferroni95family interval onlyTLT excludes0[-.114048,-.001401]. SPY[-.251724,.002757],IWM[-.162909,.009529] lose individual significance. Equalasset pooledmean-.027272,95CI[-.084595,.028682]. Realized/forecast variance ratios:GLD2.047,EEM1.397,IBIT.621; approximate95ratio intervalsGLD[1.433,2.802],EEM[1.051,1.741],IBIT[.472,.800]. Do not retrofit favorable ratio into validation. Assumptions: fixed blocklength, stationary resampling, selected survivingETF universe, IEXsample; no event/jump identification or volatility risk premium. Counterexample: GLD highvol regime invalidates stable diffusion forecast. Valid conclusion limited to TLT forecast component worth prospective checking; does not establish cheap options. Source https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=SPY,IWM,GLD,XLE,EEM,TLT,IBIT&timeframe=1Day&feed=iex&adjustment=all . Scripts research_vol/analyze.py,uncertainty.py; all coefficients,traininglosses,holdoutrows,regimes andbootstrap figures recorded as observations.

Premises1 fully paginated2025-Jan1 to2026-Sep7 05UTC hourly counts14740/14740/14735, duplicates0, missing2/2/7.2 Compare only24-hour UTC aggregates to raw daily:217/612 BTC,211/612ETH,205/611SOL days differ at1e-6 relative tolerance; most are smallclose differences, max0.09466/0.09229/0.11537%. Larger open discrepancies6.2064/7.9068/10.2992%; maxlow3.9861/6.8874/11.2848%; volume96.6146/99.3558/95.3367%. Last mismatchesBTC Aug30,ETH/SOL Apr10.3 LatestSep6 passing audit does not validate older execution opens or ranges. Source https://runway.adityaperswal.workers.dev/data/crypto-bars . Counterargument bar definitions or revised trades may explain discrepancies; missing correction metadata means cannot identify which series is authoritative. Conclusion raw-history backtests are diagnostic, no allocation even for positive results. Implement explicit calendar gap fencing and matched hourly-derived daily sensitivity, without treating this as repaired ground truth. Binding constraint data provenance plus independent netedge; no accusation feed fraud.

Premises1 query_trades fundcrypto Sep7 returned one closed BTC tradeid10: entry81614.8 Sep3 19:42:24.677UTC, exit81548.7 20:30:20.345UTC, operator liquidation; notional137.2644226836.2 Return from prices -0.080989%; holding47.9278minutes; stopped/targeted exits0, wins0, losses1, no winning-payoff estimate. Fees/quote-at-order/actual slippage unavailable.3 Operator interruption is a competing exit reason, not a completed3-4week signal outcome. Conclusion no empirical trading edge or reliable quarterKelly from this sample; require independent costed history and future observations. Counterexample early manual exit may save future loss but cannot establish intended strategy payoff. Recorded price return is not fee-adjusted realized cash P&L. Source venue query_trades fundcrypto trade10.

1. Using issuerQ2roundedunitprices,costsandpounds, decomposeGPchange atpriorunitspread: volume(428425-340305)*8.47=+746376.4;price428425*(13.70-28.32)=-6263573.5;cost428425*(19.85-13.34)=+2789046.75;net-2728150.35,within75.65ofexact-2728226fromrounding. Sourcehttps://www.sec.gov/Archives/edgar/data/101538/000110465926094035/uamy-20260630x10q.htm. Identity notcausalregression: mix/NRV/processingtiming canaffectunitcosts. FourQ1/Q2comparatorsarenotindependentstatisticalevidence, no quarterlyregressionclaimed. Inventory13.964876mfinished /21.380103mSb=65.3172%; endfinishedvalue /Q2pureSbCOGS*91=222.37days descriptivecoverage, not turnoverforecast. 2. https://runway.adityaperswal.workers.dev/data/stock-bars SIPdaily2022-Jan3–2026-Sep4:1172returns. Dailyquantiles1/5/50/95/99=-14.3081/-8.75544/-0.266698/11.42564/19.10013%;SPYcorrelation0.2198245,beta1.24149. Poolednonstationarysampleincludingpennystockperiod; relationshipnotcausalantimonyexposure. 3. Fixed35tradeholdoutIIDmeanCI[-3.1859,4.1258]% widens4-tradeblock[-3.9867,4.7908]%; resamplingpreservesonlyshortlocaldependence, notfuturepolicyregime. Rejectedrulecannotbequalifiedbypositivearithmeticmeanwhilecompoundedreturnnegative. 4. Live query_trades fund=supply ids6/3 onlytwooperatorliquidations,FRO3.80516% andSTNG0.75226%,mean2.27871%,zeroobservedloss. Reuseanalysis469; noindependentnewoutcomes. MissinglossandcensoredexitsmakefundrecordKellyundefined. Conclusion: sourcesestablishcommercialsqueeze, notstockpayoffprobability;approvedsize0.

