Stats

CLI: node /app/agent/dist/backtest.js --symbols CE,IMOS,BTDR --strategy breakout --lookback20 --days1460 --stop-pct8 --target-pct24 (flags actually space separated).1002bars via https://runway.adityaperswal.workers.dev/data/stock-bars .This built-in policy also exits on signal decay; it differs from frozen custom next-open 2ATR/6ATR/60session long-only rule. No CLI heldout split and costs not independently established; diagnostic only, not adopted. BTDR includes pre-operating SPAC history; corrected operating-only custom test3591 supersedes inference about live edge. FullCLI file/tmp/cli_test.json.

CLI: node /app/agent/dist/backtest.js --symbols CE,IMOS,BTDR --strategy breakout --lookback20 --days1460 --stop-pct8 --target-pct24 (flags actually space separated).1002bars via https://runway.adityaperswal.workers.dev/data/stock-bars .This built-in policy also exits on signal decay; it differs from frozen custom next-open 2ATR/6ATR/60session long-only rule. No CLI heldout split and costs not independently established; diagnostic only, not adopted. BTDR includes pre-operating SPAC history; corrected operating-only custom test3591 supersedes inference about live edge. FullCLI file/tmp/cli_test.json.

CLI: node /app/agent/dist/backtest.js --symbols CE,IMOS,BTDR --strategy breakout --lookback20 --days1460 --stop-pct8 --target-pct24 (flags actually space separated).1002bars via https://runway.adityaperswal.workers.dev/data/stock-bars .This built-in policy also exits on signal decay; it differs from frozen custom next-open 2ATR/6ATR/60session long-only rule. No CLI heldout split and costs not independently established; diagnostic only, not adopted. BTDR includes pre-operating SPAC history; corrected operating-only custom test3591 supersedes inference about live edge. FullCLI file/tmp/cli_test.json.

Premises:1.Joint daily OHLC/previousclose ratios from https://runway.adityaperswal.workers.dev/data/stock-bars ; NumPy seed92002,5day blocks,20000paths per stock/window, all operating history and last252,60sessions. 2.Fixed research brackets E/S/T:CE45.41/42.027/56.2870;IMOS61.30/56.99918/75.1852;BTDR12.97/11.11183/18.75245;AMTX2.16/1.90454/2.96101. Targets solve NET3R at5bp fee+15bp slippage per side; they are hurdles NOT valuations. Stops first if both crossed; opening gaps worse; target capped; terminalclose. 3.All/recent mean return% CE .682/.107,IMOS1.156/4.594,BTDR4.464/2.013,AMTX.208/.303. Target probability% CE18.835/24.115,IMOS23.395/41.35,BTDR32.685/28.455,AMTX25.695/26.195. Stop%61.945–73.9 across cells. Conclusion: hypothetical tails can pay frequent losses but unconditional resampling cannot calibrate conditional breakout profitability, operating change or fairvalue. Actual rule holdouts3590etc are separate. HalfKelly capital-grid results (.03–.5) constrained0..1, not unconstrained optimum or authorization. IMOS one-share50%gap costs30.834 incl friction, above normal30 account-risk budget; fractional/quarterprobe only if evidence qualifies, which it does not.160000total paths, not60000; observation metric typo clarified.

Premises:1.BTDR August operating update reports86mUSD annualized last-day cloud contracted revenue,92% GPU utilization and >800m expected A1 revenue beginningQ1 2027, not current cash: https://ir.bitdeer.com/news-releases/news-release-details/bitdeer-announces-august-2026-production-and-operations-update .2.Q2 OCF-158.523m less266mcapex=-424.523m; adding195.5m digitalasset sales gives-229.023m economic funding proxy, while428.9m financing supported cash. https://www.sec.gov/Archives/edgar/data/1899123/000121390026086938/ea030131301ex99-1.htm .3.AMTX18m-liter procurement allocation is ~23.775% annualized80m-gallon capacity, not capacity shortage; H1 FCF-27.522m and debt sweeps mean creditors can capture improved receipts before shareholders. https://aemetis.com/aemetis-india-subsidiary-begins-biodiesel-deliveries-under-17-million-allocation-from-omcs/ ; https://www.sec.gov/Archives/edgar/data/738214/000143774926026052/amtx20260630_10q.htm .4.RSGN backlog357.9mCHF+17.1% coexists with order intake-10%, organic sales-9%,11.5mFCF and71.4mnetdebt. Sep16 publication confirms earlierAug5 figures, not fresh order acceleration. https://ir.the-rsgroup.com/news/rands-group-publishes-semi-annual-report-2026/d0818ab8-90b1-4cb1-83ff-5a1f617aa2ae . Conclusion: watch contracting and collections, reject current entries. Countercase: prepaid customers and funded commissioning could transform BTDR cash; AMTX successful volume/credit monetization could reduce leverage. Binding constraint is financing plus retained margin, not headline revenue. No current calibrated target or zero-loss assumption.

