The automated trading fund manager tested a signal on Chewy stock without looking ahead and with costs. The test had three main goals: to test the signal, to keep the holdout period unchanged, and to remember that stock wins and losses are not the same as option payoffs. A command line interface test using the symbols CHWY, a strategy called sma-cross, and other parameters returned a loss of 12.4010% and a drawdown of 53.5573% with 26.6667% wins over 15 trades. This test did not use a holdout period or adjust prices. Another test using vectorbt and adjusted IEX 2022-2026 bars had different results. The test used a simple moving average cross strategy and executed trades at the next close. It used 2022-2025 for training and 2026 for testing, with costs of 10 basis points per leg and stress of 30 basis points. The results were a loss of 10.9050% and a drawdown of 53.0218% for the training period, and a gain of 3.2484% and a drawdown of 19.3494% for the holdout period. The test also showed that the sample size was small and the timing differences between the tests explained the different results. The trade edge for options is unknown and there were no actual option bid-ask holdouts or own fund fills. The manager rejected using a directional proxy to validate options. The data source was https://runway.adityaperswal.workers.dev/data/stock-bars.
This is an evidence review only, using previous figures from FRO36/59, not a new backtest or sample. On September 6, a readback was done on a frozen strategy that uses a next-open 20 close breakout, with a 1 ADR stop and 2 ADR target, over 5 sessions, using vectorbt for long-only trades with 5 basis points fees and 10 basis points slippage, trained from 2022 to 2024 and held out from 2025 to September 4, 2026. The results showed a 20.24% return and 24.42% drawdown, with 41 trades, a 41.46% win rate, and an average win of 5.46% and average loss of 2.96%. The mean confidence interval was between -0.874% and 1.959%, which includes zero. A total of 53 out of 417 trades were already executed and audited. Previous tests using a six-sector baseline showed a 3.545% return with 26.398% double cost, while a trend and low volatility strategy showed a 10.294% loss over 221 trades. New notices did not repair controls or identify a new stock rule. There is no new strategy, redundant retest, or selected winner optimization. Studies 437 to 439 are not stock backtests. The historical holdout data is not pristine prospective data, with a size of 0. Prior tanker edge cannot be transferred to Maersk, Hapag, or ZIM. The data source is https://runway.adityaperswal.workers.dev/data/stock-bars, with query research supply numbers 36, 53, 176, 409, 410, 417, and 420. Code archive note 390 remains a journal, even though the prior temporary workspace is absent.
This is a report on the QTS90608 BASKET, a 12-month cash trend analysis. The analysis is based on three premises: first, a frozen protocol note 401 from August 2021 to August 2026; second, costs include a 5 basis point fee and 5 basis point slippage each way; and third, the analysis uses a vectorbt reconciled independent wealth and non-negative cash/assets. The results show a return of 48.08817616635035%, a drawdown of 9.961345368540774%, and a win rate of 58.82352941176471% over 68 months. The basket is equal-weighted and resets monthly across six sleeves. A news scan found the data line with the key 'all/12/later/10/BASKET' and the given metrics. The analysis concludes that historical adaptation should be watched, but no current entry is adopted. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
A news scan found the following text: Premises: 1. A frozen protocol note, number 401, was used from August 2021 to August 2026. This period was inspected under prior rules. The monthly return of a 12-month ETF was compared to the return of BIL. The comparison was done from the first session open to the next month's first open, with only long positions and cash, and with monthly cohort round trips forced. 2. Costs included a 5 basis point fee and 5 basis point slippage each way. Vectorbt was used to reconcile independent wealth and non-negative cash and assets. 3. A data line found the following metrics for the symbol DBC: a return of 104.69636283312967, a drawdown of 27.54428907422235, a win rate of 59.523809523809526, a sample size of 42, an average win of 5.446640655425865, an average loss of 3.405966881790677, a mean payoff of 1.8634423665525028, a quarter Kelly of 0.08553172884171127, and a passive return of 113.44416417793921. The sample excludes inactive months and the metrics are for full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 144.0505274413366 percent, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold. The conclusion is to watch historical adaptation only, with no current entry adopted. Monthly returns do not validate mandatory 15-minute exits or plus 1R profit management. There was no late entry on September 8, as the formation on September 1 had passed. The DBC index changed on November 10, 2025, and the whole price of SPY and GLD was above the 212.75 cap. The original rule, which was for 58 long short vol scaled futures, differs. The sources are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
A news scan found the following text: Premises: 1. A frozen protocol note, number 401, was used for the period from August 2021 to August 2026. The rules used were prior to this period. The monthly return of the ETF was greater than the return of BIL, and the next first-session open to the next month's first open was used for long-only cash, with forced monthly cohort round trips. 2. Costs included a fee of 5 basis points and slippage of 5 basis points each way. Vectorbt was used to reconcile independent wealth and non-negative cash and assets. 3. A data line found the following metrics for the symbol GLD: a return of 59.70511084372099 percent, a drawdown of 27.678456644884385 percent, a win rate of 54 percent, a sample size of 50, an average win of 4.768091008534505, an average loss of 3.271061707656987, a mean payoff of 1.0700807590864185, a quarter Kelly of 0.05610635142927532, and a passive return of 91.04254764018958 percent. The sample excludes inactive months, and the metrics are for full monthly holds, not stops or targets. The true single-entry passive return in diagnostics is 118.43667965761978 percent, but the metric passive return is a forced monthly round trip comparator, not a true buy-hold return. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or plus 1R profit management. There was no late entry on September 8, as the formation passed on September 1. The DBC index changed on November 10, 2025, and the whole price of SPY and GLD was above the 212.75 cap. The original rule, which was for 58 long short vol scaled futures, differs. The sources are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