Premises:1 European no-dividend parity is not an equality for American dividend-paying stock options.2 Standardpairedsameexpiry/strike contracts need cashdividends,exercise andbidask.3 A cheaplegmust clear independentvalue,entryask,exitspread,feesandrisk. Collected252indicativecontracts Sep7,quotesSep4;first2expiriesSep18/Oct16 priced86legs,43pairs,29bothpositivebid. LastIEXtradeS46.15; unusableIEXclosingquote39.34/53.58 isnotNBBO, so alltestsdiagnostic. D2.61,exSep18open13.729167calendar daysafterSep4close; r4%assumed,T14/365 or42/365. Pairinterval[Cbid-Pask,Cask-Pbid].17/29exclude S-Kexp(-rT);0violateconservativeAmericanband[S-PV(D)-K,S-Kexp(-rT)]. Thisshowscompatibility,notmarket-efficiencyproof. Model1 escrowCRR X=S-PV(D),Xnodes=X*u^(2j-i),u=exp(sigma sqrt(dt)),p=(exp(rdt)-1/u)/(u-1/u); stocknodesaddremainingPV(D);backwardmax(discountedcontinuation,intrinsic). Sigma40.8179926%from20adjustedlogreturns,n800,check1600maxdiff.00078686. Baseline86/86 fair-ask-assumed$.002/share<=0. Sep18K46 feedcallIV44.96% put99.22%;no-divmidresidual-1.775521 becomes+.830555afterdividend; modeledexercise differential+.943039. OctK46IV26.41%/64.85%,residual-2.351740 becomes+.254337;exercise+.279875. Do notcomparefeedIVasmodel-independentcheapness. Model2 independentstockGBM implicitFD,logS.001–250,2400space1800time,exactcashjumpinterpolation andAmericanprojection;fine3600/3000checknearK46. Sep18call1.565900 vsask1.72,put2.992254 vs3.63;Octcall1.878755 vs2.009,put3.820327 vs4.14. Unscaledescrow valuesSepcall1.483639/put2.926155 reveal modelrisk largerthanmesherror. Full86FDrevealsONEpositivepreexitspreadflag: Sep18P42fair.78105845 vsbid.58/ask.73;net+$4.905845/contractwithassumed$.20fees. Escrowfair.70732511 rejects. Halfcurrentquotedspread.075 givesfairbidproxy.70605845,net-$2.594155;ask73dollars exceeds8.5099fullpremiumrisk. Reject,notrobustcheapness. Actualfeesunknown; evenzerofeesleaveexitspreadproxynegative. Broadbounds/cheaplegmodels arenot executablearbitrage andcannotbeimplementedasforbiddenspreads. Counterexample: truejumpvolormoreaccurateexercise canchangefairvalues; model-sensitiveP42alreadydemonstratesit. Requiredrevision: executablequotes,verifiedcarry/deliverables,independentactualoptionreturnholdout. Sources https://runway.adityaperswal.workers.dev/data/option-chain/FRO?feed=indicative ; https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=iex&adjustment=all ; https://infomemo.theocc.com/infomemos?number=54262 ; https://docs.alpaca.markets/us/docs/historical-option-data . Models/files research_fro/analyze.py,verify.py,fd_all.py;eachquote/modelrowrecorded.

