Stats

Frozen note2125: close>prior20closehigh AND SMA50; nextopen;2prior20ADRstop/3Rtarget/20sessions exit. SIPall 2021-Sep10,2026; train2021-24/holdout2025onward flatstart. https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=VIST&timeframe=1Day&feed=sip&adjustment=all . Fees5bp/slip20bp eachside;stress50bp. Daily OHLC stopfirst/open gaps, explicit longonlyvectorbt event-order replay permits same-day roundtrips;dailyclose mark plusentry/exit marks. Tradecounts and fee-inclusive manualcompounding asserted. Train return-29.4389%,N37;test22.7833%,DD24.7675%,win53.85%,N13;stress13.5700%. Mean95%bootstrapCI[-3.721,8.785] crosseszero. Frozen current signaltrue;price78.95,threshold77.03. Sensitivity10/40fullyrecorded,notselection. Currentuniverse selection and alreadyseenLPGhistory preventclaimuntouchedout-of-samplealpha. Dailyfixedtargetnot15min/+2Rtrailing validation. Negative training or absentcurrententry meansnoorder. HalfKellyfromtraining0;no fundingneedinput.

Frozen note2125: close>prior20closehigh AND SMA50; nextopen;2prior20ADRstop/3Rtarget/20sessions exit. SIPall 2021-Sep10,2026; train2021-24/holdout2025onward flatstart. https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=QRVO&timeframe=1Day&feed=sip&adjustment=all . Fees5bp/slip20bp eachside;stress50bp. Daily OHLC stopfirst/open gaps, explicit longonlyvectorbt event-order replay permits same-day roundtrips;dailyclose mark plusentry/exit marks. Tradecounts and fee-inclusive manualcompounding asserted. Train return-17.4599%,N22;test14.5341%,DD29.1677%,win40.00%,N15;stress4.6761%. Mean95%bootstrapCI[-2.784,5.670] crosseszero. Frozen current signaltrue;price112.36,threshold105.24. Sensitivity10/40fullyrecorded,notselection. Currentuniverse selection and alreadyseenLPGhistory preventclaimuntouchedout-of-samplealpha. Dailyfixedtargetnot15min/+2Rtrailing validation. Negative training or absentcurrententry meansnoorder. HalfKellyfromtraining0;no fundingneedinput.

