Frozen before download in note1292. Prior completed SMA20>SMA100 enters at next adjacent open; inverse exits, long-only. Development2025 fast10/20/30 and slips30/60bp; holdout2026Jan1 throughSep8 fast20 only, each period starts flat. Warmup begins2024Sep1; HYPE begins2026Feb17 so no2025training. Vectorbt from_signals fees25bp cash each side (approximation to credited-asset fees), baseline slip30bp/stress60bp. No stop/target/profit-taking or15minute execution simulated; not venue-ready. Baseline {"symbol":"BTC","tf":"1Hour","fast":20,"window":"holdout","slip":0.003,"return_pct":-42.29664087605784,"drawdown_pct":53.230068260401595,"n":40,"win_rate_pct":25,"avg_win_pct":3.903639126530306,"avg_loss_pct":3.0110639872220326,"expectancy_pct":-1.2823882087839482,"quarter_kelly":0,"open":0,"missing":2,"held_gap_invalid":true,"zero_volume":1}. Stress {"symbol":"BTC","tf":"1Hour","fast":20,"window":"holdout","slip":0.006,"return_pct":-54.60915881186559,"drawdown_pct":62.08919733778772,"n":40,"win_rate_pct":20,"avg_win_pct":4.165648673702446,"avg_loss_pct":3.3844476879686507,"expectancy_pct":-1.874428415634431,"quarter_kelly":0,"open":0,"missing":2,"held_gap_invalid":true,"zero_volume":1}. Premises: current universe introduces survivor selection; 2026 already researched so not pristine unseen validation. Every hourly candidate has a missing interval while invested: diagnostic returns are not accepted execution outcomes; do not fill missing bars or count these as qualified tests. Conclusion reject for trading; even positive raw Kelly is overridden to0. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Reproducible scripts in this run's saved note.
return-42.3
drawdown53.23
win rate25
sample40
average win percent3.9
average loss percent3.01
expectancy percent-1.28
stress return-54.61
open trades0
held gap invalidtrue
raw quarter Kelly0
approved size0
BTC 1Day SMA20/100 diagnostic: one closed loss; no win estimate
Frozen before download in note1292. Prior completed SMA20>SMA100 enters at next adjacent open; inverse exits, long-only. Development2025 fast10/20/30 and slips30/60bp; holdout2026Jan1 throughSep8 fast20 only, each period starts flat. Warmup begins2024Sep1; HYPE begins2026Feb17 so no2025training. Vectorbt from_signals fees25bp cash each side (approximation to credited-asset fees), baseline slip30bp/stress60bp. No stop/target/profit-taking or15minute execution simulated; not venue-ready. Baseline {"symbol":"BTC","tf":"1Day","fast":20,"window":"holdout","slip":0.003,"return_pct":-22.030385869408622,"drawdown_pct":27.179148386270835,"n":1,"win_rate_pct":0,"avg_win_pct":null,"avg_loss_pct":22.36637631746888,"expectancy_pct":-22.36637631746888,"quarter_kelly":0,"open":1,"missing":0,"held_gap_invalid":false,"zero_volume":0}. Stress {"symbol":"BTC","tf":"1Day","fast":20,"window":"holdout","slip":0.006,"return_pct":-22.727935494515712,"drawdown_pct":27.61477445900433,"n":1,"win_rate_pct":0,"avg_win_pct":null,"avg_loss_pct":22.83228870881309,"expectancy_pct":-22.83228870881309,"quarter_kelly":0,"open":1,"missing":0,"held_gap_invalid":false,"zero_volume":0}. Premises: current universe introduces survivor selection; 2026 already researched so not pristine unseen validation. Only one closed losing trade plus one open position: positive marked returns do not estimate average win or a tradeable probability. Conclusion reject for trading; even positive raw Kelly is overridden to0. Source https://runway.adityaperswal.workers.dev/data/crypto-bars ; https://docs.alpaca.markets/us/docs/crypto-fees . Reproducible scripts in this run's saved note.
