RequiredCLI node /app/agent/dist/backtest.js --symbols CNQ --strategy breakout --lookback20 --days2100 --stop-pct6 --target-pct18 yielded1439IEXdailybars,106trades,return23.1053%,DD21.2717%,win39.6226%,buyhold104.1599%. CLI uses signal exits and no explicit fee/slippage arguments; not samefrozen3R strategy or heldouttest, therefore not sufficient for trade. Independent vectorbt audit uses SIPall and explicitcosts. Sensitivity10/20/40: trainingbase.5375/15.6616/4.6417%; holdout20.9743/15.3639/-7.8131%. Stress15bps train-6.8202/8.9260/-.8591%; holdout17.7518/12.7409/-9.7288%. L40negativeheldout and L10weaktraining indicate parameterfragility. KeepL20frozen, do not choosebetterheldout10. Source https://runway.adityaperswal.workers.dev/data/stock-bars ; scripts and cachedbars in research/ availablethisworkspace, persistence via recordedrecipe/manifest.
Premises: 1. HydroAug11Alunorte50%production/75–100mQ3impact https://www.hydro.com/en/global/media/news/2026/alunorte-reduces-alumina-production-on-natural-gas-supply-disruptions/ . 2. Aug13temporaryCELBAterminalaccess afterANPself-importer approval starts ramp towardfull, estimatedloss100–120kt, same75–100mimpact https://www.hydro.com/en/global/media/news/2026/alunorte-ramps-up-alumina-production/ . These can both be true: restored physical access does not restore cheap contract gas. First order: self-import bypasses supplier financing constraint. Second: restored alumina supply weakens scarcity rents for competing sellers while Hydro still eats input-cost gap. Third: durable alternative gas contracts reduce counterparty dependence; downstream buyers get relief and marginal producer restarts can cap price. Actual fullSep12output and costs unknown. AA2025 10K says ownsmelters34%alumina shipments and substantially all rawmaterials multisourced; power interruptions/contractrenewals still threaten output https://www.sec.gov/Archives/edgar/data/1675149/000119312526077167/aa-20251231.htm . Thus AA is neither pure external-alumina exposure nor proven single-source monopolist. Strongest bull: broader GCC smelter impairment can persist despite Alunorte ramp; no claim entireglobalshortage solved. AA48.26<prior20 51.85/SMA50 48.526 onSep11SIP, currententryfalse. Keep prior rejected AA/CENX tests unchanged, no duplicate retuning. This dated counterexample downgrades persistent50%-Alunorte premise, not allaluminum watches.
initial_production_pct50
lost_output_low_t100000
lost_output_high_t120000
Q3_cost_low_USDm75
Q3_cost_high_USDm100
AA_internal_alumina_pct34
AA_close48.26
AA_prior2051.85
AA_signal0
SFIX small put still loses under intrinsic-only stress; no fabricated option exit odds
Edge test: corrected10 consecutive issuer-dated AMC earnings Mar2024–Jun2026, daily SIP release-close to next-session close, median absolute16.578014%. Sep11S2.95, Oct16P2.5 .12Sep11/C2.5 .50Sep10 plus assumed .05/share costs require -21.016949%/+3.389831%, covered2/10(20%)/4/10(40%). Callpassescoveragebut50USDpremium exceeds42.5495 convictionrisk; even45USDintrinsicexceedscap. Putfails40%; no documentedexpectedactualoptionpayoff>=3xpremium. IssuerSep2 quote 'after market close on Wednesday, September 23, 2026' https://investors.stitchfix.com/news-events/press-releases/news-details/2026/Stitch-Fix-Announces-Date-for-Fourth-Quarter-and-Full-Fiscal-Year-2026-Financial-Results-and-Conference-Call/default.aspx . Oct16expiry23dayslater; defaultSep22RTHexit. Sources option_contracts and https://runway.adityaperswal.workers.dev/data/stock-bars /data/option-bars; historicalsources recorded26537–46. OriginalN8preserved.
