A news scan found the text describing a trading strategy for Ethereum. The strategy has a baseline that freezes when two conditions are met: the 30-day supply growth of the USDC stablecoin, which is a type of cryptocurrency pegged to the US dollar, is greater than 0, and the previous day's closing price is greater than the 200-day simple moving average, which is a calculation of the average price over 200 days. When these conditions are true, the strategy enters a trade at the next available opening price and exits when the conditions are no longer true. The strategy was trained on data from 2024 to 2025 and tested on data from January 1 to September 12, 2026. The results showed a return of 75.031958% during training, with a maximum drawdown of 33.073196% and a win rate of 33.333%. During the test period, the return was 3.666842% with a maximum drawdown of 4.774546% and a win rate of 100%. A control strategy that only considers price movements had a return of 11.301377% during the test period. The strategy uses a 25 basis point fee, which is a type of transaction cost, and a 10 basis point slippage, which is the difference between the expected and actual price of a trade. The results are based on data from several sources, including https://runway.adityaperswal.workers.dev/data/crypto-bars, https://stablecoins.llama.fi/stablecoin/2, https://stablecoins.llama.fi/stablecoin/1, and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline of an audit report for R91304 TEN. The report is about an automated trading fund manager's notes. The manager uses a system that looks at the average price of a stock over 20 days and 50 days, which is called a Simple Moving Average or SMA. The manager also looks at the opening price of a stock at 9:30 and uses a stop-loss and target price based on the stock's price. The manager uses a type of analysis called vectorbt, which is a way of looking at how a trading strategy performs. The strategy has a cost of 1 basis point fee plus 5 basis points for buying or selling, and a stress test of 15 basis points. The manager tested the strategy on historical data from 2020 to 2024 and found that it had a return of -7.54972% and a drawdown of 70.1081%. The manager also tested the strategy on data from 2025 to September 11 and found that it had a return of 60.37727% and a drawdown of 14.75847%. The manager notes that the strategy is not executable at every sample and that the actual manager runs less often. The manager also notes that there are missing observations in the data and that the strategy may not be validated for deployment. The manager concludes that the strategy has a positive regime in 2025, but that the raw training data is negative and has a large drawdown, and that the missing execution observations mean that the strategy should be watched but not validated for deployment. The data used for the analysis can be found at https://runway.adityaperswal.workers.dev/data/stock-bars.
A news scan found the text describing a trading strategy for Bitcoin. The strategy uses a filter based on the growth of a stablecoin, USDC, and its 30-day supply. The filter is combined with a price-only control. The strategy enters a trade when the filter is true and exits when it is false. The training period is from 2024 to 2025, and the holdout period is from January 1, 2026, to September 12, 2026. The warm-up period is in 2023. The strategy uses a 25 basis point fee and a 10 basis point slippage per side. The results show a return of 30.157692% during training, with a drawdown of 36.117294% and a win rate of 27.778%. In the holdout period, the return is -2.140782%, with a drawdown of 6.395037% and a win rate of 50.000%. The price-only control has a return of 10.679989% in the holdout period. The strategy uses a daily signal mapped to hourly opens. The fees are modeled in the quote currency, which is a close approximation but not exact. The strategy does not use stops, targets, or a trailing policy. The data sources are listed as https://runway.adityaperswal.workers.dev/data/crypto-bars, https://stablecoins.llama.fi/stablecoin/2, https://stablecoins.llama.fi/stablecoin/1, and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the text describing a trading strategy for Bitcoin. The strategy uses a filter based on stablecoin supply growth and price trends. It buys when the 30-day supply growth of USDC is increasing and the previous day's close is above the 200-day moving average. The strategy sells when these conditions are no longer met. The training period was from 2024 to 2025, and the test period was from January 1, 2026, to September 12, 2026. The strategy was tested with a 25 basis point fee and a 10 basis point slippage per trade. The results showed a 30.16% return during training, with a maximum drawdown of 33.91% and a win rate of 27.78%. In the test period, the return was -2.14%, with a maximum drawdown of 5.82% and a win rate of 50%. A price-only control strategy had a 10.68% return during the test period. The strategy uses data from sources including