Premises: (1) Frozen main rule has29later active cohorts inside68calendar months, January2021–August2026; it is a selected, previously inspected sample. (2) Resample20,000 circular six-calendar-month blocks with seed90704, retaining inactive months; divide payoff sums by active counts. Mean trade-0.016256%,95%interval[-0.765399,+0.793882]%; positive resample fraction0.46225 is not a Bayesian probability of true edge. (3) Matched monthly minus swing comparison gives swing-minus-monthly mean-1.228586percentage points, interval[-2.884591,+0.429574], so the population difference is not established at95%. Descriptive correlation0.484933; OLS swing on monthly returns beta0.221734 with HAC3-lagSE0.055223, intercept-0.285071%; this is attenuation association, not causal identification. Removing the three best matched monthly trades cuts total from37.3643% to7.29442%. Main swing excludes2022 =>-12.5114%; only2eligible cohorts begin afterNov2025 methodology change. Conclusion reject adoption for lack of demonstrated positive payoff, not proof every monthly-trend exit must fail. Prior unknown implementation transfer updates toward fragile/nonpositive observed payoff; a preregistered independent sample with positive net lower confidence bound and executable15min fills would change the decision. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; observations7817/7820/7821, scriptsstats.py/audit.py.

Premises:1 mechanicaldividendpricechanges differ from stochastic totalreturn shocks.2 Removingcashdrops may improvevolforecast but doesnotguaranteeit.3 Forecastcomparisonmust keep2026unseen and useknownpastdata. Python collected1172matchedIEXraw/allOHLC Jan3 2022–Sep4 2026; IEXmissesothervenues. logreturns,annualstdsqrt252:20draw=adjusted40.8180%;60draw46.4959% versusadjusted49.1060%;252draw42.7511% versusadjusted42.9774%. Adjusted60dvolhigherisvalidcounterexample to 'dividendadjustmentalwayslowersvol.' Fifteenraw/adjustedratiochange daysabove.002log inferred,notissuer-certifiedexdates. Fixed20pastsessionmeansquare forecasts next10sessionadjustedmeansquare; nonoverlappingfuturewindows,step10 fromindex21. Trainend<=2025N98; forecaststart2026N16; discardonecrossyearwindow. QLIKE=log(vforecast)+vreal/vforecast. Deltaadjustedminusrawtrain+.00528588;holdout-.01078931. Circularblockbootstrap3windows,10000resamples seed7092026:trainCI[-.00292170,.01828023],holdout[-.03095147,.00772074]. HoldoutvarianceRMSEadjusted.0006111583 vsraw.0006165529. Conclusion: no statisticallyclear forecastinggain; useadjustmentforcorrectcashmechanics, not as provennewcheap-optionsignal. Validcomparison conditionalonIEXadjustments; forecasts share20dlookbackso blockbootstrap, smallNstillweak. Prior: correctionhelpsmechanicalmeasurement; evidencewould supportforecastedge only frozenlargerholdoutCIbelowzero, plusactualcostedoptionreturns. Nooptionhistorybacktest or monetaryreturn/drawdown/winrate measured; ownoptionN0. Do nottranslate variancecalibration intoKelly. Files research_fro/analyze.py,forecast_holdout.csv,stats.json. Sources https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=FRO&timeframe=1Day&feed=iex&adjustment=raw ; sameadjustment=all; datesandordinarypolicy issuer/OCC observations7422/7450.

AlpacaIEX452barsNov2024–Sep42026,excludezero-volumeandpriorzero-volumeobservations; train173dailyreturns through2025,test170in2026. OLS r_t=intercept+lag20daymomentum+lagTR/close. Coefficients0.00031832,-0.00449544,-0.09196510;trainmomentumcorrelation-.0228925;holdoutR2=-.0226811, worse thanholdoutmeanbenchmark. Holdoutdailymean+.337241%,bootstrap20000resamples95%CI[-.192086,+.877430]%. 1/5/50/95/99percentiles[-8.08934,-4.99560,.12338,6.37649,9.70074]%. Thisisdescriptive,notcausalorcalibratedeventforecast. No independentmarketfactor/ASX-FXadjustment; selectedissuer,regimechange,missingprints andiidbootstrapundercountdependence. Positive2026meanfailsuncertainty anddoesnottransporttoPDUFAwindow. CurrentpublicFDAdata cannotbebackdatedto2025signalformation. Source/dataandfullfiguresobservations7678,7722; reproducible research/analyze_final.py,extra.py.

