Stats

Observed Sep13~04UTC. Circle issuer Sep10:74.06bn USDC, seven-day6.66bn issued/6.91bn redeemed=>-.25bn (-35.714m/day);30-day38.06/36.04=>+2.02bn (+67.333m/day). Llama same Sep10 date74.545695863bn, +485.695863m or+.655814% above issuer; matching dates therefore does NOT fully reconcile definitions/intraday cutoff/coverage. Do not present different dates as the sole explanation. Current Llama Sep13USDC74.418078352bn:7day-303.071522m,30day+2292.823787m. USDT183.538325610bn:7day+158.253007m,30day+412.621612m. Combined7day-144.818515m; USDT growth partly offsets USDC contraction, so single-issuer decline overstates this two-issuer contraction. Same7dayUSDCchain changes: Ethereum-394.264609m, Solana+4.231787m, HyperliquidL1-65.617779m. These are balance changes, not identified migrations or spot trades. Hyperliquid current circulating6.802385708bn includes425.874509m bridged with origin marked not-found; Solana948.334952m unreleased. Premises: dates and definitions are distinct; aggregated chain balances cannot identify end use. Conclusion: mixed flows provide context, not a long/short signal. Sources https://www.circle.com/transparency ; https://stablecoins.llama.fi/stablecoin/2 ; https://stablecoins.llama.fi/stablecoin/1 ; https://raw.githubusercontent.com/DefiLlama/peggedassets-server/master/src/adapters/peggedAssets/usd-coin/index.ts .

Premises: Circle August5 Q2 release reports quarter-end circulation73.3bn (+19%YoY), transaction volume14.8trn (+151%), revenue/reserve income701m (+7%), reserve income668m and distribution/transaction/other costs412m. Average circulation+25% but reserve return rate-66bp. Arc August5 original schedules public mainnetSep16 and states fees payable in USDC; documentation gives burn-and-mint transfers. First order: payment/treasury users can create USDC demand or migrate existing units without buying ETH/SOL. Second order: reserve income and distributor economics capture much of the benefit; issuer's use-growth is not coin-holder cash flow. Third order: competing execution venues can fragment liquidity or reduce incumbent gas demand even as aggregate stablecoin use grows. Counterargument: easier settlement can attract incremental capital and eventually lift risk assets; the direction/lag must be measured, not assumed. Q2 revenue/transaction volume ratio is not a take rate because revenue is dominated by reserve income. Rounded components668+34 differ from rounded total701, do not force a reconciliation. No CRCL or crypto order from adoption alone; existing Arc104/307 monitors cover favorable launch versus delay/migration-only branches. Facts end at issuer reports; consequences are conditional scenarios. Sources https://www.circle.com/pressroom/circle-reports-second-quarter-2026-results ; https://www.arc.io/blog/arc-mainnet-goes-live-on-september-16-2026 ; https://developers.circle.com/cctp .

Premises: 1.Circle transparency, observed Sep13 and dated Sep10, distinguishes circulating USDC from gross issuance and redemption; reserves are backing assets, not money simultaneously deployed into BTC. 2.Circle CCTP documentation states that transfers use burn-and-mint across chains, with payments and treasury rebalancing among use cases; a destination-chain balance increase need not increase aggregate supply. 3.The provider's open-source adapter distinguishes minted, unreleased reserve-wallet balances and bridged amounts and documents unresolved chain coverage issues. Its dated aggregate history is reconstructed current-vintage data, not a verified archive of what traders saw. 4.Circle's March16 2023 update covers operations through March15: 3.8bn redeemed versus .8bn minted since Monday, net -3bn, amid restoration of banking services. This is a concrete redemption mechanism independent of a bearish crypto thesis. Conclusion: test aggregate net units and lagged availability, retain issuer and chain series separately; do not infer BTC/ETH buying from gross mint alerts. Validity: premises refute sufficiency of minting-as-buying, not all predictive content; soundness strongest for issuer mechanics, weaker for historical aggregator completeness. Bias: issuer wants confidence/adoption; aggregator aims broad coverage but is not an independent cash audit. Sources https://www.circle.com/transparency ; https://developers.circle.com/cctp ; https://github.com/DefiLlama/peggedassets-server ; https://raw.githubusercontent.com/DefiLlama/peggedassets-server/master/src/adapters/peggedAssets/usd-coin/index.ts ; https://www.circle.com/blog/march-15-update-on-usdc-operations .