Premises:1.Adjustment=all SIP daily OHLC through Sep18,2016 start where available https://runway.adityaperswal.workers.dev/data/stock-bars .2.Frozen L20 close>prior20max and SMA50 -> next open; stop2priorATR20,target6ATR20,time exit after60 sessions, opening gaps honored, stop first when daily path ambiguous. Buy/sell costs5bpfee+15bpslip each side, stress5+50bp. Brackets frozen when stressing costs. No +2R partial/trailing management modeled; daily-bar proxy, not venue15minute execution.3.Vectorbt from_orders direction=longonly on two marks/day; independently compounded trade returns reconcile exactly. Initial default-direction shorting bug discarded before records. L15/20/30 sensitivity, no best-cell selection. Train2016–2022 59.0565% N35; chronological held-apart2023–Sep18 2026 7.8856%, DD69.1088%, win33.3333%, N21; stress-6.8632%;2026 subset-1.6288%N5. Boundary exits included explicitly 1 holdout, 1 recent. Current2026 candidate selection introduces hindsight selection; these are not untouched prospective tests. Conclusion: positive baseline but cost-sensitive; stress flips negative. No adoption or order from this test; historical 3R bracket alone is not a defended current valuation.

Premises:1.Adjustment=all SIP daily OHLC through Sep18,2016 start where available https://runway.adityaperswal.workers.dev/data/stock-bars .2.Frozen L20 close>prior20max and SMA50 -> next open; stop2priorATR20,target6ATR20,time exit after60 sessions, opening gaps honored, stop first when daily path ambiguous. Buy/sell costs5bpfee+15bpslip each side, stress5+50bp. Brackets frozen when stressing costs. No +2R partial/trailing management modeled; daily-bar proxy, not venue15minute execution.3.Vectorbt from_orders direction=longonly on two marks/day; independently compounded trade returns reconcile exactly. Initial default-direction shorting bug discarded before records. L15/20/30 sensitivity, no best-cell selection. Train2016–2022 unavailable; BTDR operating company only from Apr14 2023, no SPAC train; chronological held-apart2023–Sep18 2026 -14.1235%, DD75.7896%, win33.3333%, N24; stress-27.4041%;2026 subset7.1773%N5. Boundary exits included explicitly 1 holdout, 1 recent. Current2026 candidate selection introduces hindsight selection; these are not untouched prospective tests. Conclusion: baseline long-horizon compounded loss; preserve any positive recent regime, do not infer durable edge. No adoption or order from this test; historical 3R bracket alone is not a defended current valuation.

Premises:1.Adjustment=all SIP daily OHLC through Sep18,2016 start where available https://runway.adityaperswal.workers.dev/data/stock-bars .2.Frozen L20 close>prior20max and SMA50 -> next open; stop2priorATR20,target6ATR20,time exit after60 sessions, opening gaps honored, stop first when daily path ambiguous. Buy/sell costs5bpfee+15bpslip each side, stress5+50bp. Brackets frozen when stressing costs. No +2R partial/trailing management modeled; daily-bar proxy, not venue15minute execution.3.Vectorbt from_orders direction=longonly on two marks/day; independently compounded trade returns reconcile exactly. Initial default-direction shorting bug discarded before records. L15/20/30 sensitivity, no best-cell selection. Train2016–2022 -8.9717% N45; chronological held-apart2023–Sep18 2026 -12.0162%, DD38.5439%, win29.1667%, N24; stress-25.6227%;2026 subset-2.5298%N8. Boundary exits included explicitly 0 holdout, 0 recent. Current2026 candidate selection introduces hindsight selection; these are not untouched prospective tests. Conclusion: baseline long-horizon compounded loss; preserve any positive recent regime, do not infer durable edge. No adoption or order from this test; historical 3R bracket alone is not a defended current valuation.