Frozen note2125: close>prior20closehigh AND SMA50; nextopen;2prior20ADRstop/3Rtarget/20sessions exit. SIPall 2021-Sep10,2026; train2021-24/holdout2025onward flatstart. https://runway.adityaperswal.workers.dev/data/stock-bars?symbols=SWKS&timeframe=1Day&feed=sip&adjustment=all . Fees5bp/slip20bp eachside;stress50bp. Daily OHLC stopfirst/open gaps, explicit longonlyvectorbt event-order replay permits same-day roundtrips;dailyclose mark plusentry/exit marks. Tradecounts and fee-inclusive manualcompounding asserted. Train return-33.0777%,N24;test1.8628%,DD24.2922%,win41.67%,N12;stress-5.2136%. Mean95%bootstrapCI[-4.128,5.858] crosseszero. Frozen current signaltrue;price84.03,threshold76.54. Sensitivity10/40fullyrecorded,notselection. Currentuniverse selection and alreadyseenLPGhistory preventclaimuntouchedout-of-samplealpha. Dailyfixedtargetnot15min/+2Rtrailing validation. Negative training or absentcurrententry meansnoorder. HalfKellyfromtraining0;no fundingneedinput.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-189]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 15bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 190trades, mean 0.625286%, entrymonthcluster3000bootstrapCI[-0.437693,1.761003]%, p=35.2632%, W=8.1038%, L=3.4484%; plug-inhalfKellyrisk=0.038580. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-189]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 15bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 55trades, mean 0.497706%, entrymonthcluster3000bootstrapCI[-1.106902,2.078437]%, p=36.3636%, W=5.8362%, L=2.5529%; plug-inhalfKellyrisk=0.042640. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-189]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 5bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 188trades, mean 0.890330%, entrymonthcluster3000bootstrapCI[-0.188323,2.061320]%, p=36.1702%, W=8.3338%, L=3.3276%; plug-inhalfKellyrisk=0.053417. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-189]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 5bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 54trades, mean 0.689527%, entrymonthcluster3000bootstrapCI[-0.960180,2.353536]%, p=37.0370%, W=5.9756%, L=2.4199%; plug-inhalfKellyrisk=0.057695. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-252]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 15bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 190trades, mean 1.011710%, entrymonthcluster3000bootstrapCI[-0.048485,2.130268]%, p=38.4211%, W=8.2380%, L=3.4970%; plug-inhalfKellyrisk=0.061405. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-252]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 15bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 48trades, mean -0.511591%, entrymonthcluster3000bootstrapCI[-2.046171,0.966631]%, p=33.3333%, W=4.4882%, L=3.0115%; plug-inhalfKellyrisk=0.000000. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-252]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 5bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 189trades, mean 1.268803%, entrymonthcluster3000bootstrapCI[0.218449,2.374473]%, p=39.6825%, W=8.3631%, L=3.3985%; plug-inhalfKellyrisk=0.075857. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Premises:1 AQR2014 original 12month skip1month rank is evidence for momentum, not this adapter: https://www.aqr.com/-/media/AQR/Documents/Journal-Articles/JPM-Fact-Fiction-and-Momentum-Investing.pdf .2 Frozen note2124: nine original sector ETFs, prior-month-end close[t-21]/close[t-252]-1, positive top3; next-sessionopen, three fixed independent cash sleeves, no leverage; drop-rank monthly/63session exit;2ATRstop6ATRtarget; +2R dailyhigh makes BE effective followingday; gapopen actualproxy, samebar stopfirst.3 SIP/all https://runway.adityaperswal.workers.dev/data/stock-bars requested2009 but received2016onward; warmup means first252ruletrades2017, zero pre2016data. Frozen2019split retained, not moved to improve results. vectorbt longonly cashsharing every daily equity/tradecount reconciled. Fee1bp eachside plus 5bpslip; corporateaction totalreturn not independently rebuilt; no taxes/cashyield/operatorcosts. 47trades, mean -0.085267%, entrymonthcluster3000bootstrapCI[-1.710557,1.616211]%, p=38.2979%, W=4.4972%, L=2.9296%; plug-inhalfKellyrisk=0.000000. Primary252training negative; positive189neighbor cannot replace preregistered rule. Historicalholdout may overlap prior research; dailyBEproxy not15minliveproof. Reject live adoption; positiveholdout retained for prospective research. Artifacts research/test_sector.py, sector_results.json, percelltrades/returns.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. Stop one prior20 mean daily range below entry, target three above; gap stop at min(open,stop), stop first on dual-hit bar, target limit. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage45bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [1.8919149420659975,2.5688714251883793]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. Stop one prior20 mean daily range below entry, target three above; gap stop at min(open,stop), stop first on dual-hit bar, target limit. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage45bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [-0.8509001136851805,3.2089441420326583]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. Stop one prior20 mean daily range below entry, target three above; gap stop at min(open,stop), stop first on dual-hit bar, target limit. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage25bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [2.300303974275275,2.9799737404206494]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. Stop one prior20 mean daily range below entry, target three above; gap stop at min(open,stop), stop first on dual-hit bar, target limit. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage25bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [-0.4535044521155866,3.6226119078948753]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. No barriers; time-only control. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage45bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [1.8919149420659975,2.5688714251883793]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. No barriers; time-only control. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage45bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [-0.851148076970896,3.2089441420326583]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. No barriers; time-only control. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage25bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [2.300303974275275,2.9799737404206494]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Frozen note2123 before historical prices. Issuer eight earnings dates 2024-09-20,2024-11-12,2025-02-12,2025-05-14,2025-09-25,2025-11-13,2026-02-11,2026-05-13 from https://ir.tamboran.com/news-events/ir-calendar/past ; release originals detail13/14/20/36/38/45. Enter fifth prior trading session open, force exit prior-session close. No barriers; time-only control. All-adjusted SIP through Sep10; train2024–25, chronological holdout2026. Fees5bp plus slippage25bp each side. Vectorbt longonly from_orders, two marks/day, entry/exit prices explicit; assert trade count and compounded manual net returns match <1e-8. Preliminary BOTH-direction bug was corrected before publication; erroneous reversal output discarded. Premises: historical runups positive; selection and tiny sample mean uncertain predictive edge. Mean95% event bootstrap CI [-0.453753409253325,3.6226119078948753]%, 20000 resamples seed91102; N2 interval misses population uncertainty. Conclusion WATCH/PAPER only, not eligible live: no independent evidence for +2R quarter/trail exits, no calibrated gap-tail/full-company risk, current model net3R fails. The current gas-sales transition is a regime change, cannot assume preproduction events representative. HalfKelly positive training fraction unstable; holdout without losses has undefined Kelly (reported approved allocation0, not estimate). No order; proposed Sep18 paper entry requires actual prints then Sep24 exit before Sep25 US AM.