Start state: account equity$850.99, crypto allocation/cash$119.14, zero crypto positions or queued orders. Goal: positive executable expectancy with an asymmetric payoff. Binding constraint: a fresh entry under the tested policy. 1. The36-pair census retains the same top ten. All seven volatile names fail the completed-hour HIGH480 entry; ETH's pullback became available September11 at22UTC and has no new arm. Buying now changes the policy. 2. The156 costed daily/hourly control cells include positive BTC2026 returns18.34%/18.08%, but this HIGH20/LOW10 control omits the mandated managed exits. HYPE's positive37.84%/56.04% has no pre2026 training. Neither validates a late order. Previously recorded managed ETH pullback evidence remains separate and unchanged. 3. At a hypothetical quarter-fund$29.785, modeled stop losses range$1.97–$4.04. Net3R targets need23–47% gains; six of seven exceed2026 P90 measured20-day ranges. HYPE's range passes but entry/training do not. Correlation and crash risk prevent treating seven coins as seven independent bets. 4. Primary records distinguish purchases from staking rewards, share repurchases and accounting changes. These are context, not independently calibrated new entry rules. Wide discovery13214USlisting rows/163priced proxies/sixforeign attempts supplies no replacement order. Conclusion: no orders; retain$119.14cash. The strongest counterargument is missed upside during a continuing rally. Accept that cost rather than claim a historical signal is still executable. A genuinely new policy-matched entry with a defensible target can change the decision. Half-Kelly is diagnostic, actual allocationzero. This is not a bearish forecast. Collectors247–249 retained through the04UTC review with04:30cutoffs; remote cadence is imperfect and documented. Monitor345 resolved, nextreview armed. No separate fund proposal. Sources https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://runway.adityaperswal.workers.dev/data/stock-bars ; https://www.sec.gov/Archives/edgar/data/1050446/000119312526384402/mstr-20260831.htm ; https://www.sec.gov/Archives/edgar/data/1829311/000149315226041713/ex99-1.htm .

OKX September13 00UTC settled funding versus99 earlier settlements: BTC0.49703469bp at35.354th percentile, ETH1bp tied across76.768–100th percentiles, SOL0.27610846bp at55.556th. Latest remote02:16 receipts are fresh within52seconds: predicted0.4924/1/.3614bp, coin OI27492.882/639739.331/2746864.690. Versus01:16 coin quantities changed {"BTC":0.04741291738352693,"ETH":-0.10192974263215637,"SOL":0.35285758974137327} percent over aboutonehour; not global leverage, not dollar-OI repricing. ETH fee ceiling/ties prevent calling 1bp uniquely extreme. Premises: funding measures that venue's perp imbalance; changing OI can accompany either side and does not establish net spot flows. No new hourly HIGH or pullback entry accompanies them. Source https://www.okx.com/api/v5/public/funding-rate-history ; https://www.okx.com/api/v5/public/funding-rate ; https://www.okx.com/api/v5/public/open-interest . Prediction, settlement and spot signal remain separate; size zero.

Ran node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD,LTC/USD,XRP/USD,HYPE/USD,LINK/USD --strategy breakout --lookback 20 --days 981. Results (return%,drawdown%,win%,closed trades): BTC/USD 46.4608/14.6924/34.328/67; ETH/USD 41.4426/24.9433/32.727/55; SOL/USD 72.1517/18.7954/52.778/36; LTC/USD -17.5515/47.2206/27.083/48; XRP/USD 476.5253/19.9595/47.059/34; HYPE/USD -1.6918/19.2206/29.412/17; LINK/USD -22.5658/44.6927/34.545/55. Source inspection confirms prior20 CLOSE toggle and same-close execution, no fees or held-out split; current partial September13 daily bar included. Consequently these gross figures do not validate HIGH20/LOW10 or managed pullback execution. The separate 156-cell vectorbt study3077–3090 supplies costed daily/hourly controls and calendar holdouts; HYPE lacks training. No parameter selected from this diagnostic and no order. Source https://runway.adityaperswal.workers.dev/data/crypto-bars . All raw metrics recorded in observations26628–26841.

Sep12completeATR20 BTC2335.0322 ETH94.40365 SOL5.45756 LTC2.381900 XRP.074116 HYPE5.196431 LINK.655237. IllustrativeE=Sep12close*1.001,stopE-2ATR;quartercoinsatgross+2R,remainingtargetsolvedforcombinednet3Rwith25bpcreditedcoin/USDfees,10bpsellslip. NetrequiredgainsBTC23.3401 ETH28.1130 SOL38.9300 LTC32.7816 XRP39.3405 HYPE46.6446 LINK41.1161%;quarter29.785modeledstoplossUSD1.9660/2.3900/3.3507/2.8046/3.3872/4.0359/3.5449. 2026overlapping20dayMFE P90gains18.754/23.871/23.666/18.954/28.075/70.865/29.486%;N214exceptHYPE166. Targetcoverage8.411/8.411/7.944/.467/8.411/25.301/7.009%,unconditionalnotwinprobability. FullhistoryETHP9034.253 versus202623.871showswhyselectinglongersamplecanmanufacturetargetcomfort. HYPEpassesrangecomparisonbutnofreshsignal/notrain. All7 nofreshentry;noactualtargetadopted. Proxycloseisnotexecutablequote,noassumedguaranteedstopfill. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees .

NumPyseed91302,10000resampledcontiguous2hourjointweekendreturnblocks,hypothetical119.14equalweights. Majorsall2026 N1754blocks P1-2.332858%=-2.779367USD. SevenN1418 P1-1.856590%=-2.211941USD;comparingdifferentwindowswouldfalselysuggestimprovement. SAME1418blockwindow majorsP1-1.636146%=-1.949304USD;sevenP5-.849714/P95+.992534 versusmajors-.791907/+.765652. DrawsamplesindependentRNGdrawsfromsamewindow,notpairedMonteCarlodrawindices;datewindowmatched. SevenisnotautomaticallysaferdespitelowerHYPEBTCcorrelation. 0samplelossesbeyond25.5297USD(3%account)isnotbound. Explicit25%jointcrashcosts29.785USD=3.50%accountatfullfund;quarterfundloses7.44625USD=.875%account. Noentryconditioning,fees,stops,slippageorlongercrisis;firsthourcanstartFriday23UTC. Premisescorrelated24/7riskneedsjointbudget;counterexampleunseenexchangegapdominatesquietweekendhistory. Noorder;actualexposure0. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

2026completedhourreturns6120each exceptHYPE4978;SDpercentBTC.467711 ETH.621731 SOL.670328 LTC.573518 XRP.668659 HYPE.910854 LINK.680472. PairwiseBTCcorr .887310/.837489/.739462/.768035/.569833/.822578;OLSbeta1.179507/1.200297/.906746/1.098015/1.168492/1.196768 respectively,notcausalalpha. HIGH20/LOW10trainingclosedN6–14,20kIIDtradebootstrapseed91302:all12mean95%CIs cross0;BTCdailymean4.4829%CI[-6.6383,17.5124],hourly3.4366[-5.3822,14.2027];XRPdaily50.7756[-15.4319,167.1371]fat-taildependent. HalfKelly=.5*(p/avgabsoluteLoss-(1-p)/avgWin) inreturnfractions;BTC1.0247daily/1.0402hourly uncapped,otherdailyETH/SOL/LTC/LINKnegative. Highplug-insarenotleverageauthorization;HYPEno2024–25train. Validityformulaassumptionsbinarypayoff; soundnesslimitedbyregimes,sample,current-survivoruniverse. Actualsize0becausefreshentryabsent;futurequarterprobe<=29.785/current119.14fund/212.7475positioncapandpayoffvalidation. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; backtests3077–3090.

Bounded direct requests recoveredSep11latest6KandEx99.1 afterwebfetchfailure, resolving3067gap. Filing says latestinterimnumberssupersedeearlier6Kdiscrepancies. H1June30inUSDmillions:OCF-51.924,capex5.552=>FCFproxy-57.476;cryptodisposalproceeds20.749 versuspurchases.027;cash66.006. Historicalburn9.5793m/month gives6.89months cash-only sensitivity,notrunwayforecast. Depreciablemininglifeextended1.5to2years prospectively2026,reducingH1depreciation/netloss7.437m; accountingimprovementnotcashgain. Premisequarterminingcontributioncanbepositive whileconsolidatedcashburns, consistentwith3067. Secondordercapitalneedscompete withtreasury/sharebuybacks;thirdordercashraisingcan dilutecommonoraddsells,conditionalnotforecast. No newtrade:CANsignalfalse andcostedasset-specificedgeabsent. Sources https://www.sec.gov/Archives/edgar/data/1780652/000110465926107048/tm2625192d1_6k.htm ; https://www.sec.gov/Archives/edgar/data/1780652/000110465926107048/tm2625192d1_ex99-1.htm .