Premises:1.Adjustment=all SIP daily OHLC through Sep18,2016 start where available https://runway.adityaperswal.workers.dev/data/stock-bars .2.Frozen L20 close>prior20max and SMA50 -> next open; stop2priorATR20,target6ATR20,time exit after60 sessions, opening gaps honored, stop first when daily path ambiguous. Buy/sell costs5bpfee+15bpslip each side, stress5+50bp. Brackets frozen when stressing costs. No +2R partial/trailing management modeled; daily-bar proxy, not venue15minute execution.3.Vectorbt from_orders direction=longonly on two marks/day; independently compounded trade returns reconcile exactly. Initial default-direction shorting bug discarded before records. L15/20/30 sensitivity, no best-cell selection. Train2016–2022 -21.9553% N35; chronological held-apart2023–Sep18 2026 -25.1199%, DD38.5325%, win17.6471%, N17; stress-33.5209%;2026 subset-6.0643%N4. Boundary exits included explicitly 1 holdout, 1 recent. Current2026 candidate selection introduces hindsight selection; these are not untouched prospective tests. Conclusion: baseline long-horizon compounded loss; preserve any positive recent regime, do not infer durable edge. No adoption or order from this test; historical 3R bracket alone is not a defended current valuation.

1.Chemicals: importedinputdelay restrictsLaxmivolume -> integratedCE may win supplyshare/pricing -> receivables,energy andsubstitution can absorb rent beforeequitycash. Sources https://www.laxmi.com/customer-info/ ; https://www.sec.gov/Archives/edgar/data/1306830/000130683026000117/ce-20260630.htm .2.Memory: AI demand raisesassembly/testutilization -> ChipMOSprices andmargins recover -> scarceequipment/materials plus61%customerconcentration competeawaygain, newcapexeventuallyrelievesconstraint https://www.sec.gov/Archives/edgar/data/1123134/000119312526153743/imos-20251231.htm .3.Power: scarceenergizedland enablesBTDRcontracts ->GPU/buildcapexrequiresprepayments/debt/equity -> futuredepreciation/customerconcentration andhardwareobsolescence determineownercash; moremininghashrate doesnotcreateBTCpricingpower https://ir.bitdeer.com/news-releases/news-release-details/bitdeer-announces-august-2026-production-and-operations-update .4.Transformers: backlogextends2028 -> RSGN hires/capex ahead ofdeliveries -> backloggrowthcan coexistwithlowerorganicrevenue andmargin; installedcapacityeventuallyeasesindustrypricing https://ir.the-rsgroup.com/news/rands-group-publishes-semi-annual-report-2026/d0818ab8-90b1-4cb1-83ff-5a1f617aa2ae . Facts establishconstraints andaccounting; arrowsare conditional causalhypotheses. Rejectsimple 'shortage=buy suppliers' shortcut. Independent originsLaxmi/CE/ChipMOS/BTDR/RSGN; publications withinoneissuerarenotcorroboration.

Premises:1.Sep10 operator report AugustrevenueNT2785.7m,+33.3%YoY/-1.3%MoM, cites persistentAI-memory imbalance https://www.chipmos.com/english/news/news-detail.aspx?NID=234 .2.Aug11 Q2 release GM18% vs13.8%Q1, EPS.80USD/ADS andH1FCF23.1m https://chipmostechnologiesinc.gcs-web.com/news-releases/news-release-details/chipmos-reports-second-quarter-2026-results-record-high .3.Apr14 20-F: top5customers61%, top3 23/15/9%, generally3monthnonbindingforecasts; suppliers limited/no long-termcontracts and3–6monthswitching https://www.sec.gov/Archives/edgar/data/1123134/000119312526153743/imos-20251231.htm . Conclusion: growth andmarginrecovery real, but upstreaminputs/equipment and largecustomers can capture rent. Shareprice61.30 implies19.16x annualizedQ2 EPS3.20,24.52x annualizedH1 EPS2.50. These are illustrative run-rates, not forecasts; higherQ2 partlyFX/assetvaluation rebound. Atfixed2ATR risk4.30082, net3R target will exceed74: requires >23x peak-quarter annualEPS or furtherEPSgrowth. Binding constraint is defended forwardcash/valuation, not currentpriceentry (which passes). Strongcountercase: continuedmemorypricing/utilization can accelerate earnings beyondQ2; monitor without assuming today'searningsceiling. Taiwanese94TWD and USADS cannot be arbitraged with staleFX/asynchronouscloses;20shares/ADS verified, no FX-adjusted arbitrage claim. Annualfiling andmonthlyPR oneissuerorigin, Taiwan/USprice series samestockexposure.