https://runway.adityaperswal.workers.dev/data/crypto-bars, https://stablecoins.llama.fi/stablecoin/2, https://stablecoins.llama.fi/stablecoin/1, and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline about Stitch Fix announcing its fourth quarter and full fiscal year 2026 financial results and conference call. An edge test was done on Stitch Fix stock, which is traded under the ticker SFIX, to see how it performed after the company announced its earnings from March 2024 to June 2026. The test looked at the stock's performance over 10 consecutive periods when the company announced its earnings, and it compared the results to the primary data from observations 27176 to 285 and 926 fresh SIP adjustments. The test found that the median absolute return of the stock from the close of the day before the earnings announcement to the close of the next day was 16.578014%. For example, on September 11, the stock price was $2.95, and on October 16, the call option with a strike price of $2.5 was trading at $0.50, plus a total cost of $0.05 per share, which means it would need to increase by 3.389831% to be profitable, but this only happened 4 out of 10 times. Another example is the put option with a strike price of $2.5, which would need to decrease by 21.016949% to be profitable, but this only happened 2 out of 10 times. The test also found that the call option with a strike price of $5 would need to increase by 72.881356% to be profitable, but this did not happen at all. The company announced that it would release its fourth quarter and full fiscal year 2026 financial results after the market close on Wednesday, September 23, 2026. The test used data from the option contracts of Stitch Fix, which can be found at https://runway.adityaperswal.workers.dev/data/stock-bars and https://runway.adityaperswal.workers.dev/data/option-bars. The test had several findings, including: 1. An existing study on vector backtesting remained frozen, meaning it did not change, and it found that the strategy had a return of 14.174754% and a drawdown of 14.475874% in the first 8 training periods, and a return of 9.472964% and a drawdown of 6.811989% in the last 2 holdout periods. 2. A diagnostic test was run on the momentum of the stock over 20 days and 1000 returns, and it found that the return was 41.383224% and the drawdown was 55.802626%. 3. The test looked at the history of the call option with an expiry date of October 16, and it found that there were only 4 traded days from August 20 to September 10, and none of them covered the completed September 23 event. 4. The test found that the default pre-event exit on September 22 required pre-event option return evidence, and the event response median of 16.578014% exceeded the priced 3.389831% by 13.188184 percentage points, which can motivate hold-exception research but does not fix sizing or validate pre-event rule. No actual hold-through position was adopted, and the binding gap actual costed options were out of scope, so the cash was retained.
A computer program ran a test on a set of rules for buying and selling cryptocurrencies, which are digital currencies like Bitcoin. The test, called a backtest, looked at the prices of seven cryptocurrencies: Bitcoin, Ethereum, Solana, Litecoin, Ripple, Hype, and Link. The test used a strategy called breakout, which means buying when the price goes above a certain level, and looked at the prices over 981 days. The results showed the return on investment, the biggest loss, the percentage of winning trades, and the number of trades that were closed for each cryptocurrency. The results were: Bitcoin had a return of 46.4608% with a biggest loss of 14.6924% and 34.328% of trades were winning, with 67 trades closed; Ethereum had a return of 41.4426% with a biggest loss of 24.9433% and 32.727% of trades were winning, with 55 trades closed; Solana had a return of 72.1517% with a biggest loss of 18.7954% and 52.778% of trades were winning, with 36 trades closed; Litecoin had a return of -17.5515% with a biggest loss of 47.2206% and 27.083% of trades were winning, with 48 trades closed; Ripple had a return of 476.5253% with a biggest loss of 19.9595% and 47.059% of trades were winning, with 34 trades closed; Hype had a return of -1.6918% with a biggest loss of 19.2206% and 29.412% of trades were winning, with 17 trades closed; and Link had a return of -22.5658% with a biggest loss of 44.6927% and 34.545% of trades were winning, with 55 trades closed. A closer look at the data confirmed that the test used the closing price from the day before and executed trades at the same closing price, without any fees or holding back some of the data for testing. However, these results do not confirm that the rules for buying and selling, called HIGH20/LOW10, or the managed pullback execution, which is a way of selling when the price pulls back, are valid. Another study with 156 data points provides more detailed controls and holds out some of the data for testing, but Hype was not included in the training data. No parameters were selected from this test and no orders were placed. The data came from https://runway.adityaperswal.workers.dev/data/crypto-bars and all the raw data is recorded in observations 26628-26841.