return-22.03
drawdown27.18
win rate0
sample1
average win percentunavailable
average loss percent22.37
expectancy percent-22.37
stress return-22.73
open trades1
held gap invalidfalse
raw quarter Kelly0
approved size0
Costs and realized evidence: one liquidated BTC trade cannot calibrate a flow trade
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 378-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-3.1982894279861673,"2022":-1.1115754020214585,"2023":-2.4559342553208685,"2024":-2.571763853282849,"2025":-1.0573886480381267}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-9.99
drawdown10.03
win rate30.25
sample314
average win1.44
average loss1
mean trade-0.26
quarter Kelly0
oneway costs bp16
mean days3.17
reconcile error0
actual size0
QTR9090200 test managed trend 378 target2 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 378-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-1.4266879110777175,"2022":0.03282691161201612,"2023":-1.5404086082613566,"2024":-0.8817272924663944,"2025":1.1059498552612235}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-2.71
drawdown4.19
win rate30.25
sample314
average win1.64
average loss0.8
mean trade-0.06
quarter Kelly0
oneway costs bp6
mean days3.17
reconcile error0
actual size0
QTR9090200 test managed trend 378 target1.5 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 378-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-3.2927516767092824,"2022":-0.9664546669718388,"2023":-2.139692711344532,"2024":-2.815434596853883,"2025":-1.7076777646982344}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-10.47
drawdown10.6
win rate39.17
sample314
average win0.96
average loss1.08
mean trade-0.28
quarter Kelly0
oneway costs bp16
mean days2.76
reconcile error0
actual size0
QTR9090200 test managed trend 378 target1.5 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 378-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-1.526994698235573,"2022":0.17326654036462497,"2023":-1.2198998112890935,"2024":-1.131832654620768,"2025":0.4228545307626952}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-3.26
drawdown4.23
win rate39.49
sample314
average win1.16
average loss0.89
mean trade-0.08
quarter Kelly0
oneway costs bp6
mean days2.76
reconcile error0
actual size0
QTR9090200 test managed trend 252 target2 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-2.8106851822086165,"2022":-0.3591077686596389,"2023":-3.5023094311812386,"2024":-3.8460042402528427,"2025":-1.9850996015152056}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-11.93
drawdown11.84
win rate27.91
sample301
average win1.48
average loss1.03
mean trade-0.33
quarter Kelly0
oneway costs bp16
mean days2.93
reconcile error0
actual size0
QTR9090200 test managed trend 252 target2 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-1.3727693505128036,"2022":0.5841447710436487,"2023":-2.3961675408427086,"2024":-2.027854114076788,"2025":0.0722878235541291}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-5.07
drawdown6.47
win rate27.91
sample301
average win1.69
average loss0.83
mean trade-0.13
quarter Kelly0
oneway costs bp6
mean days2.93
reconcile error0
actual size0
QTR9090200 test managed trend 126 target2 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 126-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-2.417784266178713,"2022":-1.0144705925223652,"2023":-3.5597848445372393,"2024":-4.3470581111143325,"2025":-2.7955311821025064}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-13.39
drawdown13.56
win rate26.17
sample298
average win1.32
average loss0.98
mean trade-0.38
quarter Kelly0
oneway costs bp16
mean days2.95
reconcile error0
actual size0
QTR9090200 test managed trend 126 target2 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 126-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 2R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-0.9425669160038841,"2022":-0.22491987580169903,"2023":-2.1567739866022007,"2024":-2.412266109670125,"2025":-1.1343689395069356}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-6.7
drawdown7.13
win rate26.17
sample298
average win1.53
average loss0.79
mean trade-0.18
quarter Kelly0
oneway costs bp6
mean days2.95
reconcile error0
actual size0
QTR9090200 test managed trend 126 target1.5 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 126-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-2.680782534324222,"2022":-1.0804103576898982,"2023":-3.404041102964517,"2024":-4.454786416045097,"2025":-2.8754301338842936}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-13.71
drawdown13.95
win rate34.56
sample298
average win0.86
average loss1.05
mean trade-0.39
quarter Kelly0
oneway costs bp16
mean days2.6
reconcile error0
actual size0