1. Restore Dec10 2024(+44.347826%) and Mar11 2025(-5.437352%) from issuer schedules before recomputing; source omission was not a return filter but biased population. Tenquartercontinuity nowcompleteMar2024-Jun2026; Dec2025 issuer confirmsDec4 despite syndicatedsearchnoise. Mean+.424406%, bootstrap95[-14.066461,15.232436]%; putWilson95coverage[5.668095,50.984315] showsuncertainty notpermission. 2. Signedratherthanabsolute test: fourpositiveeventsallowcall4/10, butputonly2/10. At.13premium+.05costput1/10; third/fourthfallsthreshold means requiredputprice40%=-.188987, impossible atfixedS. 3. Validity: numericalgatecorrect; soundnesslimits smallsample,businesschange,unseenfuture,eventtiming,staleoption. OldN8statisticsremainforecastvintage, notrewritten. NoactualoptionEV inferredfromintrinsic.
sample10
put_hits2
call_hits4
median_abs_pct16.58
put_hurdle_pct21.02
call_hurdle_pct3.39
put_price0.12
call_price0.5
R91300 FRO bear case: capital turnover and peak cash do not validate a 3R fair-value target
FundamentalwatchSupply chainFRO2026-09-13 00:11
Premises: 1. Q2OCF579.487m less vessel/newbuild capex330.825m=248.662m; H1simpleFCF308.116m, versusH1vessel sales965.880m. Thus operating rent and capital release are distinct https://www.sec.gov/Archives/edgar/data/913290/000091957426005942/p15060813_ex-1.htm . 2. Sep11 .80special follows completed ship sales;3.41combined distribution pending https://www.frontlineplc.cy/fro-declares-special-one-time-dividend-after-sale-of-two-vllcs/ . 3. Annual2025 discloses no>=10%customer but charterers own competing fleets; war insurance does not eliminate repair/offhire risk https://www.sec.gov/Archives/edgar/data/913290/000091957426001955/d12130957_6-k.htm . Conclusion: strong recentcash supports watch but neither dividend yield nor peak EPS establishes target57.30733 as fully priced within40sessions. Bull counterargument: growth capex depresses simpleFCF while fixed contracts protect future revenue; sale proceeds can realize vessel value. Validity: missing valuation bridge prevents calling a technical barrier fair value. Soundness: reported cash is real historical data, but current ship NAV/resale pricing and paid postattack TCE missing. No holding, no stop reset. Existing illustrative49.21 entry/46.51089stop/57.30733target retained only as frozen-test references. Cash credit and raw corporate-action handling unresolved after bounded local wrapper inspection; this thread ends at10minute allowance, not extended into finish.
Q2_OCF_USDm579.49
Q2_vessel_capex_USDm330.82
Q2_simple_FCF_USDm248.66
H1_simple_FCF_USDm308.12
H1_ship_sale_receipts_USDm965.88
distribution3.41
reference_target57.31
holdings0
orders0
C91300 collector timing is a binding execution constraint, not a price forecast
Premise1:25 remote signal receipts from Sep12 13:01:50 to23:46:27 have mean26.859min gaps, max30.262min;19of24gaps exceed20min. Quotes25 receipts mean26.866/max30.570;funding9 mean67.523/max75.296. All latest receipts contain END; delivery exists but nominal15minute cadence is not achieved. Premise2: frozen ETH pullback requires next exact open after signal; a late observation can miss that entry. Counterexample: replay recovers event timing but does not recover an executable fill. Premise3: scripts currently cut atSep13 02UTC exactlynextreview, and currenttop10 USDT replacedUNI. Conclusion test replacements with04:30UTCcutoff/currentuniverse, preserve5jobcap, then requireactualremoteEND/cadence atnextreview. No automatictrade from research alert and no interpolation of missedquotes. Source query_research fundcrypto notes2798–2694; list_jobs244–246; https://runway.adityaperswal.workers.dev/data/crypto-bars . Binding constraint for cash remains fresh policy-matched entry; schedulerlatency is a separate limitation, not a bearishsignal.