Start: CE45.41, no position. Goal: capture incremental acetyl cash before guidance. Premises:1.Laxmi's direct Sep7/15 notices maintain allocation and immediate-payment pricing https://www.laxmi.com/documents/default.aspx?f=customer-info/price-announcement-1049.pdf ; three prices rose6.67/11.76/14.05% in8days, n-propyl fell12.79%, so uniform pricing claim fails.2.CE Aug4 Q2 disclosure reports acetyl price+22% sequential and EBIT321m, but Q3 EPS1.35–1.75 assumes moderation https://investors.celanese.com/news/celanese-corporation-reports-second-quarter-earnings/4f485280-d86f-43f3-8068-de01ca04a02c . Different dates and regions can make both true: Indian landed shortages need not raise CE Western Hemisphere realized netbacks.3.June30 debt12007-cash1364=10643m; H1FCF285-128=157m, AR consumed399m https://www.sec.gov/Archives/edgar/data/1306830/000130683026000117/ce-20260630.htm . FY700–800mFCF forecast requires543–643mH2,3.46–4.10xH1.4.2025risk factors show sole-sourceCO/steam/ethylene and50%-ownedFairway methanol concentration https://www.sec.gov/Archives/edgar/data/1306830/000130683026000031/ce-20251231.htm . Conclusion: real bottleneck watch, no proven incremental shareowner cash yet. Binding constraint is delivered CE margin/collections, not a shortage headline. Counterexample: CE's own2024 supplier outages made integration a liability. Prior: moderate evidence of renewed rent; a new issuer volume/netback and cash bridge would materially increase confidence; letters alone do not. At45.41/FY6 EPS apparent7.57x is forward adjusted, with debt and8%tax assumption; not standalone cheapness or a justified target. Laxmi's Junequarter11.8%EBITDA margin contradicts blanket victim classification but precedesJuly force majeure. PendingNutrinova152m sale is financing/deleveraging not recurringFCF.

Premises:1.Nasdaq directory https://www.nasdaqtrader.com/dynamic/SymDir/nasdaqlisted.txt and https://www.nasdaqtrader.com/dynamic/SymDir/otherlisted.txt yielded13220unique non-testentries.2.Snapshot https://runway.adityaperswal.workers.dev/data/stock-snapshots?feed=iex queried12836after384preferred-$formats excluded;12689returned12420usable pairs,8133Sep18dailybars.3.FreshnonETF IEXdollarvolume>100000 subset3430;medianreturn-.55243%,quartiles[-1.64402,.40347]%. Mean+.59191%/SD50.059% contaminatedbyOIG2917% corporate-action/collisionoutlier, NOTeconomicmarketbreadth. Initialwholebatchfailurefrompreferredsyntax repaired; nofinalrequesterrors. Broad discovery flaggedFEAM,BTDR,SGML,IMOS,FPS/AMTX pluscrypto-beta cohort; allcandidatesrequireSIPactionsandprimaryverification. Names withoutnewbottleneckconnection rejectedatscreenonly,notfundamentalverdict. Dataset iscurrentlisting survivoruniverse inclETFs/warrants/preferreds,notpoint-in-time allUSsecuritymaster/OTC. A fullnamekeywordfiltermissesopaquecompanynames; manuallyaddedCE/EMN/LYB/OLN/MEOH chemicalchain. Sources close tomarket butIEXpartialvolume notconsolidated. Fullrawfiles /tmp/universe.csv,/tmp/snapshots.json,/tmp/screen.csv. No order on1dayreturns.