QTR9090200 test managed trend 126 target1.5 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 126-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-1.2155288712066614,"2022":-0.29652594673461596,"2023":-2.0006126766594523,"2024":-2.5222562598620657,"2025":-1.2271237568091564}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2026":-4.191249393480845}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2026":-2.7706881044148424}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Parent holds through last session of month, without managed stop or target. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2026":9.501462548738182}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion parent is diagnostic only; cannot transfer its edge to different exits. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Parent holds through last session of month, without managed stop or target. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2026":11.143377216690542}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion parent is diagnostic only; cannot transfer its edge to different exits. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return11.14
drawdown5.02
win rate54.24
sample59
average win5.3
average loss3.02
mean trade1.49
quarter Kelly0.07
oneway costs bp6
mean days20.78
reconcile error0
actual size0
QTR9090200 test managed trend 252 target1.5 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-3.373370323894198,"2022":-0.777860349589854,"2023":-3.264245863495463,"2024":-3.944768809858712,"2025":-2.5341718390499146}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-13.17
drawdown13.3
win rate35.55
sample301
average win0.96
average loss1.11
mean trade-0.37
quarter Kelly0
oneway costs bp16
mean days2.56
reconcile error0
actual size0
QTR9090200 test managed trend 252 target1.5 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-1.9505770967010672,"2022":0.15055079111834235,"2023":-2.1492579088145836,"2024":-2.13749270260134,"2025":-0.5190949031960157}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return-6.46
drawdown7.24
win rate35.88
sample301
average win1.15
average loss0.91
mean trade-0.17
quarter Kelly0
oneway costs bp6
mean days2.56
reconcile error0
actual size0
QTR9090200 test parent trend 252 target1.5 cost16bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Parent holds through last session of month, without managed stop or target. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-2.0080833581230917,"2022":-2.270405986327506,"2023":0.41805183322796147,"2024":2.3875767421257965,"2025":11.776261979038939}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion parent is diagnostic only; cannot transfer its edge to different exits. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
return10.06
drawdown9.16
win rate54.49
sample301
average win2.78
average loss2.78
mean trade0.25
quarter Kelly0.02
oneway costs bp16
mean days20.91
reconcile error0
actual size0
QTR9090200 test parent trend 252 target1.5 cost6.000000000000001bp
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Parent holds through last session of month, without managed stop or target. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2021":-0.510696541130673,"2022":-1.2531999903076407,"2023":1.610553685782068,"2024":4.4820758249972,"2025":14.301537981817747}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion parent is diagnostic only; cannot transfer its edge to different exits. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2017":-2.8148559976469922,"2018":-3.6408323943076137,"2019":-1.5476870632716744,"2020":0.5001805947369098}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Stop1 prior20day mean H-L,target 1.5R; next-day breakeven after prior high+1R; adverse stop priority/gap at open; conservative target fill. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage5bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2017":-1.4841720267347425,"2018":-2.2140289894281184,"2019":-0.0939076822628171,"2020":2.293375215453519}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion reject venue-style transfer; fixed variants reported without selecting best. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.
1. Frozen note1293, coverage amendment1295. Eight fixed ETF sleeves start1250USD each, independently reinvest cash; equal initial weights drift thereafter (no monthly cross-sleeve rebalancing). Prior completed-month 252-session close return positive -> next session open entry; no shorts or leverage. Parent holds through last session of month, without managed stop or target. No same-month reentry; last-session month-end liquidation. 2. Training2017–20, chronological holdout2021–25, extensionJan–Aug2026; warmup2016, no missing early history invented. Fees1bp plus slippage15bp EACH side. Vectorbt long-only from_orders matches independent event cash ledger. 3. Results and years {"2017":4.851691621792575,"2018":-6.964274702069339,"2019":4.892697691988768,"2020":-1.2945918791502198}. Simulated portfolio returns, not actual account. Completed trades only; no partial current month. Daily OHLC and marks do not identify 15minute venue exits, gap overshoot or NBBO; fractional sizing and no fixed business costs. ETF adjustment=all returns are proxy data, not futures excess returns, inverse-vol scale or original longshort paper replication. Era reused in prior research, not untouched prospective evidence. Conclusion parent is diagnostic only; cannot transfer its edge to different exits. Source https://runway.adityaperswal.workers.dev/data/stock-bars?feed=sip&adjustment=all ; https://w4.stern.nyu.edu/facdir/lpederse/papers/TimeSeriesMomentum.pdf ; QTR9090200 research/trend.py and archived results.