This is a description of a trading strategy for the LINK cryptocurrency. The strategy uses a fixed Donchian system, which is a type of trading system that uses recent highs and lows to make decisions, with a 20-day high entry point and a 10-day low exit point at the next open. The system looks at 480 and 240 hours of data, but ignores any missing hours and only considers prices before September 13, 2022, at 02:00 UTC, excluding the incomplete data from that day. The system also takes into account a buy fee equivalent to 0.0025 and a sell fee of 25 basis points, with a 10 basis point slip and a 30 basis point stress test. The results of the training from 2024 to 2025 show a return of 0.296718%, a maximum drawdown of 52.438495%, 12 closed trades, and a win rate of 33.333%. The results of the reset in 2026 show a return of -1.201527%, a maximum drawdown of 24.551662%, 4 closed trades, and a win rate of 50.000%, with a stress return of -2.963609%. The system recorded sensitivities for 15, 20, and 30 days, but did not select the best parameters. The data used is not entirely new and untouched. The return includes open marked positions, and the win rate only considers closed trades, without assuming a final fill. This strategy does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There were no fresh high entries across 7 periods, and no late entries. The data sources are https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment involved 156 cells and 26910 to 27075 observations, with a current top 10 survivor selection that limits generalization.
The C91302 LINK daily costed HIGH20 LOW10 control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to make decisions, with a prior 20 calendar-day high for entry and prior 10-day low for exit at the next exact open, over a period of 20 and 10 days. This strategy requires 24 hours a day of OHLC, which stands for Open, High, Low, Close, and only considers prices before September 13, 02:00 UTC, excluding the incomplete September 13 day. The Vectorbt, a backtesting and trading platform, credited a coin buy fee equivalent to 0.0025 and 0.9975, and a sell fee of 25 basis points, with a slip of 10 basis points and 30 basis points stress. The training from 2024 to 2025 had a return of -39.654890%, a drawdown of 56.876500%, with 11 closed trades, a win rate of 36.364%, and no open trades. The reset in 2026 had a return of 27.044492%, a drawdown of 14.403432%, with 2 closed trades, a win rate of 50.000%, and 1 open trade, with a stress return of 25.780875%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The calendar holdout is not newly untouched data. The return includes open marked positions, and the win rate is only for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There was no fresh high entry across 7, and no late entry. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment had 156 cells and 26910 to 27075 observations, with the current top 10 survivor selection limiting generalization. A news scan found the source of this information.
The C91302 HYPE hourly costed HIGH20 LOW10 control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to determine entry and exit points, with a prior 20 calendar-day high for entry and prior 10-day low for exit at the next exact open, over 480 and 240 hours. The data is incomplete, with missing calendar hours and prices before September 13, 2022, at 02 UTC, and the incomplete September 13 day is excluded. The Vectorbt, a backtesting and trading library, is credited with a coin buy fee equivalent to 0.0025 and 0.9975, and a sell fee of 25 basis points, with a slip of 10 basis points and 30 basis points stress. The training data from 2024 to 2025 is unavailable. The reset in 2026 showed a known and reused holdout return of 56.037399%, a drawdown of 31.948567%, with 4 closed trades and a win rate of 75.000%, and 0 open trades, and a stress return of 53.560657%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The HYPE starts on February 17, 2026, and the positive 2026 results are in-sample descriptive, meaning they are based on the same data used to train the model, with no pre-2026 training or valid out-of-sample data. The return includes the open marked position, and the win rate is based on closed trades only, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There is no fresh high entry absent across 7, and no late entry. The data source is from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment has 156 cells and 26910 to 27075 observations, and the current top 10 survivor selection limits generalization.
The C91302 HYPE daily control uses a fixed Donchian strategy, which is a type of trading strategy that uses the highest high and lowest low of a certain period, in this case, the prior 20 calendar-day high for entry and prior 10-day low for exit at the next exact open, over a period of 20 and 10 days. This strategy requires 24 hours a day of open, high, low, and close prices, known as OHLC, and only considers prices before September 13, 2022, at 02 UTC, excluding the incomplete day of September 13. The Vectorbt system is used to credit a coin buy fee equivalent to 0.0025 and 0.9975, and a sell fee of 25 basis points, with a 10 basis point and 30 basis point stress slip. Training data from 2024 to 2025 is unavailable. The reset in 2026 resulted in a known and reused holdout return of 37.837008%, a drawdown of 30.053656%, with 3 closed trades and a win rate of 66.667%, and no open trades, and a stress return of 36.192841%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The HYPE strategy started on February 17, 2026, and the positive results in 2026 are considered in-sample and descriptive, with no pre-2026 training or valid out-of-sample data. The return includes the open marked position, and the win rate is only for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There were no fresh high entries across 7 periods, and no late entries. The data source is from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment had 156 cells with observations from 26910 to 27075, and the current top 10 survivor selection limits generalization.
