Notes

A news scan found the following data line: a train hourly weekend return distribution for BTC/USD, also known as Bitcoin to US Dollar, with the following details. The data was collected from a training window, which is a period of time used to train a model, with 4990 data points, from 2024 to 2025, and a holdout period, which is a period of time used to test a model, from January 1, 2026, to September 12, 2026. The data was downloaded from a source venue on September 13. The mean, or average, return was 0.38835636790084094, and the standard deviation, which is a measure of how spread out the data is, was 35.60555933003448. The absolute value of the 99th percentile, which is a measure of the worst 1% of returns, was 1.3497227486760532%, and the worst hourly return was -4.290659668464924%. The data includes log close returns, which are returns calculated based on the natural logarithm of the closing prices, with no bridging of missing hours, and cross-coin correlation, which is a measure of how different coins move together, was calculated using complete cases, which means that only data points with complete information were used. This data is descriptive, meaning it describes what happened, and is not a strategy for timing weekend trades. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91306 train hourly weekday return distribution'. The data shows the return distribution for Bitcoin, symbol BTC, from a training dataset. This dataset has 12549 entries and only includes weekdays. The average hourly return is 0.4318934484333188, with a standard deviation of 57.65163458883257. Standard deviation is a measure of how spread out the returns are. The absolute 99th percentile return is 2.112540798162487 percent, and the worst hourly return is -4.895889045284707 percent. The data was collected from log close returns, without filling in missing hours, and was split into training and holdout periods, from 2024-2025 and 2026 January 1 to September 12, respectively. The data source is venue bars downloaded on September 13. The standard deviation and mean are given in basis points, which are a unit of measurement for returns. The absolute 99th percentile and worst returns are given as percentages. The correlation with other coins is also complete. This data is descriptive and does not include a weekend timing strategy. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about the SOL/USD C91306 calendar history quality for a 1 Hour time frame. The symbol is SOL and the time frame is 1 Hour. There are 32424 rows of data, with 10085 rows missing. The first piece of data is from 2023-01-01 00:00:00+00:00 and the last piece of data is from 2026-09-12 23:00:00+00:00, with 100 events. This information was retrieved on September 13 at 06:03-04 UTC. Some data from September 13 is not included. Missing calendar slots mean that rolling high and Average True Range, which is a measure of price volatility, signals are not used, and only mark-to-market, which is a way of valuing assets, is used to fill in missing close prices. There are 418 missing daily spans of data from 2023 to 2024, so the data from 2024 to 2025 should not be used for full training. Duplicate timestamps, which are repeated time stamps, have been removed. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline is not available, but a data line was retrieved on September 13 at 06:03-04UTC. The data line is about the SOL symbol with a 1Day time frame, which has 1351 rows of data, but 418 rows are missing. The first piece of data is from January 1, 2023, at 00:00:00+00:00, and the last piece of data is from September 12, 2026, at 00:00:00+00:00. There are 29 events in the data. The data is incomplete, with 418 missing calendar slots from 2023 to 2024, which affects the rolling high and Average True Range, or ATR, signals. The ATR is a measure of how much a stock's price moves. The missing data only allows for mark-to-market calculations going forward, using the close price. The data from 2024 to 2025 should not be used for full training because it is incomplete. Any duplicate timestamps have been removed. