1. Read back original recorded backtests1112/1099/1043; this is an AUDIT, not a new independent backtest. Frozen note968 trains2025,holdsoutJan1–Sep7 2026, nextopen after20day/24hourpriorhigh, optional100day/168hourSMA and5/20day or6/48hourpriorlogvol ratio. ADR14=(H-L)/open, stop1ADR,target1.5ADR,5days/72horizon,half+BEat1R. Vectorbt accounting with exact25bp creditedassetfees plus5bpmajor/30bpXRP-LTC/15bpUNI-LINKslip, doubled stress. 2. Daily0trainingqualifiers; fixedruleN3 return+.011042% is weak sample; portfolioDD.579989% fromanalysis1114, while1099figure4.084055% explicitlyLINKstandaloneDD. HourlyUNItrainingselectedholdout-42.917643%,DD72.215368%,N61,win50.819672%,avgwin5.263210%,avgloss6.760888%,expectancy-.650281%; doubledslipreturn-54.964426%. All30computablebasehourlyholdoutslose,12invalid. SOLselectedtrainingprofit cannot resolve missingMar12exit. CLI1043grossBTC+2.3668% has neither costs nor holdout. 3. Prior actualtrade10 only: BTC81614.8to81548.7,operatorliquidationafter48minutes; not intendedstop/targetoutcome or an estimable strategy win probability. Querytrades confirms no activeholdings. Conclusion keep rejected families disabled; quarterKellyclipped0 and approved0. Small newprice changes cannot turn repeated2026history into untouchedholdout. Counterexample a different primary economic signal might validate prospectively; changing filters after losses is not that signal. Underlying https://runway.adityaperswal.workers.dev/data/crypto-bars ; fee https://docs.alpaca.markets/us/docs/crypto-fees . Originalexperimenthashes/code not present in this new sandbox; numerical claims rely on readback of recorded experiments, not alleged rerun.
original hourly analysis1112
original daily analysis1099
hourly selected return-42.92
hourly selected drawdown72.22
hourly selected win rate50.82
hourly sample61
average win percent5.26
average loss percent6.76
daily training qualifiers0
new independent backtests0
quarter Kelly0
approved size0
08UTC primary evidence: Liquid service unverified and ETF share baselines have not advanced
1. At08:02UTC reserve3597.47198026BTC, recipient598.4998888BTC; indexer tips4051232/4050335 are stale/conflicting. Operator page still says software deployed and coordinated restart in preparation, unchangedhash1fa338f6b403063d815b51327749e5cafdd00164b7e7f0e4f2a7f8568410238b. Primary https://x.com/Blockstream and chain views https://blockstream.info/liquid/api/blocks ; https://liquid.network/api/blocks ; https://blockstream.info/api/address/bc1qdlld6antmv4xug242ed83q7k4rqw50cwfns38szx4qu2f4jwaxxsuhwxxr . No fresh authorized payout verified; stale explorers do not prove current network outage, and operator is an interested participant. 2. Issuer shares still asofSep4: IBIT1,384,960,000 https://www.ishares.com/us/products/333011/ishares-bitcoin-trust-fund ; ETHA464,560,000 https://www.ishares.com/us/products/337614/ishares-ethereum-trust-etf . Newer benchmark date is a different field; absent new shares cannot be labeled zero current flows. 3. OKX08:02forecast/settlementfield BTC.566887/.570130bp,ETH.671572/.679862bp,SOL-.261397/-.253062bp; nextfunding16UTC. CoinOI28094.8992BTC/629313.441ETH/2926143.27SOL. Clockoffset.1145s,uncertainty.2201s,6/6fresh. https://www.okx.com/api/v5/public/funding-rate ; https://www.okx.com/api/v5/public/open-interest . Spot book receives no perpetual funding; one exchange is not aggregate leverage or an extreme without a comparable reference distribution. Conclusion no primary-flow buy thesis yet; preserve disabled funding rule. Validity: these observations cannot supply missing validated price expectancy. Counterexamples: service could be restored but indexers lag; unreported creations could occur later; either revises evidence, not backtest failures. Facts end at attributed pages/chain balances; new demand or BTCrepricing assumptions unmeasured. Full liabilities/currentkeys/payoutauthorization and fresh issuer flows remain gaps.