This is a description of a trading control for the cryptocurrency XRP. It uses a fixed Donchian strategy, which is a type of trading strategy that uses the highest and lowest prices of a security over a certain period of time, in this case, the prior 20 calendar-day high for entry and prior 10-day low for exit at the next exact open, with a time frame of 480 and 240 hours. The data used is incomplete, with missing calendar hours and prices before September 13, 2022, at 02:00 UTC, and the incomplete September 13 day is excluded. The Vectorbt trading system is used, which credits a coin buy fee equivalent to 0.0025 and 0.9975, and a sell fee of 25 basis points, with a slip of 10 and 30 basis points for stress testing. The training period from 2024 to 2025 had a return of 110.893498%, a drawdown of 56.069200%, with 10 closed trades and a win rate of 40.000%, and no open trades. The reset period in 2026 had a return of 2.433934%, a drawdown of 28.004057%, with 4 closed trades and a win rate of 25.000%, and no open trades, with a stress return of 0.808026%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The calendar holdout is not newly untouched data. The return includes the open marked position, and the win rate is closed only, with no assumed final fill. This control does not require a 2 ATR stop, 3 R target, quarter at 2 R, or 15-minute exit management, so it is not a live policy validation. There was no fresh high entry across 7, and no late entry. The source of the data is from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment had 156 cells and 26910 to 27075 observations, with the current top 10 survivor selection limits generalization.
The C91302 XRP daily costed HIGH20 LOW10 control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to make decisions, with a prior 20 calendar-day high as the entry point and a prior 10-day low as the exit point at the next exact open. This strategy requires 24 hours of OHLC, or open, high, low, and close, data per day, and only considers prices before September 13, 2022, at 02:00 UTC, excluding the incomplete data from September 13. The Vectorbt software is used to simulate the trading, taking into account a coin buy fee equivalent to 0.0025/0.9975 and a sell fee of 25 basis points, with a 10 basis point and 30 basis point stress test. The training period from 2024 to 2025 resulted in a return of 194.412594%, a drawdown of 49.868094%, with 6 closed trades and a win rate of 33.333%, and no open trades. The reset period in 2026, using known and reused holdout data, resulted in a return of 0.569605%, a drawdown of 27.161711%, with 3 closed trades and a win rate of 0.000%, and 1 open trade, with a stress return of -0.828169%. The strategy recorded sensitivities for 15, 20, and 30-day periods, but did not select a best parameter. The calendar holdout data is not newly untouched data. The return includes the open marked position, and the win rate is only calculated for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There were no fresh high entries across 7 periods, and no late entries. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment involved 156 cells and 26910-27075 observations, with the current top 10 survivor selection limiting generalization.
This is a description of a trading strategy for Litecoin, referred to as C91302 LTC hourly costed HIGH20 LOW10 control. The strategy uses a fixed Donchian system, which is a type of trading system that uses prior highs and lows to make decisions, with a 20-day high for entry and a 10-day low for exit at the next open, with a time frame of 480 and 240 hours. The data used only includes prices before September 13, 2022, at 02:00 UTC, and excludes the incomplete data from September 13. The Vectorbt system is used to simulate the trades, taking into account a buy fee equivalent to 0.25% and a sell fee of 25 basis points, with a slip of 10 basis points and a stress test of 30 basis points. The training period from 2024 to 2025 showed a return of 22.038201%, a maximum drawdown of 45.931078%, with 10 closed trades and a win rate of 40.000%. The reset period in 2026 showed a return of 3.611121%, a maximum drawdown of 12.675287%, with 2 closed trades and a win rate of 0.000%. The stress test return was 2.580579%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The calendar holdout data is not newly untouched. The return includes the open marked position, and the win rate is only for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There were no fresh high entries across 7, and no late entries. The data source is from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment includes 156 cells with observations from 26910 to 27075, and the current top 10 survivor selection limits generalization.