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about the ETH/USD C91306 calendar history quality for a 1 Hour time frame. The data shows the symbol is ETH, with a time frame of 1 Hour, and it has 32424 rows of data, but 6 rows are missing. The first piece of data is from 2023-01-01 00:00:00+00:00 and the last piece of data is from 2026-09-12 23:00:00+00:00, with 172 events in total. This information was retrieved on September 13 at 06:03-04 UTC. Some data from September 13 is not included, and the missing calendar slots mean that rolling high and Average True Range, which is a measure of price volatility, signals are not used, and only mark-to-market, which is the value of assets at current prices, is used to fill in the missing close prices. There are also 418 missing daily spans for SOL from 2023 to 2024, so the full 2024 to 2025 training data should not be used. Any duplicate timestamps have been removed to ensure the data is accurate. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline is not available, but a data line was retrieved on September 13 at 06:03-04UTC. The data line is about the Ethereum symbol, with a 1-day time frame, and it has 1351 rows of data with no missing rows. The first data point is from January 1, 2023, at 00:00:00, and the last data point is from September 12, 2026, at 00:00:00. There are 46 events in the data. The data quality is affected by missing calendar slots, which suppresses certain signals, and only allows forward filling of close prices. The Solana daily data has 418 missing spans from 2023 to 2024, so the 2024 to 2025 training data should not be used. Duplicate timestamps have been removed. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the data line: {"symbol":"BTC","tf":"1Hour","rows":32424,"missing":4,"first":"2023-01-01 00:00:00+00:00","last":"2026-09-12 23:00:00+00:00","events":188}. This is a data summary retrieved on September 13 at 06:03-04UTC. It shows that some data is missing, which affects certain calculations, and some data from September 13 is not included. The data for SOL is also missing for 418 days between 2023 and 2024, so the training data for 2024-2025 should not be used. Duplicate timestamps have been removed. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline about a calendar history for BTC/USD with the following details: it has a symbol of BTC, a time frame of 1 day, 1351 rows of data, and no missing data points. The first data point is from 2023-01-01 00:00:00+00:00 and the last data point is from 2026-09-12 00:00:00+00:00, with 56 events in total. This information was retrieved on September 13 at 06:03-04 UTC. Some data from September 13 is not included. Missing calendar slots affect signals, so only mark-to-market data is used going forward. There are 418 missing daily data points for SOL from 2023 to 2024, so the 2024-2025 training data should not be used. Duplicate timestamps have been removed. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay4 slip30bp'. The trade returns percentage is given as a list: [-2.6855360880560952, 17.572886666666676, 17.323570646651888, -12.061626145520854, 17.318134178998324, -0.6635701138521632, 6.102041728247842, -11.02047517707212, -12.391346453778507]. This data was retrieved on September 13 at 06:03-04UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a known holdout period in 2026. The model used a gross return of 3R, 2ATR, and 20 days, with sampled open exits and terminal liquidation. A vectorbt exact coin fee equivalent of 10/30bpslip was applied, with no model fills. The full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but this did not change the result. The trade had a return of 13.784420505082947, a drawdown of 33.73180647313123, and a win rate of 44.44444444444444, based on a sample of 9 trades, with a mean trade percentage of 2.166008804698332 and a half Kelly fraction of 0.9567173990314317, over 731 days, using the SOL symbol, a 1Hour time frame, a delay of 4, and a slippage of 30 basis points. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay4 slip10bp'. The trade returns percentage is given as a list: [-2.295496984083034, 18.044122097902115, 17.793806812278245, -11.709166673781047, 17.78834855510962, -0.26542691347724023, 6.527301690865928, 3.526560045925109, -12.040208510292793]. This data was retrieved on September 13 at 06:03-04UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a holdout period in 2026. The model uses a gross return of 3R, 2ATR, and 20 days, with sampled open exits and terminal liquidation. The vectorbt model includes an exact coin fee equivalent and a 10/30bps slip. The model does not fill orders. A full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. The trade statistics are: return of 36.69165039103146, drawdown of 26.838415518202975, win rate of 55.55555555555556, sample size of 9, mean trade percentage of 4.152204457827434, and half Kelly fraction of 2.478271835536935, with 2 skipped entries over 731 days, using the SOL symbol, 1Hour timeframe, 4 delay, and 10 slippage bp. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay2 slip30bp'. The trade returns for SOL/USD were calculated with the following results: return of 39.16, drawdown of 30.90, win rate of 66.67, sample size of 9, mean trade percentage of 4.27, and half Kelly fraction of 2.34. The trade used a 1 Hour time frame, a delay of 2, and a slippage of 30 basis points. The results were based on 731 days of data and a training window. The trade returns percentage was calculated as follows: -2.55, 18.82, 11.39, -11.90, 15.77, 2.85, 6.99, -11.18, and 8.24. The data was retrieved on September 13 at 06:03-04 UTC and included complete UTC bars through September 12. The model used a full gross 3R, 2 ATR, and 20-day period, with sampled open exits and terminal liquidation. The vectorbt exact coin fee equivalent was used, with a 10/30 basis points slippage. The model did not fill, and the full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay2 slip10bp'. The trade returns percentage is given as a list: [-2.1554368980076233, 19.29834347729793, 11.83162410298635, -11.551044503403762, 16.238146550641353, 3.2574745277854644, 7.41618277618068, -10.823205255564483, 8.677443711121358]. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a known holdout period in 2026. The model used a gross return of 3R, 2 ATR, and 20 days, with sampled open exits and terminal liquidation, which means the model closed trades when they were open and also considered the final liquidation of assets. Vectorbt, a backtesting library, was used with an exact coin fee equivalent plus 10/30 basis points slippage, which refers to the difference between the expected and actual price of a trade. The model does not fill, meaning it does not adjust for trading fees. A full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but this did not change the result. The return was 44.26454674413809, the drawdown was 30.34690065215855, the win rate was 66.66666666666666, the sample size was 9, the mean trade percentage was 4.687725387670808, the half Kelly fraction was 2.5778743326517204, and the number of skipped entries was 5, over 731 days, for the symbol SOL, with a 1 Hour time frame, a delay of 2, and a slippage of 10 basis points. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay0 slip30bp'. The trade returns for SOL/USD are as follows: the return is 24.128878339405176, the drawdown is 39.27212760725306, the win rate is 54.54545454545454, the sample size is 11, the mean trade percentage is 2.744779504223641, and the half Kelly fraction is 1.1035837157870332. The trade details include a symbol of SOL, a time frame of 1 hour, a delay of 0, and a slippage of 30 basis points. The trade returns percentage is [-1.364949246044389, 21.969392425005175, 16.52598396343432, -13.342296650279584, 16.84638769157647, 3.2992046776715744, 10.408181387216263, -10.990765815137538, -13.203511918316401, 8.964738874730639, -8.91979084339648]. This data was retrieved on September 13 at 06:03-04 UTC and includes complete UTC bars through September 12. The training period is from 2024 to 2025, with a holdout period of 2026. The model uses a vectorbt exact coin fee equivalent, plus 10 and 30 basis points slippage, and does not fill orders. A full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Hour train delay0 slip10bp'. The trade returns percentage is 25.80951170226426, with a drawdown of 38.75805309385727 and a win rate of 54.54545454545454. This is based on a sample of 11 trades, with a mean trade percentage of 2.906327207954995. The half Kelly fraction is 1.1074234688921152. There were no skipped entries over 731 days. The symbol is SOL, with a 1 Hour time frame, 0 delay, and 10 basis points slippage. The trade returns percentage for each trade is -0.9696171929261643, 22.45824917477497, 16.993023379237915, -12.994970141884377, 17.314711294519448, 3.713230790686617, 10.850700478505004, -10.952523542354708, -12.855629156818383, 9.401472607897077, and -10.98904840413245. This data was retrieved on September 13 at 06:03-04 UTC and includes complete UTC bars through September 