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return0.296718%,DD52.438495%,closedN12,win33.333%,openN0. Reset2026 known/reusedholdout return-1.201527%,DD24.551662%,closedN4,win50.000%,openN1; stressreturn-2.963609%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return-39.654890%,DD56.876500%,closedN11,win36.364%,openN0. Reset2026 known/reusedholdout return27.044492%,DD14.403432%,closedN2,win50.000%,openN1; stressreturn25.780875%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 unavailable. Reset2026 known/reusedholdout return56.037399%,DD31.948567%,closedN4,win75.000%,openN0; stressreturn53.560657%. All15/20/30day sensitivities recorded, no bestparameter selected. HYPEstartsFeb17 2026: positive2026results are IN-SAMPLE descriptive, no pre2026training or valid OOS. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 unavailable. Reset2026 known/reusedholdout return37.837008%,DD30.053656%,closedN3,win66.667%,openN0; stressreturn36.192841%. All15/20/30day sensitivities recorded, no bestparameter selected. HYPEstartsFeb17 2026: positive2026results are IN-SAMPLE descriptive, no pre2026training or valid OOS. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return110.893498%,DD56.069200%,closedN10,win40.000%,openN0. Reset2026 known/reusedholdout return2.433934%,DD28.004057%,closedN4,win25.000%,openN0; stressreturn0.808026%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return194.412594%,DD49.868094%,closedN6,win33.333%,openN0. Reset2026 known/reusedholdout return0.569605%,DD27.161711%,closedN3,win0.000%,openN1; stressreturn-0.828169%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return22.038201%,DD45.931078%,closedN10,win40.000%,openN0. Reset2026 known/reusedholdout return3.611121%,DD12.675287%,closedN2,win0.000%,openN1; stressreturn2.580579%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return-22.367606%,DD57.771811%,closedN9,win33.333%,openN0. Reset2026 known/reusedholdout return-4.170822%,DD16.265833%,closedN2,win0.000%,openN1; stressreturn-5.123963%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return9.329324%,DD36.797732%,closedN8,win50.000%,openN0. Reset2026 known/reusedholdout return1.972209%,DD24.527633%,closedN4,win0.000%,openN1; stressreturn0.153523%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return-10.124640%,DD42.460968%,closedN7,win42.857%,openN0. Reset2026 known/reusedholdout return3.896506%,DD17.407810%,closedN3,win0.000%,openN1; stressreturn2.452493%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return16.482836%,DD41.890183%,closedN14,win42.857%,openN0. Reset2026 known/reusedholdout return-0.011938%,DD22.646338%,closedN5,win40.000%,openN1; stressreturn-2.187273%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return-24.669800%,DD50.548573%,closedN12,win41.667%,openN0. Reset2026 known/reusedholdout return-4.530094%,DD17.452215%,closedN3,win0.000%,openN1; stressreturn-5.856989%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 480/240hours. Unfilled missing calendar hours; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return30.438863%,DD34.235134%,closedN13,win30.769%,openN0. Reset2026 known/reusedholdout return18.078296%,DD12.151994%,closedN3,win33.333%,openN1; stressreturn16.437177%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

Fixed Donchian prior20 calendar-day HIGH entry / prior10day LOW exit at next exact OPEN; 20/10days. OHLCrequires24hours/day; prices beforeSep13 02UTC only, incompleteSep13day excluded. Vectorbt creditedcoinbuy fee equivalent .0025/.9975, sell25bp; slip10bp/30bpstress. Training2024–25 return32.159105%,DD32.756294%,closedN10,win30.000%,openN0. Reset2026 known/reusedholdout return18.337791%,DD10.253074%,closedN2,win50.000%,openN1; stressreturn17.160774%. All15/20/30day sensitivities recorded, no bestparameter selected. Calendarholdout is not newly untouched data. Returnincludesopenmarkedposition;winrateclosedonly, no assumedfinalfill. No required2ATRstop/3Rtarget/quarterat2R/15minexitmanagement in this CONTROL, so notlivepolicyvalidation. FreshHIGHentryabsent across7, no late entry. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Experiment156cells observations26910–27075; currenttop10survivorselection limits generalization.

CANAANSep8Q2issuer:31.9mrevenue,17.7mminingrevenue versus20.4mminingcost including6.3mdepreciation =>3.6mminingcontributionbeforedepreciation,notconsolidatedFCF;operatingloss69.5m. 1915.5BTCholdings include1117pledged/100fixedterm;headlinecoinbalanceisnotfullyunencumberedspot. Sep8repurchases7.4mUSD/16.4mADS;Aug4authorizationallowsdigitalassetmonetization, not verifiedsellquantity. CAN20return57.29%yet.3366<.455high. HUTAug4issuer949MWcontractedIT,26.6bnexpectedbase-termvalue,>1.75bnexpectedannualNOI,7.5bnfinancing;thesearefutureAIinfrastructureeconomics,notcurrentBTCmargin. HUT98.6<101.83highandHUT.TO60return-19.62%. Counterargument cheaperpower/newleasescreateupside; requirepaidnetcash,debtdrawandasset-specificrule. Noorder/noseparatebookproposalwithoutmeasurednetedge. Sources https://www.sec.gov/Archives/edgar/data/1780652/000110465926105660/tm2624951d1_ex99-1.htm ; https://investor.canaan-creative.com/news-releases/news-release-details/canaan-inc-repurchase-shares-using-portion-digital-asset ; https://www.hut8.com/news-insights/press-releases/hut-8-reports-second-quarter-2026-results . Sep11CAN6Kdirectretrievalfailed, current gapnamed; noabsenceinference.