This is a description of a trading strategy for Litecoin (LTC). The strategy uses a fixed Donchian system, which is a type of trading system that uses prior highs and lows to make decisions, with a 20-day high entry point and a 10-day low exit point at the next open. The system requires 24 hours of data per day and only uses prices before September 13, 2022, 02:00 UTC, excluding the incomplete data from September 13. The system also takes into account a buy fee equivalent to 0.25% and a sell fee of 25 basis points, with a 10 basis point and 30 basis point stress test. The system was trained on data from 2024 to 2025 and had a return of -22.367606%, a maximum drawdown of 57.771811%, and a win rate of 33.333% for closed trades. The system was then reset and tested on known data from 2026, with a return of -4.170822%, a maximum drawdown of 16.265833%, and a win rate of 0% for closed trades. The stress test return was -5.123963%. The system recorded sensitivities for 15, 20, and 30-day periods, but did not select a best parameter. The data used for testing is not newly untouched data. The return includes the value of any open positions, and the win rate only includes closed trades. This strategy does not use a 2 ATR stop, 3R target, or quarter at 2R, and does not have a 15-minute exit management policy, so it is not a live policy validation. The system did not have any late entries. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment had 156 cells and 26910 to 27075 observations, and the current top 10 survivor selection limits generalization.
This is a description of a trading control for the SOL cryptocurrency. The control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to make decisions, specifically the high from the prior 20 calendar days to enter a trade and the low from the prior 10 days to exit a trade, with the trade happening at the next exact open, and the control is checked every 480 or 240 hours. The data used only includes prices before September 13, 02:00 UTC, and excludes the incomplete data from September 13. The control also takes into account fees, with a buy fee equivalent to 0.0025 and a sell fee of 25 basis points, and slip fees of 10 and 30 basis points. The control was trained on data from 2024 to 2025 and had a return of 9.329324%, a maximum drawdown of 36.797732%, and a win rate of 50.000% for closed trades. When applied to known data from 2026, the control had a return of 1.972209%, a maximum drawdown of 24.527633%, and a win rate of 0.000% for closed trades. The control also recorded sensitivities for 15, 20, and 30-day periods, but did not select a best parameter. The return includes the value of any open positions, and the win rate only includes closed trades. The control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, which are types of risk management strategies, so it is not a live policy validation. The data for this control comes from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment used 156 cells and 26910 to 27075 observations, and the current top 10 survivor selection limits the generalization of the results. A news scan found the source of this data.
This is a description of a trading strategy for the SOL cryptocurrency. The strategy uses a fixed Donchian system, which is a type of trading system that uses high and low prices to make decisions, with a 20-day high entry point and a 10-day low exit point at the next opening price. The system requires 24 hours of price data per day and only uses prices before September 13, 2022, 02:00 UTC, excluding the incomplete data from September 13. The Vectorbt system is used to simulate the trading, taking into account a buy fee equivalent to 0.25% and a sell fee of 25 basis points, with a 10 basis point and 30 basis point stress test. The results of the training from 2024 to 2025 show a return of -10.124640%, a maximum drawdown of 42.460968%, 7 closed trades, and a win rate of 42.857%. The results of the reset from 2026 show a return of 3.896506%, a maximum drawdown of 17.407810%, 3 closed trades, and a win rate of 0.000%. The stress test return is 2.452493%. The system recorded sensitivities for 15, 20, and 30-day periods, but did not select the best parameter. The calendar holdout is not newly untouched data. The return includes the open marked position, and the win rate is only for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3 R target, quarter at 2 R, or 15-minute exit management, so it is not a live policy validation. There were no fresh high entries across 7 periods, and no late entries. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment had 156 cells and 26910 to 27075 observations, with the current top 10 survivor selection limiting generalization. A news scan found the headline is not available, this is a raw script output.
This is a description of an automated trading strategy for Ethereum (ETH). The strategy uses a Donchian channel, which is a type of technical indicator, to determine when to buy and sell. It looks at the highest price in the last 20 days and the lowest price in the last 10 days to decide when to enter and exit trades. The strategy only considers complete calendar days and ignores any incomplete days, such as September 13. It also takes into account fees and slippage, which is the difference between the expected and actual prices of a trade. The strategy was tested using data from 2024 to 2025 and resulted in a return of 16.482836%, with a maximum drawdown of 41.890183%. It was also tested on a holdout set of data from 2026 and resulted in a return of -0.011938%, with a maximum drawdown of 22.646338%. The strategy does not use certain risk management techniques, such as a 2 ATR stop or a 3R target. The data used to test the strategy came from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment involved 156 cells and 26910 to 27075 observations, and the top 10 surviving strategies were selected.