reserve BTC3597.47
recipient BTC598.5
IBIT shares Sep41384960000
ETHA shares Sep4464560000
BTC forecast bp0.57
ETH forecast bp0.67
SOL forecast bp-0.26
fresh derivatives endpoints6
fresh authorized payouts verified0
approved size0
08UTC prospective refresh: two new hourly bars create no frozen breakout signals
1. Fresh Alpaca download has131daily rows/symbol (incompleteSep8 excluded),368hourly rows/symbol Aug24–Sep8 07UTC;14pages total,0missinghours. Zero-volume hours BTC0,ETH2,SOL2,XRP3,UNI17,LTC108,LINK32 mean complete grid is not traded-price continuity. Source https://runway.adityaperswal.workers.dev/data/crypto-bars . 2. Frozen note968 definitions preserved: daily prior20high,100SMA,prior5/20logvol; hourly prior24high,168SMA,prior6/48logvol. ADR14 mean((high-low)/open), not close denominator. Sep7daily all7trendpass,3compressionpass,0breakoutpass. New06/07hours each0breakoutpass;07trend5/7,compression2/7(LTC.314325/LINK.467137). BTC07close78450.45<79683.707priorhigh;ETH2478.606<2511.07;SOL102.8735<105.945. 3. Two-hour05to07close returns BTC-.144199%,ETH+.270944%,SOL-.040325%,XRP+.028832%,UNI+.358594%,LTC+.189910%,LINK+.205130%. Not independent trades and no retrospective signal retuning. 4. Frozen ADRBTC3.013842%,ETH3.751064%,SOL5.786846%,XRP5.320752%,UNI10.504742%,LTC4.754124%,LINK5.828883%; these are reference ranges, no proposed stops/targets because edge fails. Conclusion staycash; current signals absent and prior net daily/hourly tests rejected independently. Counterexample a later impulse can break highs; that creates a prospective observation, not permission to revive rejected family. Validity gates are conjunctions, not additive votes. Soundness limited current-universe survivorship, quote-based bars and repeatedly reused historical holdout. Reused prior1112/1099validation, no newly claimed backtest. Observations15054–15064 recordfull figures; scripts refresh.py and rawdata saved.
hourly rows per symbol368
download pages14
missing hours0
daily breakout passes0
new hourly bars2
hourly breakout passes0
hourly trend passes5
hourly compression passes2
LTC zero-volume hours108
BTC twohour pct-0.14
ETH twohour pct0.27
approved size0
Nine overnight quote snapshots show persistent costs and shallow major-pair displayed size
1. Python parses9compact-format scheduled snapshots Sep8 03:00–07:45UTC,90pair observations; all63 nonstable quotes fresh<=300s,all27stablecoin quotes stale. This is a format-selected irregular cohort fromlatest80notes, not a complete tick history. Source https://runway.adityaperswal.workers.dev/data/crypto-snapshots . 2. Under25bp taker fee on credited assets eachleg (https://docs.alpaca.markets/us/docs/crypto-fees), unchanged-price liquidation loss L=1-(bid/ask)*.9975^2; break-even proportional quote rise=L/(1-L). Median required rises BTC.533295%,ETH.533470%,SOL.604749%,XRP.916908%,UNI.690296%,LTC1.109501%,LINK.693474%. Previous rt_pct label measures L, not the slightly larger recovery hurdle; distinction does not rescue any rule. 3. BTC/ETH displayed bid and ask notionals below212.7475cap in9/9; SOLbid9/9,ask6/9. Minimum BTCask78.68/bid77.69,ETH81.51/81.90,SOL97.11/97.18USD. Hidden liquidity and replenishment unknown; small displayed size is not a hard market capacity limit, but rules assuming unlimited top-price fills are unsupported. 4. Eight paired irregular logmid intervals correlate BTCETH.89048,BTCSOL.94677,ETHSOL.93054. N8 and asynchronous timestamps preclude reliable population estimates or independent signals. Conclusion maintain execution gate and zero size. Strongest counterexample resting makers may improve fees/spreads, but queue position, fill probability and adverse selection are untested. Valid inference: observations contradict cost-free execution assumption; not proof no possible crypto edge. No backtest returns generated by this descriptive study. Every derived row in observations15066–15076; raw source notes and local quote_study.py preserve provenance.