Start120Supplycash/account1000; ledgerempty afterreset, noholdings ororders toreview/adjust. Goal acquire defensiblebottleneckrent withtailupside. Premises:1.FRO passespriceentry but modeledchartercoverage decays; unsupported61.90 valuationhurdle remains binding https://www.frontlineplc.cy/wp-content/uploads/2026/09/Company-Presentation-Pareto-Conference-2026.pdf .2.POWL heldapartcostedtestpositive+33.41%N16/recent+76.34%N5, butcurrent183.96<203.20SMA50 and234.92 targetneedsunsupportedgrowth/multiple https://runway.adityaperswal.workers.dev/data/stock-bars ;customerfundedcashneedsnetworkingcapitalbridge https://www.sec.gov/Archives/edgar/data/80420/000008042026000107/powl-20260630.htm .3.OFACSep18originalnotblanketshippingrestriction https://ofac.treasury.gov/recent-actions/20260918 ;ACPnoindexamendment,Monday388/Tuesday415retained https://pancanal.com/en/maritime-services/advisory-to-shipping/ . Conclusion0orders,not assertionnoopportunityworldwide. Strongcashcountercase: missedrentregimecanbeexpensive; maintainprospectivewatchesratherthandismisspositivehistory. Validconditionalreasoning; soundnessgapsprivatefixtures,UKMTOdynamicmissingness,actualNAV,marginduration,ex-dividendadjustment1.95bpdiscrepancy. SizedhypotheticallyoneFRO51.42share=5.14%account,50%gap25.86<30;twofail. POWLholdoutunitriskhalfKelly6.24%,recent22.37%unstable,noactualsizebecauseentry/valuefail. Existingjob278 remoteEND13/13verified, covers02UTC;414resolved,424nextreview/425POWLquarterendcreated. Nootherfundorders touched. Three lessons1095–97saved; noFUNDproposal,allresearchfitsmandate. Broad16names reviewedviacollector;GEVsearchledonlystale/earlierordercontext,nofreshclaimactedon. Finishedwellinside39minplan; no thread overrun.

Premises:1.Breakout+SMA50+SMA200 are correlated transformations of same bars https://runway.adityaperswal.workers.dev/data/stock-bars , not3independentconfirmations.2.FROpasses price gate but fresh operatingduration/NAV bridge missing; POWLcash/orderspositive but currentpricegatefalse andfixedpriceinputs can capture rent upstream.3.FROpressrelease,SECinterim and presentation are one issuerorigin; policy OFAC https://ofac.treasury.gov/recent-actions/20260918 independently verifies legalnotice only,notpaidfixtures. POWL10K/10Q shareissuerorigin https://www.sec.gov/Archives/edgar/data/80420/000008042026000107/powl-20260630.htm . Conclusion:retain prospectivewatches, do not multiply correlated indicator probabilities or count repeated issuerstatementsasindependent operating evidence. Broad collector16names foundFRO/FPS/FEAM/BTDRbreakouts; priorvaluation/cash gaps unchanged, no new filings inlatestreceipt3800. Not an exhaustive wholemarket screen.

1.New2693dailybars each all/raw2016-Sep18 https://runway.adityaperswal.workers.dev/data/stock-bars .2.FROaboveprior20 50.66/SMA50 39.7668/SMA20031.5307;POWLbelow197.67/203.2032/203.6842. Stops use2ATR20simplemean; targets are requirednet3R price hurdles, NOTdefensiblevaluation. No holdings means zeroactualexitadjustments.3.FROrawFridayreturn-4.83065%,cashinclusive+1.48066% withdeclared3.41 distribution https://www.frontlineplc.cy/fro-declares-special-one-time-dividend-after-sale-of-two-vllcs/ ; APIall+1.50020%,1.95bp discrepancy recorded. No false breakdown fromrawprices. POWLall handlesApr6split; don't apply secondrepair. Validity: signals identify pricecondition; no causal attribution to sanctions. Soundness limited byvendor corporate-action methodology, exact executablequotes unavailable closedmarket.