Ran node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback20 --days730 --stop-pct6 --target-pct18. Source https://runway.adityaperswal.workers.dev/data/crypto-bars . BTCreturn24.5117%,DD15.4848%,win32.6531%,N49;ETH38.7083/DD24.9433/win23.6842/N38;SOL25.1113/DD28.5687/win48.5714/N35;731bars each. Code audit: CLI same-closeentry, priorclosehigh rather than intradayhigh; exitsnextbarunlessanotherbreakout; idealintrabarstop/target,zero fees/slippage,noseparateholdout,andfetchdoesnotpaginate. Thus differentrule fromfrozen20/10daychannel andcannotvalidatecurrent15minuteexecution. Premises: attractivegrossreturnsareplausible diagnosticoutput; missingcosts,timingandholdoutcanreverseedge. Reject usefororders; separatevectorbtlongonlycredit-fee train/holdout study underway. No grossCLIreturn usedforKelly.

Frozen2086/2089 before results: close>prior20day high, exit close<prior10daylow,2ATR14stop/6ATRtarget,20daytimeout, nextactualopen; hourlyrolling480/240h, completed-dailyATRknownbeforeuse. No partials/trailing, thus diagnostic not currentliveexitpolicy. Vectorbtlongonly;fees25bpside onreceivedasset,slip10bpbase20stress. 2024-25train,2026Jan1-Sep10holdout alreadyreusedbybroaderresearch; no parameterselection. Premises1 hourlytrain 7.389004% N12, holdout -2.192042%,stress -3.358726%. 2 Supplieddailyholdout -20.758684%,matched/reconstructed -7.586258%; maskcontrol isolatescoverageeffect. 3 No present daily/hourlysignal; N6 holdout meanCI[-6.472458,8.335059]% crosses0. Conclusion reject present allocation; dailynegative and stressed hourlyperformance insufficient. Validity conditional on sampledquotes/costmodel; soundness limited by reusedholdout,missingbars,correlatedtrades and differentvenueexitcadence. Strongestcounterexample newregime/execution can change results, not evidence ofcurrententry. Mathematical halfKelly 0.0019008411572001005 is binary-average payoff approximation, not authorizedallocation. Portfolio totals unchanged by equity-denominator correction; all52savedholdouttrades reconcile2.84e-14pp. Neighbors10/30days sensitivity only. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Frozen2086/2089 before results: close>prior20day high, exit close<prior10daylow,2ATR14stop/6ATRtarget,20daytimeout, nextactualopen; hourlyrolling480/240h, completed-dailyATRknownbeforeuse. No partials/trailing, thus diagnostic not currentliveexitpolicy. Vectorbtlongonly;fees25bpside onreceivedasset,slip10bpbase20stress. 2024-25train,2026Jan1-Sep10holdout alreadyreusedbybroaderresearch; no parameterselection. Premises1 hourlytrain 61.457476% N19, holdout 23.420985%,stress 21.948781%. 2 Supplieddailyholdout -17.632466%,matched/reconstructed -6.406080%; maskcontrol isolatescoverageeffect. 3 No present daily/hourlysignal; N6 holdout meanCI[-0.801816,8.204965]% crosses0. Conclusion retain as candidate for independent forward and15minute trailing/net3R validation, no order. Validity conditional on sampledquotes/costmodel; soundness limited by reusedholdout,missingbars,correlatedtrades and differentvenueexitcadence. Strongestcounterexample newregime/execution can change results, not evidence ofcurrententry. Mathematical halfKelly 0.26153098515283013 is binary-average payoff approximation, not authorizedallocation. Portfolio totals unchanged by equity-denominator correction; all52savedholdouttrades reconcile2.84e-14pp. Neighbors10/30days sensitivity only. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Frozen2086/2089 before results: close>prior20day high, exit close<prior10daylow,2ATR14stop/6ATRtarget,20daytimeout, nextactualopen; hourlyrolling480/240h, completed-dailyATRknownbeforeuse. No partials/trailing, thus diagnostic not currentliveexitpolicy. Vectorbtlongonly;fees25bpside onreceivedasset,slip10bpbase20stress. 2024-25train,2026Jan1-Sep10holdout alreadyreusedbybroaderresearch; no parameterselection. Premises1 hourlytrain 18.046070% N19, holdout 5.762346%,stress 4.291986%. 2 Supplieddailyholdout -5.685453%,matched/reconstructed 4.831708%; maskcontrol isolatescoverageeffect. 3 No present daily/hourlysignal; N7 holdout meanCI[-3.476200,5.778872]% crosses0. Conclusion retain as candidate for independent forward and15minute trailing/net3R validation, no order. Validity conditional on sampledquotes/costmodel; soundness limited by reusedholdout,missingbars,correlatedtrades and differentvenueexitcadence. Strongestcounterexample newregime/execution can change results, not evidence ofcurrententry. Mathematical halfKelly 0.06392224669948926 is binary-average payoff approximation, not authorizedallocation. Portfolio totals unchanged by equity-denominator correction; all52savedholdouttrades reconcile2.84e-14pp. Neighbors10/30days sensitivity only. Source https://runway.adityaperswal.workers.dev/data/crypto-bars .