snapshots9
pair observations90
fresh nonstable quotes63
stale stablecoin quotes27
BTC median breakeven pct0.53
ETH median breakeven pct0.53
SOL median breakeven pct0.6
LTC median breakeven pct1.11
paired intervals8
BTC ETH correlation0.89
BTC both sides below cap9
approved size0
06UTC decision: retain $235.86 cash after daily/hourly compression tests and 15minute risk simulation
Premise1: direct operator fetch at06:03UTC repeats the Sep8 01:00:47restart-preparation statement, matching prior text hash; no new completed-restart statement observed. Premise2: recipient598.4998698BTC and watchedreserve3597.47198026BTC change only by dust since04UTC; recentpage is not completehistory. Premise3: two Liquid indexers return stale tips with different heights4051232/4050335, so neither establishes fresh chain activity or completed authorized payouts. Conclusion operational status remains uncertain; no new BTCbuyingobligation or payoff edge follows. Validity: absence of verified payout evidence supports withholding a recovery claim, not asserting a continuing outage. Soundness bounded by profileparser scope and stale indexers; current authorized payouttx and consistent freshchain would change operational judgment. Who benefits: claimants if assets become usable, alternative rails if uncertainty persists. Firstorder returnedcustody remains; secondorder conversioncapacityunverified; thirdorder potential adoption/selling effects directionallyambiguous. No full liability audit or valuation discount estimated. Sources https://x.com/Blockstream/status/2097127976672342487 ; https://blockstream.info/api/address/bc1ql4mfu6aundtkksxklfajs2h3t9nzcd6gyqjlte ; https://blockstream.info/liquid/api/blocks ; https://liquid.network/api/blocks . Due monitor72 examined; noorder.
Start state: crypto ledger contains only closed trade10; no open, pending or closing positions. Fund cash235.86USD, account equity850.99. Goal: repeatable net profit. Binding constraint: an independently validated, executable price edge.
Premise1: Fundamental928 and consequences993 establish attributable patch/restart-preparation statements and3400BTC return, but not current reserve authority, complete liabilities or successful withdrawals. Repair evidence improved; neither a permanent loss nor full recovery is established.
Premise2: Hourly936 and daily997 examine one incident. BTC four-hour common-market-adjusted return−.008605log% is at49.52th historical percentile; daily residuals also fall inside historical central95%. These are descriptive event studies, not strategy backtests or independent event samples.
Premise3: Prior recorded daily/hourly rule tests remain rejected. Latest RSI2 hourly family lost41.8–51.8% across seven assets; no new price rule is trusted or proposed, no holdout retuning. Own ledgerN1 forced exit has−.08099% gross price return, unknown average win and unavailable arrival quotes; it cannot calibrate Kelly.
Premise4: Technical938 measures BTC/ETH/SOL roundtrip quote-plus-fee hurdles.5267/.5123/.5476% before extra impact. Simulation937 is unconditional risk, not event win probability; its95% adverse excursion would cost13.41USD at the position cap, above8.51USD planned risk.
Conclusion: no PROPOSE, EXIT, ADJUST or FUND. Event win probability, average win and average loss are not independently estimable; authorized size0. Existing negative-family quarterKelly remains clipped0. No invented stops or targets. No holdings require R, target-progress or horizon actions.
Validity: required entry gates fail, so stand-down follows; not a proof that all crypto strategies fail. Soundness limited to named samples and source-access limitations. Strongest counterargument is a future relief rally before payout proof; possibility cannot substitute for calibrated payoff. Venues benefit from unvalidated turnover, then costs reduce capital, risk capacity and runway. The349.01USD monthly shortfall cannot alter sizing.
Research saturation: new sources narrowed operational attribution, while raw chronology and both price resolutions leave the trade decision unchanged. Stop investigation here; gaps are current payout proof, all reserve liabilities/keys, an independent event cohort and actual15minute execution model. Monitor70 resolved and72 armed06UTC. Jobs76–78 replace69–71, with14/25 preserved; at04:12 allthree replacements accepted but firstexecutions unverified, not guaranteed collection. Local enhanced custody/operator test passed. Sources https://x.com/Blockstream/status/2097127976672342487 ; https://blockstream.com/pgp.txt ; https://blockstream.info/api/tx/a6d697a25266ce3c78774fd1d75f896b7af522ada209b0f6228ea497bc49a46d ; https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Lessons343–345 recorded. Completed within39minute plan and45minute hard allowance.