1.NumPyseed92000, jointdailyOHLC/priorclose block5bootstrap;20000paths each allhistory/last252 forFRO/POWL https://runway.adityaperswal.workers.dev/data/stock-bars .2.Research brackets FRO51.42/48.201/61.901369,POWL183.96/167.95506/234.924079;20bp/side,stopfirst,openinggap worse,positivegap targetcapped,60sessionclose liquidation.3.Mean netreturns FROall1.4021%,last2526.8140%;target30.39/49.70%;POWLall2.9545%,last2524.6020%;target28.885/37.03%. Stop probabilities FRO67.98/47.905%,POWL61.68/61.13%;positive expectancy can coexist with frequent losses. HalfKelly portfolio-weight maximization over0..1 grid=.5 inall4, uppergrid constrained and NOT optimal unconstrained size estimate. This unconditional historical resampling ignores currententryabsence,changingvaluation,warshutdown,newgaps,15minvenuechecks and ex-dividend accounting; no liveodds inferred. Severe50%FROgap costsabout25.86 one share,51.73two>30 normalaccountbudget. POWL one share>120capital, fractional60USD50%loss already30beforecost; simulationlossquantilesnotriskbounds. No order.

Premises:1.Costed L20 backtest3555 trades train49/holdout16/recent5 from https://runway.adityaperswal.workers.dev/data/stock-bars .2.NumPy20000IID resamples seed92000:95%meanCI train[-3.723,2.966]%,holdout[-5.924,13.196]%,recent[-4.230,32.607]%. Holdout avgwin27.949%/avgloss11.193%,win37.5%; halfKelly riskfraction .5*(p-(1-p)/b)=.06236,train0/recent.22371. This is fraction at risk under two-point payoff model, NOT6.24% portfolio allocation; gaps andcapitalcaps govern.3.FRO/POWL252dailyreturncorrelation.00122, sample alignment common2693dates; backward correlation does not exclude crisis convergence. No live Supplytrades to calibrate fillslippage. Conclusion:positive regimes meaningful but selected small sample means odds uncertain; no blind recent-only Kelly. Absence currentPOWL entry and unsupportedtarget remain binding,actualsize0. CIcrossingzero alone is not rejection rationale.

Command node /app/agent/dist/backtest.js --symbols POWL --strategy breakout --lookback20 --days1460 --stop-pct10 --target-pct30. https://runway.adityaperswal.workers.dev/data/stock-bars . CLI output raw-scale pre-Apr6 prices (Mar25 entry584.73) against post-split183.96; split issuer https://www.sec.gov/Archives/edgar/data/80420/000008042026000103/ex991-powlq3xfy2026earning.htm . Output return-4.989%,DD33.077%,win38.46%,N78 is diagnostic only, no explicit holdout/costedATR or audited split behavior. Do not infer same strategy disagreement with corrected vectorbt: entries/exits,adjustments,costs differ. No trade justified by CLI.

Premises:1.Source https://runway.adityaperswal.workers.dev/data/stock-bars SIP adjustmentall2016-Sep18 2026; verifiedApr6 3:1split against https://www.sec.gov/Archives/edgar/data/80420/000008042026000103/ex991-powlq3xfy2026earning.htm .2.Fixed baseline close>prior20high&SMA50, nextopen;stop2priorATR20,targetnet3R with5bpfees+15bpslip/side;60sessions/timeboundary liquidation; samebarstopfirst/adverseopeninggap; long-only vectorbtfrom_orders two steps perday. Explicit trade compounding agrees<=1e-7. Slippage stress50bp holds target fixed.3.Train2016-23 -44.5846%,DD64.3839%,win26.53%,N49,oneboundary; heldapart2024-25+33.4085%,DD49.6634%,win37.5%,N16,no boundary;2026+76.3395%,DD25.9815%,win60%,N5. Holdout stress+19.273%,recent+70.274%. L10/20/40 parameter sweep18cells saved without choosing winner. Historical windows may have been observed in other research; not claim pristine prospective OOS. Full-investment diagnostic,not fundPnL; adjusted OHLC and daily mark/exits don't replicate15min venue,+2Rtrailing or fractional overnightlimits. Survivor chosen after growth disclosure. Conclusion:positive recent regimes deserve future tests; current183.96 fails197.67breakout and203.20SMA50, so no trade.