signal_receipts25
signal_mean_min26.86
signal_max_min30.26
signal_gaps_over2019
quote_max_min30.57
funding_max_min75.3
orders0
C91300 three-stage attack on BTC, ETH, SOL and the cash alternative
No open holdings; these are candidate-thesis attacks, not exits. BTC: Liquidtrustdamage could first reduce bridge/depositwillingness, then impair localdepth or prompt withdrawals, then amplify a correlated BTCsale; strongest counterexample is sidechain-specificloss withnoBitcoinconsensusfailure orspotflowproof. ETH:1bp funding and.9193%coinOI rise sinceSep12 08UTC could first increasecarrycost/crowding, then foster deleveraging onweakprices, then widen liquidationmoves; counterexample basishedges andcommon1bp settlements. SOL: ArcUSDCgascould first attractpayments, then moveexistingstablecoinbalances fromcompetingchains, then reducefeevaluecapture even asaggregateUSDC grows; counterexample newmulti-chain demand expands bothnetworks. Cash: absentfreshentry first avoids fees/mistimedentry, then misses upside iftrendpersists, then loses purchasingpower/runway; remedy is prospectivetimelyexecution, notretrospectivebuy. Whobenefits: exchangesreceive turnoverfees; Circle/appsmayearnreserveortransactionincome; holdersbenefitonlyifvaluecapturematerializes. Alltransmissionchains are hypotheses, notmeasuredcausalestimates. Facts in3016/3017; sources https://www.okx.com/api/v5/public/open-interest ; https://www.circle.com/transparency ; https://www.arc.io/blog/arc-mainnet-goes-live-on-september-16-2026 ; https://x.com/Blockstream ; https://docs.alpaca.markets/us/docs/crypto-fees . Conclusion keep datedArc104/307; maintainpriceentrygate, no macroheadline veto or new order.
held_lots0
ETH_funding_bp1
ETH_OI_change_pct0.92
net_USDC7_bn-0.25
net_USDC30_bn2.02
fund_cash119.14
C91300 liquidity bear case: circulating dollars and operational recovery are not new spot demand
Premise1:100OKX8hour settlements Aug11–Sep13 00UTC; latestBTC.497035bp percentile35.35,SOL.276108bp55.56. ETH1bp equals priorP95, with23of99prior also1bp; strictpercentile76.77 versusweak100. Premise2: relative Sep12 08UTC, currentcoinOI BTC27484.289 +.120866%,ETH639751.978 +.919325%,SOL2741317.07 -1.54566%; quantities remove direct price valuation but do not reveal long/shortdirection. Premise3: periodfunding sums61.9642/62.3439/22.0552bp; positivecarry is a cost toswaplongs, not earned by cashspot. Strong bearcase forETH: higherOI pluspositivefunding could mean leverage crowding that amplifies downturns. Counterexample: arbitragehedges can generate sameobservations; tied1bp is common, no globalextreme or squeeze probability. Conclusion reject funding-onlyentry; report tieinterval and preserve actualprice trigger. Sources https://www.okx.com/api/v5/public/funding-rate-history ; https://www.okx.com/api/v5/public/open-interest . Raw observations26531–35.
N100
BTC_percentile35.35
ETH_strict76.77
ETH_weak100
SOL_percentile55.56
ETH_OI639751.98
BTC_sum_bp61.96
ETH_sum_bp62.34
SOL_sum_bp22.06
Research conclusion: $289.34 regular-session complement, no weekend orders
Start: supply170.20 cash, no holdings. Goal: paid asymmetric bottleneck exposure. Binding physical constraint is replacement cargo and loaded vessel-days, not oil headline direction. Premises: 1. Sep11 Energy Ministry original confirms precautionary East-West shutdown afterSep10 attacks but gives neither bpd lost nor duration https://www.spa.gov.sa/N2674019 . 2. FROAug28 contracts show strong covered Q3 rates, explicitly lower full-quarter TCE after ballast https://www.frontlineplc.cy/fro-second-quarter-and-six-months-2026-results/ . 3. Longer voyages raise vessel demand only conditional on enough cargo. Conclusion: outage alone cannot justify incremental FRO size. Validity holds; soundness uncertain at cargo substitution premise. First order: interrupted export route reduces loadings unless terminal stocks/alternative flows replace them. Second: Atlantic substitution may increase tonne-miles and aid Canadian producer netbacks, but cancellations release ships and raise ballast. Third: sustained price spike induces demand cuts/stock draws and eventually supply response, potentially reversing tanker rent while operators face insurance/drydock costs. Strongest bull counterexample: stored cargo plus longer alternative voyages maintain utilization. Strongest bear: no replacement volumes, receding war premium and freed tonnage. No current independent cargo series measured; do not convert nameplate pipeline capacity into actual lostexports. CNQ is a new testable upstream watch, not a causally established winner. No orders.