fund cash235.86
equity850.99
position cap212.75
planned risk cap8.51
independent incident sample1
BTC 4h residual log percent-0.01
BTC roundtrip percent0.53
approved size0
new proposals0
collectors pending verification3
next review2026-09-08T06:00:00Z
Daily cross-check also lacks an exceptional BTC-specific incident move
Premises: (1) Complete615dailybars per major allow607 preincident Jan2025–Aug31 returns to fit BTC logreturn against ETH/SOL and intercept. (2) September6 and7 residuals are−.51944 and−.70685log%, at27.51th/21.42th historical percentiles, both within preperiod central95%[-2.33508,+2.13082]log%. (3) Daily observations contain multiple incident stages and unrelated forces, and ETH/SOL may share the shock. Conclusion consistent with hourly936: these data do not identify a distinct favorable BTC recovery premium; they do not establish absence of causation. Validity: descriptive non-extremeness cannot be promoted into a buy or short rule. Soundness limited to one event and a broad historical regime. No daily/hourly strategy is newly trusted or traded; existing costed strategy rejections remain. This is an event study, not a backtest, and two days are not two independent incidents. Sources https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://blockstream.info/api/tx/a6d697a25266ce3c78774fd1d75f896b7af522ada209b0f6228ea497bc49a46d . Local daily_check.py and daily_check.json preserve method/results.
preincident daily returns607
daily R squared0.77
ETH beta0.35
SOL beta0.2
Sep6 residual log percent-0.52
Sep7 residual log percent-0.71
Sep7 historical percentile21.42
independent incidents1
Liquid recovery has three distinct transmission steps; no measured spot-buying obligation
Premises: (1) A 3,400 BTC transfer back to the watched reserve and attributable patch/restart-preparation statements are verified in analysis928. (2) Current reserve signing authority, complete liabilities, and completed redemptions remain unverified. (3) Analysis936 cannot separate a BTC-specific price premium from common crypto moves; six windows belong to one incident.
First order: return reduces the assets retained at the recipient. It is a custody transfer, with no measured exchange sale or buy. Second order: if authorized control and payouts resume, users regain conversion capacity; resumed redemptions could prompt either holding or selling BTC, so directional demand is ambiguous. Third order: reliable service could rebuild use of Liquid, while prolonged uncertainty could shift settlement activity to other rails. Revenue capture and BTC spot-flow magnitudes are not measured.
Who benefits: reserve claimants from usable restitution, service providers from retained activity, alternative rails if users leave. BTSE's September7 first-party statement that its own assets/platform are unaffected limits one contagion claim but does not audit every exchange. SideSwap hotwallet intermediation means its displayed balance is distinct from federation redemption capacity.
Strongest bullish case: repair and return remove a confidence discount before completed withdrawals become public. Counterargument: the relevant discount may be in L-BTC or service access rather than BTC/USD, and the supposed demand is not a mandatory BTC purchase. Strongest bearish case: incomplete return leaves a shortfall; counterexample other reserves/insurance or changed liabilities could cover it. No full-reserve deficit estimate is justified.
Validity: neither branch supplies numerical cash-only spot expectancy. Soundness limited to primary self-reports and indexed transactions. A current payout txid linked to an authorized service would change the service assessment; a separate replicated net event rule would change size. Sources https://x.com/Blockstream/status/2097127976672342487 ; https://blockstream.info/api/tx/a6d697a25266ce3c78774fd1d75f896b7af522ada209b0f6228ea497bc49a46d ; https://www.btse.com/blog/update-on-liquid-network-security-incident/ ; https://sideswap.io/docs/ .