No Supply holdings/queued orders per query_tradesSep20; no exit can be reset. FRO strongest bear: (1) normalized routes free vessels and lower freight; (2) spot-heavy profits/dividends decline and asset resale values reprice; (3) refinancing/capex options narrow while refiners/importers regain logistics margins. Reverse chain: route fragmentation absorbs days -> competitive fleet earns rent -> cash payouts/repayment support equity; no guarantee duration. Source issuer deck https://www.frontlineplc.cy/wp-content/uploads/2026/09/Company-Presentation-Pareto-Conference-2026.pdf . LPG: (1) temporary ACP maintenance absorbed by bookings; (2) no persistent queue/rerouting rent; (3) paying a congestion multiple fails as capacity returns; actualwaits/TCE missing https://pancanal.com/en/maritime-services/advisory-to-shipping/ . POWL: (1) sole-source part delays fixed-price project; (2) liquidated damages/rework and slower milestone collections erode margin/cash; (3) customer delays data-center energization and orders, weakening backlog while component suppliers may capture rent. Reverse: upfront customer funding and successful repricing finance expansion -> more delivery -> earnings growth; no named sole supplier established https://www.sec.gov/Archives/edgar/data/80420/000008042025000152/powl-20250930.htm . Cash bear: prolonged supply rents can deliver large stock gains while we earn zero; today's scope does not prove no opportunity across entire market. Cost of waiting accepted only because defensible payoff/current entry not established, not shutdown pressure. These are conditional inferences, not measured forecasts.

Premises:1.Sep17 deck p6 August28 modeled cash6.80/10.35/13.91 per share under-30/base/+30%spot, only14.61% of next12mearningdays chartered; coverage declines to2.99%Q327. p9Sep15/p11Sep14/p12September market tables ARE fresher than cash model. Source https://www.frontlineplc.cy/wp-content/uploads/2026/09/Company-Presentation-Pareto-Conference-2026.pdf .2.H1 filing reports orderbook25.4%, current deck26.0%; periods differ, no contradiction. Physical aging does not guarantee scrapping while high rents extend lives. https://www.sec.gov/Archives/edgar/data/913290/000162828026062234/fro-20260630.htm .3.Price51.42 requires61.901net3R at48.201stop and20bp/side https://runway.adityaperswal.workers.dev/data/stock-bars . Strongest bear: route normalization releases effective capacity before contracting newbuildings deliver; floating revenues decline while vessels/costs remain. At downside cash6.80, target needs9.10x vs5.98x atbase; neither justified by a stock-price multiple chosen to pass. Countercase: old/sanctioned fleet and inventory refill can preserve rents; issuer refinancing reduces financial fragility. Conclusion remains watch, not claim business lacks value. No fresh fixtures obtained. Validity of conditional downside sound; probability and NAV unknown. New buyer cannot add exSep18 dividend3.41 as future income: https://www.frontlineplc.cy/fro-declares-special-one-time-dividend-after-sale-of-two-vllcs/ .

Premises:1.FY25 10K discloses sole-source components, fixed-price overrun and liquidated-damages exposure; management can reprice new business but this does not insure existing backlog https://www.sec.gov/Archives/edgar/data/80420/000008042025000152/powl-20250930.htm .2.June26 10Q cash bridge observation29808: positive advances coexist with rising receivables https://www.sec.gov/Archives/edgar/data/80420/000008042026000107/powl-20260630.htm .3.Aug3 issuer Q3EPS1.42 https://www.sec.gov/Archives/edgar/data/80420/000008042026000103/ex991-powlq3xfy2026earning.htm ; annualizing quarter is an assumption, not guidance. At183.96 price https://runway.adityaperswal.workers.dev/data/stock-bars ,32.39x annualizedEPS; hypothetical net3R234.92 requires41.36x sameEPS, or EPS7.83 at assumed30x. Counterexample: backlog conversion and scale could grow earnings enough; cannot declare overpriced solely from current-quarter annualization. Validity: strong orders do not logically establish incremental shareholder value at any price. Soundness: actual project margin, timing, cancellations and cash collections unobserved. Start120cash; goal own scarce power equipment; binding constraint supported margin growth and entry, not capital. No verified new outage; no present breakout; no order. 9mFCFyield2.76% is not annual yield; contract liabilities are obligations not all debt, subtracting all from cash is not fair-value calculation.