12. The training period is 2024-25, with a holdout period of 2026. The model uses a full gross 3R, 2 ATR, and 20-day period, with sampled open exits and terminal liquidation. The vectorbt exact coin fee equivalent is plus 10/30 basis points slippage. The model does not fill orders. A full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but this did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Day train delay0 slip30bp'. The trade returns percentage is 29.704431327084198, with a drawdown of 32.13358173937147 and a win rate of 55.55555555555556. This is based on a sample of 9 trades, with a mean trade percentage of 3.726327313674249. The half Kelly fraction is 1.350967339190396. There were 0 skipped entries over 731 days, for the symbol SOL, with a 1Day time frame, 0 delay, and 30 basis points of slippage. The trade returns percentage for each trade is -3.743604098893616, 18.16655544120327, 11.667660265636037, -11.23171079234605, 22.028455509325173, -9.79541379078145, 6.987374375452204, -13.591984767701536, and 13.049613681174211. This data was retrieved on September 13 at 06:03-04UTC, with complete UTC bars through September 12, and a training period from 2024 to 2025, and a holdout period in 2026. The model used a 3R, 2ATR, and 20-day window, with sampled open exits and terminal liquidation, and a vectorbt exact coin fee equivalent plus 10 and 30 basis points of slippage. The model does not fill, and the full timestamp ledger is in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91306 delayed HIGH 1Day train delay0 slip10bp'. The trade returns percentage is given as a list: [-3.357805761218824, 18.640170313557448, 12.11522738363593, -10.875924996566521, 22.517548985595482, -9.43387127628801, 7.41618277618068, -13.245659016292265, 13.502719707269195]. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a holdout period in 2026. The model used a 3R, 2ATR, and 20-day window, with sampled open exits and terminal liquidation. The results include a vectorbt exact coin fee equivalent plus 10/30 basis points slippage. The model does not fill orders. A full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but this did not change the result. The return was 34.458878026702806, the drawdown was 31.588470493420452, and the win rate was 55.55555555555556, based on a sample of 9 trades, with a mean trade percentage of 4.142065346208124 and a half Kelly fraction of 1.51244063285213. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay4 slip30bp'. The results of a trade simulation for the ETH symbol are as follows: the return was 80.86, the drawdown was 38.14, and the win rate was 52.94 percent. The simulation was based on a sample of 17 trades, with a mean trade percentage of 4.45. The half Kelly fraction, a measure of optimal bet size, was 1.558. There were 6 skipped entries in the data. The simulation covered 731 days, using a 1 hour time frame, a delay of 4, and a slippage of 30 basis points, which is a measure of the difference between expected and actual trade prices. The trade returns percentages were: -9.64, 18.82, 19.97, -9.20, -11.57, -0.02, -10.95, -8.84, 18.42, 8.51, 23.72, 15.10, 27.73, 3.51, 10.16, -8.82, and -11.33. The data was retrieved on September 13 at 06:03-04 UTC and includes complete UTC bars through September 12. The model used a 3R, 2 ATR, and 20-day window, which refers to a measure of price movement, and sampled open exits and terminal liquidation. The vectorbt model, a backtesting framework, was used with an exact coin fee equivalent and a 10/30 basis points slippage. The model does not fill orders. The full timestamp ledger is available in the results.json file under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay4 slip10bp'. The trade results for ETH/USD are as follows: the return is 90.02254429139484, the drawdown is 36.88848596186753, and the win rate is 58.82352941176471, based on a sample of 17 trades. The mean trade percentage is 4.765458424329873 and the half Kelly fraction is 1.5944071109014066. There were 6 skipped entries over 731 days, with a delay of 4 and a slippage of 10 basis points. The trade returns percentage is listed as follows: -10.946376815330193, 19.300372069337833, 20.455211949915352, -8.832433030194498, -11.220016085503559, 0.3813336182278082, -10.596488539128313, -8.478553689457568, 18.88976254455301, 8.94634221008166, 24.213813819935815, 15.56298316557938, 28.239601388038317, 