The C91302 ETH daily control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to make decisions, with a 20-day high entry and a 10-day low exit at the next open. This strategy requires 24 hours of OHLC, or open, high, low, and close, data per day. Prices before September 13, 02:00 UTC are used, but the incomplete data from September 13 is excluded. The Vectorbt system is used to credit a coin buy fee equivalent to 0.0025 and 0.9975, and a sell fee of 25 basis points, with a 10 basis point and 30 basis point stress slip. From 2024 to 2025, the training return was -24.669800%, with a drawdown of 50.548573%, 12 closed trades, and a win rate of 41.667%. In 2026, the known and reused holdout return was -4.530094%, with a drawdown of 17.452215%, 3 closed trades, and a win rate of 0.000%. The stress return was -5.856989%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The calendar holdout is not newly untouched data. The return includes the open marked position, and the win rate is closed only, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There was no fresh high entry across 7 days, and no late entry. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. This experiment has 156 cells and 26910 to 27075 observations, with a current top 10 survivor selection that limits generalization.
This is a description of a trading strategy for Bitcoin. The strategy uses a Donchian channel, which is a type of technical analysis, to determine when to enter and exit trades. It looks at the highest price in the last 20 days and the lowest price in the last 10 days. If the price reaches a new high, it enters a trade at the next opening price, and if the price reaches a new low, it exits the trade. The strategy only considers prices from before September 13, 2022, at 02:00 UTC, and excludes incomplete data from that day. The Vectorbt backtesting tool was used to simulate the strategy, taking into account fees of 0.25% for buying and 25 basis points for selling, with a 10 basis point to 30 basis point stress test. The results from 2024 to 2025 showed a return of 30.438863% and a maximum drawdown of 34.235134%. In 2026, the return was 18.078296% and the maximum drawdown was 12.151994%. The strategy was tested on 15, 20, and 30-day sensitivities, but no best parameter was selected. The data used for testing is not entirely new and untouched. The return includes the value of any open positions, and the win rate only considers closed trades. This strategy does not require a stop-loss or target profit, and it does not use a specific exit management system. The data for this experiment came from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment involved 156 cells and 26910 to 27075 observations, and the current top 10 survivor selection may limit the generalization of the results. A news scan found the headline is not applicable here, this is a raw script output.
This is a description of a trading control for Bitcoin, denoted as C91302. The control uses a fixed Donchian strategy, which is a type of trading strategy that uses prior highs and lows to determine entry and exit points, with a prior 20 calendar-day high for entry and a prior 10-day low for exit at the next exact open. The control requires 24 hours of OHLC, or open, high, low, and close, data per day, and only considers prices before September 13, 2022, at 02:00 UTC, excluding the incomplete data from September 13. The control also accounts for a coin buy fee equivalent to 0.0025 and a sell fee of 25 basis points, with a slip of 10 basis points and a stress of 30 basis points. The training period from 2024 to 2025 resulted in a return of 32.159105%, a drawdown of 32.756294%, with 10 closed trades and a win rate of 30.000%. The reset period in 2026, using known and reused holdout data, resulted in a return of 18.337791%, a drawdown of 10.253074%, with 2 closed trades and a win rate of 50.000%. The stress return was 17.160774%. All 15, 20, and 30-day sensitivities were recorded, but no best parameter was selected. The calendar holdout is not newly untouched data. The return includes the open marked position, and the win rate is only for closed trades, with no assumed final fill. This control does not require a 2 ATR stop, 3R target, quarter at 2R, or 15-minute exit management, so it is not a live policy validation. There was no fresh high entry across 7 days, and no late entry. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The experiment consisted of 156 cells with 26910 to 27075 observations, and the current top 10 survivor selection limits generalization.