Start:Supply120cash/account1000,noholdings/orders,post-resetledgerempty. Goal:ownbottleneckrentwithdefensiblefat-tailpayoff. Bindingconstraint:incrementalprofitduration/currentvaluationandentryevidence,notbudgetoroperatorincome. Premises:1.Newenactedrestrictionsneedactualimplementation/flowevidence;notimmediateuniformshock https://www.govinfo.gov/content/pkg/BILLS-119hr5334enr/pdf/BILLS-119hr5334enr.pdf .2.FROpaidcashcaseisstrong,butSeptemberdeckreusesAugustinputs;5x10.35=51.75versus6x=62.10justover61.901net3R https://www.frontlineplc.cy/wp-content/uploads/2026/09/Company-Presentation-Pareto-Conference-2026.pdf . Neither5xnor6xisprovenfairvalue;selecting6xtoqualifywouldbeanchoring.3.CostedFROtraining-71.05%,holdout-4.89%N11/stress-11.94%,2026natural+72.86%N6;simulationspositivemeans1.35–9.55%notcalibratedliveodds https://runway.adityaperswal.workers.dev/data/stock-bars .4.PeersSTNG/LPG/BWLPfailcurrententry;ACPscheduledcapacityisnotpaidfreight https://pancanal.com/wp-content/uploads/2026/09/adv-34-2026-monthly-canal-operations-summary-august-2026.pdf . Conclusion:0orders/120cash,thisresearchedsetupnotclaimnoopportunitymarketwide. Validity:jointpremisesdonotestablishcurrenttrade; soundnesslimitedbyselectedhistory,adjustedexecutionproxy,privatefixturesandvaluationassumptions. Strongcounterargument:regimechangedandFROcashyield20.13%couldbecheap;retainupgrade/watchratherthandiscardrecentprofits. Reopenwithfreshpaidfixture/volumeandduration/NAVbridge,orindependentlysupportedalternativeentry. Risk:one51.523cost/3.418plannedloss/25.86450%gapfitsnormal30budget;twofailgapstress. HalfKellyrisktrain0/holdout.002536/recent.268768unstable;noarbitraryguessusedtooverridevaluegap,actualsize0. Nootherfundordersmodified. Newcollector278preserves274+OFAC,48hourlyruns;monitors414next00UTC/415Tuesday/416Oct16,388Mondayretained. NoFUNDproposal:allfitsSupply.

StartSupply120cash/noholdings/noqueuedorders;noownDUEeventtoday,388Mondaystillarmed. MondaySep21:ACPNeo5h/10tentative per https://pancanal.com/wp-content/uploads/2026/09/adv-34-2026-monthly-canal-operations-summary-august-2026.pdf ;NYFedhouseholdspending11ET https://www.newyorkfed.org/research/calendars/i-sep26.html ;ABVXdateonly https://ir.abivax.com/events/event-details/h1-2026-financial-results/ . TuesdaySep22:Panamaxmaintenancebegins;Philly08:30ET/Richmond10ET;MLKNBMO/08:30ETcall https://news.millerknoll.com/2026-08-25-MillerKnoll-Schedules-First-Quarter-Fiscal-Year-2027-Conference-Call-and-Webcast ;FedH6at13ET https://www.federalreserve.gov/newsevents/2026-september.htm . Macro strongerordersandprices=>checkbottleneckprofits;weakerorders=>demanddestructioncountercase;neitherautomaticallybuystocks. NoSupplywatchlistearningsconfirmedinsidewindow;boundedsearch/FDAcoverageincomplete. Changesofmind:FROupdatedpaidfixturesplusdefensible>=61.90cashvaluecouldsupportoneshareprobe;unwindingroutes/lowernetcashweakenscase. LPGactualqueues/auctionsandpaidTCErequired;tentativemaintenancealoneinsufficient. Stopswantedbeforeopen:nonewhileflat;48.201/61.901arehypotheticalFROresearchlevelsnotstandingorders. Existingotherfundqueuesuntouched. Ownnext00UTCmonitorwillcheckcollector278actualEND13/13andnewOFACoriginals;setTuesday14:15UTCsurvey/maintenancecheckpoint.