3.926195783534636, 10.598850229627033, -8.450747493091715, -10.977057912517152. This data was retrieved on September 13 at 06:03-04UTC, with complete UTC bars through September 12, and a training period of 2024-25 with a holdout period of 2026. The model used a full gross 3R, 2ATR, and 20-day period, with sampled open exits and terminal liquidation, and a vectorbt exact coin fee equivalent plus 10/30 basis points slippage. The model does not fill, and the full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay2 slip30bp'. The trade returns percentage is given as a list: [-10.386448882967958, 18.677008146279682, 19.66617309879166, -13.800866616606289, 0.7874050793369669, -10.194829130498274, -8.554331394228253, 16.294763900613507, 10.174018518651607, 23.207863420307184, 11.646579270119028, -12.696972562121944, 27.907530460785136, 0.40806943683622254, 8.14203912554634, -7.500518802571687, -11.84405051086621]. This data was retrieved on September 13 at 06:03-04 UTC. The trade used a 1-hour time frame, or period, and had a delay of 2, which is a time delay, and a slippage of 30 basis points, or a small price difference. The results are based on a training period from 2024 to 2025, with a holdout period in 2026. The model used a vectorbt, a type of trading software, with a coin fee equivalent to 10 or 30 basis points, and did not fill all orders. The full results are available in a file called results.json. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay2 slip10bp'. The trade returns for ETH/USD are as follows: the return is 65.13182818830985, the drawdown is 29.70900019235322, and the win rate is 58.82352941176471. This is based on a sample of 17 trades, with a mean trade percentage of 3.9132250347705266. The half Kelly fraction is 1.3087744922116473. There were 8 skipped entries over 731 days. The symbol is ETH, the time frame is 1 hour, the delay is 2, and the slippage is 10 basis points. The trade returns percentage is [-10.027275253720447, 18.51711202894699, 20.145798480827025, -13.455378069512891, 1.1913638290299344, -9.834887483902143, -8.187814586639563, 16.76087657991261, 10.615599078587756, 23.701683996602707, 12.094061894896901, -12.347059576660547, 28.42018743454675, 0.8105077986221954, 8.575475455016468, -8.964707025581642, -11.49071898987315]. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a train period of 2024-25 and a known holdout period of 2026. The model uses a gross 3R, 2 ATR, and 20-day period, with sampled open exits and terminal liquidation. The vectorbt exact coin fee equivalent is plus 10/30 basis points slippage. The model does not fill, and the full timestamp ledger is in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay0 slip30bp'. The trade returns percentage is given as a list: [-11.592138960515497, 18.519482774248775, 16.961402219428656, -10.998014295926817, -13.91030110641186, 6.374793109093457, -9.816965149359437, -10.253666507666924, -9.175381535200344, 17.47246425920448, 9.46970870791728, 34.18815526197392, -14.374809081759498, -11.894408757009355, 27.996353485470582, 2.6240816269850864, 13.851722157863144, -7.500518802571676, -11.274011669148177]. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a known holdout period in 2026. The model uses a gross return of 3R, 2 ATR, and 20 days, with sampled open exits and terminal liquidation. Vectorbt is a backtesting library, which is used here to calculate the exact coin fee equivalent plus 10 and 30 basis points slippage. A basis point is a unit of measurement, equal to 0.01%. The model does not fill, meaning it does not adjust for trading fees. The full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Hour train delay0 slip10bp'. The trade fund manager reported the following results for the ETH symbol: a return of 30.213027025341177 percent, a drawdown of 46.15312671371922 percent, and a win rate of 47.368421052631575 percent, based on a sample of 19 trades. The mean trade percentage was 2.555082441190936 percent. The half Kelly fraction was 0.6655620017244688. There were 2 skipped entries over 731 days, with a delay of 0 and a slippage of 10 basis points. The trading window was 1 hour. The trade returns percentage was as follows: -11.237797770954883, 18.994512189264068, 17.430186803828928, -10.641291838027357, -13.565251175826532, 6.801146266904934, -9.455509013316387, -9.893960682930658, -8.811353914015918, 24.21878532661148, 9.908466374354852, 34.72598515630019, -14.031620912182841, -11.541279073280874, 28.509366464409204, 3.0354018278553774, 14.308043057256082, -8.964707025581664, -11.242555678040212. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a known holdout period in 2026. The gross results are based on a 3R and 2ATR with a 20-day period, using sampled open exits and terminal liquidation. The model used a vectorbt exact coin fee equivalent with a 10 and 30 basis points slippage. The model does not fill orders. A full timestamp ledger is available in the results.json file under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Day train delay1 slip30bp'. The trade returns percentage is given as a list: [-12.705597244397138, 18.95304657836776, -12.219363745936052, -4.382742454684085, -5.021752669607194, 19.523498405827723, -11.09681395354183, 11.12746615986715, 26.134479559945433, -8.472333778633189, 1.1656330316294206, -11.841269385517005]. This data was retrieved on September 13 at 06:03-04 UTC. The trade simulation used complete UTC bars through September 12, with a training period from 2024 to 2025 and a holdout period in 2026. The simulation used a gross return of 3R, 2 ATR, and 20 days, with sampled open exits and terminal liquidation, which means the simulation used a method to determine when to close trades based on the average true range, a measure of market volatility. The Vectorbt model, a backtesting framework, was used with an exact coin fee equivalent and a slippage of 10/30 basis points, which is a measure of the difference between the expected price of a trade and the actual price. The model does not fill orders, meaning it does not adjust for the actual execution of trades. A full timestamp ledger is available in the results.json file under /tmp/codex-run-3DmDYP. An oversized record was repaired, but this did not change the result. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Day train delay1 slip10bp'. The trade returns percentage is given as a list: [-12.355718826904505, 17.027208756886726, 19.434466891301728, -3.9995058014864693, -4.641077187294562, 20.002551943687386, -10.740487487444383, 11.572868164722339, 26.64003009590874, -10.56170554751411, 1.571107729575938, -11.487926717695396]. This data was retrieved on September 13 at 06:03-04 UTC. The data includes complete UTC bars through September 12, with a training period from 2024 to 2025 and a holdout period in 2026. The model used a gross return of 3R, 2 ATR, and 20 days, with sampled open exits and terminal liquidation. A Vectorbt exact coin fee equivalent plus 10/30 bps lip was applied, with no model fills. The full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. The trade had a return of 36.47267904464448, a drawdown of 21.57214775719677, and a win rate of 50, based on a sample of 12 trades, with a mean trade percentage of 3.538484334478619 and a half Kelly fraction of 1.230337788564526, over 731 days, with a symbol of ETH, a time frame of 1 day, a delay of 1, and a slippage of 10 basis points. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91306 delayed HIGH 1Day train delay0 slip30bp'. The trade results for ETH/USD are as follows: the return was -19.996206876800684, the drawdown was 45.63156263014141, and the win rate was 31.25. The sample size was 16, the mean trade percentage was -0.39026800899233693, and the half Kelly fraction was 0. The trade was delayed by 0 days and had a slippage of 30 basis points. The results were retrieved on September 13 at 06:03-04UTC and are based on complete UTC bars through September 12. The training period was from 2024 to 2025, with a holdout period from 2026. The model used a 3R, 2ATR, and 20-day window, with sampled open exits and terminal liquidation. The results include a vectorbt exact coin fee equivalent and a 10/30bps slip. The model does not fill orders. The full timestamp ledger is available in results.json under /tmp/codex-run-3DmDYP. An oversized record was repaired, but it did not change the result. The trade returns percentage was [-11.63105866750891, 18.125358679089242, -14.066927599309121, -13.238668258443232, -0.019391468247387245, -10.624097185204418, -9.096677357934425, 24.952891285441446, -7.4840796777366325, 20.133360447341442, -11.089161744663068, 26.751203289554315, -10.838407286137464, 11.334134830377018, -7.809905907419957, -11.642861523076242]. https://runway.adityaperswal.workers.dev/data/crypto-bars