Diagnostic note2092 added after initialdailytraining inspected; not independentpreregistration/no winnerselection. Period2023-01-03–2026-09-10. Priorcalendar-month finalclose above10monthlycloseSMA holds followingmonth; only firsttradingdayopen signals, no catchup atsplitinitialsession; terminalclose liquidation. Fivefixedindependent equalcapital sleeves,1000each,no monthlycapitalrebalance,cash0%;SIP/alladjustment not independentlybuilt totalreturn. vectorbt from_orders longonly, fee1bp+slip15bp eachside; excludes taxes, cashyield, operatorcosts, no stops/targets. Authorpaperuses same-monthsignalclose andTbillcash; thislaggedproxy differs. Bothperiods profitable; WATCHmonthlyresearch, notlivepermission. Missing current>=3R stop/target and15min/whole-share implementation; modifying rule requires independenttest, existingobserved2023–2026 no longer untouched. Fiveassetmatch from200dailyreturns baseline train-3.20348/holdout12.65007%,stress-7.77259/+9.01443%; differentrule/exposuretiming, cannot attribute difference solelytostops orcausality. Sources https://mebfaber.com/wp-content/uploads/2016/05/SSRN-id962461.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . Code research/monthly_control.py andmonthly_control.json retained.

Diagnostic note2092 added after initialdailytraining inspected; not independentpreregistration/no winnerselection. Period2023-01-03–2026-09-10. Priorcalendar-month finalclose above10monthlycloseSMA holds followingmonth; only firsttradingdayopen signals, no catchup atsplitinitialsession; terminalclose liquidation. Fivefixedindependent equalcapital sleeves,1000each,no monthlycapitalrebalance,cash0%;SIP/alladjustment not independentlybuilt totalreturn. vectorbt from_orders longonly, fee1bp+slip5bp eachside; excludes taxes, cashyield, operatorcosts, no stops/targets. Authorpaperuses same-monthsignalclose andTbillcash; thislaggedproxy differs. Bothperiods profitable; WATCHmonthlyresearch, notlivepermission. Missing current>=3R stop/target and15min/whole-share implementation; modifying rule requires independenttest, existingobserved2023–2026 no longer untouched. Fiveassetmatch from200dailyreturns baseline train-3.20348/holdout12.65007%,stress-7.77259/+9.01443%; differentrule/exposuretiming, cannot attribute difference solelytostops orcausality. Sources https://mebfaber.com/wp-content/uploads/2016/05/SSRN-id962461.pdf ; https://runway.adityaperswal.workers.dev/data/stock-bars . Code research/monthly_control.py andmonthly_control.json retained.