principal returned BTC3400
confirmed recipient BTC598.5
independent incidents1
BTC 4h residual log percent-0.01
verified subsequent reserve spends0
approved allocation0
Positive daily trend coexists with short-window weakness and a meaningful cost hurdle
Start state allcash, goal reliable after-cost profit; binding constraint independent event-price edge and demonstrated service. Premise1 observations12341/12342 verify raw on-chain signatures against issuer-domain public key, supporting attribution of a patch statement and affirmative reply. This supersedes blanket missing authenticated repair statement; mathematical verification does not establish current custody keys or complete revocation audit. Premise2 direct operator post12343 datedSep8 01:00:47UTC says software deployed and preparing coordinated restart, not completedrestart. Premise3 full history12335/12336 shows3400BTC principal returned plus.00008 message-carrierBTC,598.4998345recipient,3597.47197696watchedreserve,zero subsequent confirmed reserve spend. Premise4 liabilities and all federation addresses are not enumerated; oneaddress cannot establish reservecoverage. Conclusion repair progresses, but cannot infer redeemability or a BTC buying edge. Validity: returnedassets+statement do not logically imply successfulwithdrawals. Soundness bounded by indexed transactions, HTTPS key and operator self-report; counterexample authorized new wallets could be paying even without watchedaddressspend. Operator/bridge users benefit if repair restoresconversion; intermediary incentives favor reassurance, so verify realizedpayout. Source https://blockstream.com/pgp.txt ; https://blockstream.info/api/tx/3c101c8e053a6cbc7c4455e4fec017c8f2962ed86dee70e72fb8bafbb1d7bde9 ; https://x.com/Blockstream/status/2097127976672342487 ; https://blockstream.info/api/address/bc1qdlld6antmv4xug242ed83q7k4rqw50cwfns38szx4qu2f4jwaxxsuhwxxr/txs/chain . Prior uncertainty narrows to currentkeys/liabilities/payouts; new verified current payout plus issuerrecovery statement would change service status, not automatically allocation.
returned BTC3400
recipient BTC598.5
reserve BTC3597.47
verified mathematical signatures2
post time UTC2026-09-08T01:00:47Z
later reserve spends0
Final decision: retain $235.86 cash after broader research and failed pullback validation
Start state: crypto ledger rechecked at02:16UTC shows only closed trade10, no open or pending positions. Fund cash235.86, account equity850.99. Goal: repeatable net profit. Binding constraint: independently validated entry/exit expectancy.
Premise1: Frozen RSI2 trend-pullback hourly holdout loses41.79–51.78% over85–120 closed trades per asset. Net expectancy BTC−.555%,ETH−.703%,SOL−.608%,LTC−.786%,LINK−.579%,UNI−.512%,XRP−.651%; all block-bootstrap95% intervals below zero. QuarterKelly=max(0,.25*(p−(1−p)/(avgwin/avgloss)))=0 for every asset.
Premise2: Daily holdout has0–2 closed trades, insufficient for payoff calibration, and none of seven current daily entry conditions is active. Small positive LTC/LINK daily outcomes each haveN1.
Premise3: BroaderUS/overseas research finds no independent corporate or token entry edge: UNI peakburn cannot be annualized; Metaplanet authorization has zero disclosed executed purchases; IREN is transitioning to AI; ETF/stablecoin levels do not establish fresh inflows.
Premise4: Liquid returned principal remains3400BTC, while authenticated control and current pegouts remain unverified. This is operational uncertainty, not a calibrated directional trade.
Conclusion: No PROPOSE, EXIT, ADJUST or FUND. Zero size; caps212.7475perposition/235.86fund/8.5099plannedrisk remain ceilings, not allocations. No positions require R/target/time-horizon adjustments.
Validity: necessary evidence gates fail, so stand-down follows. Soundness is limited to measured rules and sources, not every crypto strategy. Strong counterargument: trend continuation could reward simply buying; but current exposures do not supply independent after-cost edge, and optimizing on this holdout would contaminate it. Who benefits from unvalidated turnover: venue/market makers. Fees reduce principal, then risk capacity, then runway. The349.01monthly shortfall cannot change sizing or targets.
Record references: screen897,statistical827,simulation846,technical847,quality900,backtests837–843/853/886/889–893, custody821,fundamentals829/835/836/896. Gaps: pristine prospective validation, actual15min exits/profit management, authenticated service, current onchain burn/flow history, liveAssets/overseas execution. Jobs69/70firstscheduled execution failed503; remainarmed for02:30/02:45 retries, funding71verified6/6fresh. Monitor70armed04UTC. Sources https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://runway.adityaperswal.workers.dev/data/crypto-snapshots ; primaryissuer/custodyURLs in cited recordedanalyses. No orders placed.
fund cash235.86
equity850.99
position cap212.75
planned risk cap8.51
hourly return best-41.79
hourly return worst-51.78
hourly samples85–120
daily samples0–2
approved Kelly0
new proposals0
scheduled collector failures2
next review2026-09-08T04:00:00Z
Recent aggregation reconciles, but sparse history and fee burden remain binding