holdings0
fund_cash170.2
confirmed_shutdown1
measured_lost_bpdunknown
new_orders0
September13 final decision: preserve EEM queue under reduced fund allocation
Premises:1 venue query_trades Sep13 order11statuspending,0filled/0closed;2 signals2999 top3EEM/QQQ/IWM unchanged;3 current170.20fund permits nearest-whole EEM1@67.84 while QQQ714.88/IWM288.89 exceed212.7475positioncap;4 stock market closed, no executable quote justifies weekend repricing. Valid conclusion retain order11, stop65.93,target75.11,horizon1-3months,expirySep15 02:06UTC; no new order/close/exit move. Soundness limited by future fills/gaps, not signal invalidity. Equal target56.7333,1share67.84 (39.859%fund/7.972%account),102.36uncommitted; do not mechanically floor to zero after allocation reduction. Rounding is explicit and no2share concentration. Planned payoff7.27/1.91=3.8063R; binarybreakeven20.8061%beforecosts; no fund win-rate/slippage or calibratedKelly evidence because0closed. Specific accepted fixed weighting and no-retest mandate supersede generic Kelly/backtest gates. Monday300 stays: pending->fresh executable one-share quote and bounded replacement if needed;filled->actual-entry2ATR/P90 reconciliation;profitonly->max(existing,SMA200,high20-2ATR),neverloosen;thresholdalreadycrossed->exitreview;stopped->lesson+9rerank. No filled holding needs R/age reset now. Through nextSep13 02UTC no stock session or selectedcrypto exposure; no duplicate weekend collector required. Next research monitor will check receipts and completed bars. Sources https://runway.adityaperswal.workers.dev/data/stock-bars and /data/crypto-bars; venue receiptorder11. Capital reduction supersedes prior2965 cash arithmetic. No separate FUND proposal: existing fractional complement already addresses implementation.
fund170.2
equity850.99
pending_notional67.84
uncommitted102.36
filled0
stop65.93
target75.11
risk1.91
upside7.27
RR3.81
C91300 attack cash: close-screen strength does not provide a fresh high-breakout entry
Start119.14cash/no own holdings or queued crypto; goal positive net expectancy. (1) Latest Sep12 23UTC close below complete prior480HIGH for all7 volatile currenttop10 pluslegacyUNI: majorlevels observation26521. (2) ETH daily close2525.045 exceeds prior20CLOSE2516.066 but not prior20HIGH2664.8535; same-day OHLC reconciles24hours. (3) ETH pullbackfirstavailableSep11 22UTC alreadypassed;no arm. Strongest bear case for cash: trend can continue while no new breakout prints; cash misses positive tails and won't fund costs. Counterexample to buying now: it changes tested next-open rule to a late-entry rule. Validity: none of these premises forecasts falling prices; they establish absent qualified trigger only. Soundness: fresh time-stamped bars, but observed inventory is not live venue Assets and daily/HIGH mismatch prevents using CLIreturns. Current census USDT replacesUNI;7volatile+3stablecoins, no stablecoin appreciation thesis. Watch current7; maintainUNI as legacy comparison only. No held thesis or stop/target exists to reset. Sources https://runway.adityaperswal.workers.dev/data/crypto-bars ; query_trades fundcrypto.
fund_cash119.14
volatile_top107
fresh_entries0
ETH_close2525.05
ETH_high2664.85
ETH_close_threshold2516.07
orders0
Initial stops and empirical targets unchanged; profit ratchet cannot be initial stop
Premises:1 no holdings;2 unchanged Sep11 prices/ATR in3000;3 initial stopE-2ATR, targetE*(1+fiveyear60gainP90),1–3months. Reference QQQ E714.88 stop699.14315 target830.47313 P90+16.169585% RR7.3454; IWM288.89/282.74895/325.93680 P90+12.823845% RR6.0326; EFA106.70/104.93826/116.97551 P90+9.630283% RR5.8326. SMA distances8.4373/5.7565/5.9863% do not protect initial2ATR losses. Profit-ratchet max(existing,SMA200,high20-2ATR) current references718.84315/299.03895/107.12826 exceed closes: never set these as initial above-price stops. Once actually profitable recalc; price already through threshold means RTHexitreview, not invalid adjustment. Targets are historical return percentiles, not90%hit odds. Recompute actual-entry exits before placement and check fills; do not widen for shortfall. Source https://runway.adityaperswal.workers.dev/data/stock-bars . No exit moved.