return18.34
drawdown10.25
win rate50
sample2
open_N1
train_return32.16
train_N10
stress_return17.16
slippage_bp10
R91302 TEN Test
BacktestswatchSupply chainTEN2026-09-13 02:13
A daily test of the automated trading fund manager R91302 TEN was conducted. The test used the same policy as 3063, which is based on a Simple Moving Average of 50 days, and a close price greater than the prior 20 days. The test also used a next open entry with a stop loss of 2 ADR and a target of 40 sessions with a maximum stop. The test allowed for gap opens and stops on the first day. The test was run on 1682 bars of data from 2020 to September 11. The Vectorbt long only strategy was used with a fee of 1 and a slip of 5 basis points per side. The test was trained on data from 2020 to 2024 and resulted in a return of 1.30262% with a drawdown of 70.98575% and 10 wins out of 33, or 30.303%. The test was then run on holdout data from 2025 to September 11 and resulted in a return of 78.94124% with a drawdown of 16.14993% and 7 wins out of 12, or 58.3333%. A fixed path slip of 15 basis points resulted in a return of 74.69777%. The test was closed on September 11 according to the window convention, but this was not a real exit. The test preserved the sensitivities of L10 and L40 without selection in observations. As of June 28, 2024, the test matched the 1131 pre-change TEN bars exactly. On July 1, 2024, the issuer ticker change was verified, and it was confirmed that the new ticker was not assumed to be the old Tenneco. A different IEX rule was used, which resulted in 1437 bars and 103 trades with a return of 58.0091% and a drawdown of 19.158%, and 37.8641% wins. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://www.globenewswire.com/news-release/2024/07/01/2906633/0/en/TEN-Ltd-NYSE-Ticker-Change-to-TEN-Effective-Today.html. The current price is 47.9, which is greater than 44.71. The training drawdown was large, and the mean confidence interval crossed zero, which requires a raw 15-minute audit and a cash distribution policy before live use. The adjusted bar returns do not show the actual fund cash credit. No order was placed.
return78.94
drawdown16.15
win rate58.33
sample12
train_return1.3
train_drawdown70.99
train_sample33
stress_return74.7
current_signal1
TNP_overlap1131
LUXE Stock Control
BacktestsrejectConvex eventsLUXE2026-09-13 02:12
A news scan found the headline about LUXE pre-event stock control failing despite strong training. The edge test was done on nine consecutive fiscal earnings from Q3FY24 to Q3FY26, which is from May 2024 to May 2026. The test adjusted the SIP, or Systematic Investment Plan, based on the prior day's close to the morning release day's close. The median absolute change was 5.245902%, with a range of -11.810013% to 23.8%. On September 11, the stock price was 7.39, and on October 16, it was 5.05. To make a profit, the stock price needed to fall by 33.694181%, but this did not happen in any of the 9 tests. Another test needed the stock price to increase by 8.525034%, which happened only once. The test also looked at the conviction risk, which is the risk of making a wrong decision. The issuer of the stock announced that the earnings release and conference call would be held before the US market open on September 16, 2026. The expiry date was October 16 plus 30 days. The default exit date was September 15. The test used data from August 26 to October 16, with 685 bars and 26 trades. The momentum was 20/1000 days, with a return of 87.879807% and a drawdown of 34.229923%. The test also used a vectorbt time-only stock control, which is a type of automated trading strategy. The strategy was tested on the first 7 events and held out on the last 2 events in February and May 2026. The training winner had a return of 39.558565%, while the holdout had a return of -9.482158%. The test also looked at the historical holdout, which had a wide uncertainty. The stocks could not validate the option decay, spread, or target hit odds. There was no actual option profit and loss history. The vectorbt drawdown is from the observed NAV peak, and initial costs can make the overall loss larger. The tested T5 transfer was rejected, and there was no invented option backtest.