Premise1 observations7454-56:892 contiguous hourly bars each Aug1-Sep7 03UTC, event windows use last pre14UTC or pre20UTC close, include announcement-hour pre-event minutes.14h cumulative BTC-.301%,ETH+.169%,SOL-2.114%;8h BTC-.306%,ETH+.311%,SOL-.488%. Premise2 pre-event877 hourly logreturns regress BTC on intercept+equal-weight ETH/SOL; beta.62826,R2.69795,residualsd.2180%/hour;BTCETHcorr.83644,BTCSOL.72017.14h BTC residual+.27497% at77.89th empirical percentile versus8h-.27427% at23.10th. Overlapping pre-event windows864/870 not independent trials; these percentiles are not win probabilities/p-values. Premise3 N1 incident, controls potentially affected and macro/weekend confounding unresolved. Conclusion timestamp/control sensitivity refutes confident causal BTC selloff or rebound attribution; it does not prove no contagion. Prior bridge-specific rather than mainchainfailure remains plausible, no numeric Bayesian posterior fabricated. Evidence that would move decision: verified recovery/service restart plus separately registered multi-event net payoff, or demonstrated wider shared-code exposure. Raw sources/operators and explorer timestamps anchor event, supplied03:27headline over7hafteroperator statement cannot be eventtime. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://x.com/Liquid_BTC/status/2096696272447218108 ; https://blockstream.info/api/tx/8db751a650ae2f12006b7e8c69a75e4df360e8afd6b9e05ae0b9fa6458a7b140 .

Premises: 1 Primary venue snapshot Sep6 21:02:06UTC yields8/10 quotes age0-300s, USDC/USDG stale. Python calculates unchanged-book loss100*(1-bid/ask*.9975^2): BTC.523215%,ETH.523151%,SOL.598964%, excluding extra slippage. Published tier1 taker25bp per side reverified; actual account fees not reconciled. 2 BTC/ETH/SOL displayed asks80.21/83.89/100.29USD, bids79.43/82.64/101.49USD. LTC bid3656.89USD now versus1.05USD prior19:41: a replenished quote counters any claim permanent illiquidity, but does not establish fill distribution. 3 One venue/snapshot cannot calibrate15min stops; no independently validated edge. Conclusion zero allocation. Venue/liquidity providers benefit from unvalidated turnover; fees reduce capital, reducing future risk capacity and runway. Sources https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; https://docs.alpaca.markets/us/docs/crypto-fees .

Premises: 1 Primary venue snapshots observation7388 show8/10 valid quotes within300seconds, one snapshot per symbol; not an execution distribution. Python computes displayed notional and unchanged-book loss100*(1-bid/ask*.9975^2). BTC/ETH/SOL costs .532631/.523781/.593599percent using reverified published tier1taker .25% per side; extra slippage excluded, actual account fee reconciliation unverified. 2 SOL ask301.47USD versusbid99.79; LTC ask3703.11 versusbid1.05; freshness and affordable entry do not guarantee exit liquidity. 3 Backtests398-400 fail eligibility; exact15minute polling/depth model unresolved. Conclusion zero allocation. Counterexample deeper levels or replenishment could permit fills; one level does not prove total illiquidity. Source https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; https://docs.alpaca.markets/us/docs/crypto-fees .