QQQ_stop699.14
QQQ_target830.47
IWM_stop282.75
IWM_target325.94
EFA_stop104.94
EFA_target116.98
Three bear chains converge on delayed exits and common capital loss
Premises:1 issuer exposures in this run's fundamental records;2 signal3000 selectsQQQ/IWM/EFA;3 correlations3001 .7068–.7494 and jointstress64.67;4 fractional exits wait RTH. QQQ: 1 expected future cash flows disappoint or discount rates rise;2 index leaders reprice together, reducing diversification;3 fractional liquidation waits RTH and reduces next-run capital. IWM: 1 financing or demand worsens;2 margins and equity valuations fall together;3 correlated gaps can bypass the2ATR stop before regular-session exit. EFA: 1 industrial demand/currency conditions weaken;2 financials and exporters transmit stress across the index;3 global repricing reaches US ETF open before fractional stops can execute. Facts end at exposure/price/rules; all causal chains are conditional scenarios, not claims these shocks occurred today. Today changed cap221.26->289.34 and completedcrypto day, not ETF daily data or complement. Strongest common objection: monthly ranking and200SMA lag fast reversals, while2ATR stop is not gap insurance. Response follows mandate: maintain empirical initial exits and stress-sized equal slots; no headline veto or lower-rankedcrypto substitution. Sources https://www.invesco.com/qqq-etf/en/about.html https://www.ishares.com/us/products/239710/ishares-russell-2000-etf https://www.ishares.com/us/products/239623/ishares-msci-eafe-etf ; https://runway.adityaperswal.workers.dev/data/stock-bars . No position means no Rmultiple, holding age or exit modification exists.
joint_stress64.67
correlation_min0.71
correlation_max0.75
positions0
EFA strongest prospective bear case; no holding to reset
Premises:1 own book has no holdings/trades;2 Issuer identifies developed large/mid caps outside US/Canada; Sep10 financials26.47%,industrials18.66%,Japan24.44%; Sep11 close106.70 matches bars. 3 fixed signal3000 remains eligible. Bear case: A global industrial slowdown plus adverse currency translation can lower USD equity returns while financial-sector losses amplify cyclical exposure. Conditional inference; no current FX or profit shock measured. Counterexample: Foreign equity strength and supportive currency moves could keep returns positive; diverse countries do not mean independent crisis returns. Who benefits: issuers collect asset-based fees; index inclusion does not establish cheap valuation. Source https://www.ishares.com/us/products/239623/ishares-msci-eafe-etf. Start flat, goal implement fixed complement, binding constraint RTH fractional execution. No fresh constituent cash-flow/valuation audit or verified Saturday event change; supplied news is unverified context and does not override rule. Cash-flow mechanism supports stress sizing, not a discretionary entry veto.
positions0
IWM strongest prospective bear case; no holding to reset
Premises:1 own book has no holdings/trades;2 Issuer identifies broad US small-cap Russell2000 exposure; Sep11 close288.89 matches bar feed. 3 fixed signal3000 remains eligible. Bear case: A credit tightening/recession shock can simultaneously weaken smaller-company revenues and raise refinancing expense. Debt sensitivity is a scenario inference, not an audited current aggregate leverage estimate. Counterexample: Easier financing and resilient domestic demand could sustain the trend; no verified new shock establishes failure. Who benefits: issuers collect asset-based fees; index inclusion does not establish cheap valuation. Source https://www.ishares.com/us/products/239710/ishares-russell-2000-etf. Start flat, goal implement fixed complement, binding constraint RTH fractional execution. No fresh constituent cash-flow/valuation audit or verified Saturday event change; supplied news is unverified context and does not override rule. Cash-flow mechanism supports stress sizing, not a discretionary entry veto.