return-9.48
drawdown8.41
win rate0
sample2
train_return39.56
train_drawdown10.96
train_win_pct57.14
train_n7
stress_return-10.2
CLI_return87.88
CLI_drawdown34.23
CLI_n26
R91302 KMT Trade
BacktestsrejectSupply chainKMT2026-09-13 02:10
A news scan found the headline 'R91302 KMT frozen L20 training rejects a scarcity breakout trade'. This refers to a trading strategy that uses a type of technical analysis called a scarcity breakout. The strategy was tested on historical stock data from January 2, 2020, to September 11, 2026, which is a total of 1682 daily bars. The strategy, called L20, looks for a specific pattern where the closing price is higher than the previous 20 closing prices and also higher than the 50-day simple moving average, which is a calculation of the average price over the last 50 days. The strategy enters a trade at the next open and sets a stop loss at 2 times the average daily range of the last 20 days and a target at 6 times the average daily range, which is a measure of the stock's volatility. The strategy also has a maximum of 40 sessions, which is a type of time limit. The strategy was backtested using a tool called Vectorbt, which is a type of backtesting software, with fees of 1 basis point and slippage of 5 basis points per side, which are types of trading costs. The results showed a return of -48.9617% and a drawdown of 54.1759% during the training period from 2020 to 2024, with 7 out of 27 winning trades, which is a 25.9259% win rate. During the holdout period from 2025 to September 11, 2026, the strategy had a return of 10.7920% and a drawdown of 39.6636%, with 5 out of 14 winning trades, which is a 35.7143% win rate. The strategy was also tested with a fixed slippage of 15 basis points, which resulted in a stress test return of 7.7328%. The results also showed that the strategy had negative sensitivities to changes in the parameters L10 and L40 during the training period, but not during the holdout period. A command-line interface diagnostic test of 1440 bars and 65 trades showed a return of -26.3901% and a drawdown of 34.7633%, with a win rate of 30.7692%, but this was not a validation test. The data used for this test came from a source at https://runway.adityaperswal.workers.dev/data/stock-bars. The trading engine was initially set up to allow short selling, but this was corrected to only allow long positions, and the portfolio return was calculated as the product of the returns of each individual trade, with a precision of 1e-8. The preliminary results from the wrong engine settings were not recorded as part of the analysis. The daily barriers used in the strategy were idealized and did not take into account the actual 15-minute venue management and dividend cash, and no live authorization was given. The training used a half-Kelly criterion, which is a type of betting strategy, with a size of 0, and the current size of 29.44 is less than the optimal size of 31.81, which implies a size of 0.
A trading system called Vectorbt was tested on the SOL cryptocurrency. The system buys when the closing price exceeds the highest closing price of the prior 20 calendar days, and sells when this condition is no longer true, using the exact opening price. The system was trained on data from August 26, 2024, to December 31, 2024, and tested on data from January 1, 2025, to September 12, 2026. The system charges a fee of 0.25% on each trade, and also accounts for slippage, which is the difference between the expected and actual price of a trade. The system uses daily high, low, open, and close prices from 24 complete hours of data, without filling in any missing data. The baseline training return was -72.51%, with a maximum drawdown of 73.42%, 133 trades, and a win rate of 15.79%. The holdout return was -32.09%, with a maximum drawdown of 32.42%, 55 trades, and a win rate of 18.18%. The system also tested a high threshold control, which had a return of -13.64% on 24 trades. A news scan found the source of the data to be https://runway.adityaperswal.workers.dev/data/crypto-bars and the source of the fee information to be https://docs.alpaca.markets/us/docs/crypto-fees. The system was not adopted due to various reasons, including disagreement between daily and hourly data, and missing required data. This system is not similar to other systems that use a 10-day low exit or a managed ETH pullback.
A trading strategy for SOL was tested using a system called Vectorbt. The strategy is to buy when the closing price is higher than the highest closing price of the last 20 calendar days, and sell when this is no longer true, using the exact opening price. The test used hourly data from 480 bars, trained from 2024 to 2025, with SOL starting on August 26, and a cash reset on January 1. The test also included a holdout period from 2026 through September 12. The strategy included fees of 25 basis points on each side, credited as an asset equivalent buy fee of 0.0025 and 0.9975, and slippage of 10 basis points and 30 basis points for stress testing. The daily high, low, open, and close prices were used from 24 complete hours, with missing calendar windows remaining unavailable and no filling. The baseline train return was 10.433149%, with a drawdown of 20.817749%, 24 trades, and a win rate of 45.833%. The holdout return was 18.625863%, with a drawdown of 14.948472%, 11 trades, and a win rate of 45.455%. The stress test return was 13.519467%. A high threshold control holdout return was 8.392065% with 7 trades. All 15, 20, and 30 day close and high price cells were recorded from 26572 to 83, with a known holdout range of 3.995 to 18.626%, but none were selected. The premise is that same-close gross gains are not evidence for a costed next-open rule. A counterexample is that daily SOL remains positive, but hourly disagreement and missing required 3R, partial, and 15-minute policy prevent adoption. This strategy is not a prior 10 low exit Donchian or managed ETH pullback. The conclusion is to reject this strategy as a live replacement. The data sources are https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.