Premises1 Python computed quote-age,spread,displayednotional and unchanged-book roundtrip loss1-(bid/ask)*(.9975^2); Sep6 16:32receipt primary snapshots. BTC/ETH/SOL ages7.85/18.00/7.65s, spreads2.813/3.422/6.408bp, asks79.924/82.618/101.338USD,bids79.475/83.187/100.913USD. 2 docs confirmtier1taker25bp/maker15bp,fees on receivedasset; actualfeeaccount reconciliation unresolved. Modelinstantaneousloss.527365/.533422/.563111% excludes extraslippage; fresh quote does not guarantee depth or fill. 3 sample3one-timequotes cannot estimate execution distribution or naturalexitwinrate; ownonlyBTCtradeoperatorliquidated. Wider ten-symbol compacttest8/10fresh; USDC/USDGstale. Conclusion size0 follows failedstrategyvalidation, not blanketclaim crypto cannotprofit. Counterexample makerfeeslower but adverse selection/wait/15minpolling unvalidated. Sources https://docs.alpaca.markets/us/docs/crypto-fees ; https://runway.adityaperswal.workers.dev/data/crypto-snapshots .

Premises 1: timestamped venue snapshots13:52:21UTC, BTC/ETH/SOLmid79812.0395/2494.2975/106.635; quoteages118.94/35.94/.94seconds, spread1.714/3.508/12.191bp. Seven of10<=300s, USDC/USDG/PAXGstale; threshold is diagnostic not executable permission. Ask USD79.95/83.17/304.10; bid79.86/83.89/101.38, one displayed level only. https://runway.adityaperswal.workers.dev/data/crypto-snapshots . 2 primary fees reverified0.25% tier1taker eachside on creditedasset https://docs.alpaca.markets/us/docs/crypto-fees . Instantaneous unchanged-book roundtrip loss=1-(bid/ask)*(.9975^2): BTC.516427%,ETH.534274%,SOL.620604%; excludes additional slippage and uncertain fee account reconciliation. 3 prior net range reward/risk431 only1.10-1.25 and polling simulation433 is not a calibrated edge. Conclusion no limit or market proposal; fresh asks exceeding a proposed notional would not guarantee exit depth. Counterexample maker orders reduce posted fee to0.15% but fill selection/wait/adverse movement untested. Zero allocation follows failed strategy validation, not claim fees prohibit all profitable crypto. Prior sole trade10 operator liquidation supplies no natural-exit probability or slippage estimate.

1. query_trades fund=supply returns only two closed trades, both operator liquidation onSep4. Python recomputation from fill prices gives FRO3.80516296%,STNG0.75225677%,mean2.27870986%. 2. No loss observed; average loss and payoff ratio undefined. 3. Common tanker exposure and operator-selected exits defeat independent-rule interpretation. Conclusion: observed100% wins cannot estimate prospective strategy edge; quarterKelly undefined, size0. IID Wilson95 lower34.238% is sensitivity illustration only, not applicable calibrated bound. Counterexample: genuine positive edge may exist despite tiny sample; that supports prospective collection, not numerical confidence. No reference quotes to measure slippage and no natural stop/target outcomes. No new backtest or Monte Carlo performed; frozen456 and prior439 remain separate historical/economic evidence. Source query_trades fund=supply ids6/3; bars historical provenance https://runway.adityaperswal.workers.dev/data/stock-bars .

Read query_trades fund=quant September6 12:32UTC; exact trade IDs9,7,1. Computed exit/entry-1 and qty*(exit-entry). N3, one win, average win7.93994%, average loss5.26143%, equal-trade expectancy -0.860974%, aggregate P&L -1.046053USD before separately billed overhead. Quarter-Kelly max(0,[p-(1-p)/(avgwin/avgloss)]/4)=0. All exits operator liquidation, none strategy stops or targets; horizons4-6weeks truncated to roughly1-2days. Premises: this is the full available quant closed ledger, but intervention and tiny sample invalidate extrapolation to intended PEAD payoff. A profitable DELL is a counterexample to 'all signals lose', not proof of aggregate edge. Slippage cannot be calculated without arrival quotes/order benchmark. Do not pool this empirical sample with monthly ETF cohorts. Conclusion cash; no simulation is used to manufacture exit odds from3intervened trades. Source internal query_trades fund=quant; recorded exact trade-level calculations observation7276.