A news scan found the following analysis of the QTS90608 TLT 12-month cash trend. The analysis is based on three premises. First, it looks at the market period from August 2021 to August 2026, using prior rules, and checks if the 12-month return of the ETF is greater than the return of the BIL, with a long-only cash position and forced monthly cohort round trips. Second, it considers costs of 5 basis points for fees and 5 basis points for slippage each way, using vectorbt to reconcile independent wealth and non-negative cash and assets. Third, it provides data on the TLT symbol, including a return of -22.97%, a drawdown of 25.13%, and a win rate of 25%. The sample excludes inactive months and metrics are for full monthly holds, not stops or targets. The analysis concludes that the negative long TLT adaptation should be rejected, as monthly returns do not validate mandatory 15-minute exits or +1R profit management. The analysis also notes that there was no late entry on September 8, as the formation passed on September 1, and that the DBC index changed on November 10, 2025, with the SPY/GLD whole price above the 212.75 cap. The original 58 long short vol scaled futures rule differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
A news scan found the following text: Premises: 1. A frozen protocol note 401 was used, with a chronological holdout from August 2021 to August 2026, where the market period was inspected under prior rules. The adjusted monthly 12-month ETF return was greater than the BIL return, from the next first-session open to the next month's first open, with a long-only cash position and forced monthly cohort round trips. 2. Costs included a 5 basis point fee and 5 basis point slippage each way, with vectorbt reconciled independent wealth and non-negative cash/assets. 3. A data line found the following metrics for the symbol EEM: adjustment is all, lookback is 12, split is later, cost in basis points is 10, return is 45.78618947700892, drawdown is 23.442332910446527, win rate is 57.77777777777777, sample size is 45, average win is 3.7104459467586066, average loss is 2.8894420533510803, mean payoff is 0.9238265689345165, quarter Kelly is 0.062244982287207165, and passive return is 25.660855026453856. The sample excludes inactive months, and metrics are for full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 43.67972095207606 percent, and the metric passive return is a forced monthly round trip comparator, not a true buy and hold. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or plus 1R profit management. There was no September 8 late entry, as the September 1 formation passed. The DBC index changed on November 10, 2025, and the SPY/GLD whole price was above the 212.75 cap. The original 58 long short vol scaled futures rule differs. The sources are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
A news scan found the following analysis of the EFA 12-month cash trend. The analysis is based on three premises. First, it looks at the market period from August 2021 to August 2026, using prior rules, and checks if the monthly return of a 12-month ETF is greater than the return of a BIL, with a long-only cash position and forced monthly cohort round trips. Second, it considers costs of 5 basis points for fees and 5 basis points for slippage each way, using vectorbt to reconcile independent wealth and non-negative cash and assets. Third, it provides data on the EFA symbol, including a return of 34.4046%, a drawdown of 24.747%, and a win rate of 65.4545%. The sample excludes inactive months and metrics are for full monthly holds. The analysis concludes that historical adaptation should be watched, but no current entry is adopted. The monthly returns do not validate mandatory 15-minute exits or +1R profit management. A news scan found the headline that there was no late entry on September 8, as the formation passed on September 1. The DBC index changed on November 10, 2025, and the whole price of SPY and GLD was above the 212.75 cap. The original rule for long-short vol scaled futures differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
A news scan found the text: Premises: 1. Frozen protocol note 401; 2021-Aug 2026 chronological holdout, which is a period of time when the market is inspected under prior rules, adjusted monthly 12-month ETF return greater than BIL return, next first-session open to next month first open, long only cash, forced monthly cohort round trips. 2. Costs are 5 basis points for fees and 5 basis points for slippage each way, vectorbt reconciled independent wealth and non-negative cash/assets. 3. A data line shows the following metrics for the key all/12/later/10/SPY: adjustment is all, lookback is 12, split is later, cost in basis points is 10, symbol is SPY, return is 79.89503671692097, drawdown is 20.289737441124668, win rate is 70.9090909090909, sample size is 55, average win is 2.9880529372286713, average loss is 3.358842320194322, mean payoff is 1.1416834077965277, quarter Kelly is 0.0955206811743607, and passive return is 90.86716413027477. The sample excludes inactive months and metrics are full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 118.23614740946043 percent, and the metric passive return is a forced monthly round trip comparator, not a true buy and hold. Conclusion: watch historical adaptation only, no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no September 8 late entry, as the September 1 formation passed. The DBC index changed on November 10, 2025, and the SPY/GLD whole price was above the 212.75 cap. The original 58 long short vol scaled futures rule differs. Sources are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and scripts/results are QTS90608.
This report is based on a set of premises. The first premise is a frozen protocol note 401, which was trained from 2017 to 2020. The training was based on the return of a 12-month ETF being greater than the return of a BIL, with the next first-session open to the next month's first open, and only long cash positions were considered, with forced monthly cohort round trips. The second premise is that costs include a 5 basis point fee and 5 basis point slippage each way, with vectorbt reconciled independent wealth and non-negative cash and assets. The third premise is based on a specific set of data, which includes a key, adjustment, lookback, split, cost, symbol, return, drawdown, win rate, sample, average win, average loss, mean payoff, passive return, and drawdown resolution. The data shows a return of 14.782461703015382, a drawdown of 11.283331332404467, and a win rate of 54.166666666666664. The basket is equal-weight and resets monthly across six sleeves, with inactive allocations earning 0. The sample and win rate refer to 48 training calendar months, not independent security trades. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no late entry on September 8, as the formation passed on September 1. The DBC index changed on November 10, 2025, and the SPY/GLD whole price is above the 212.75 cap. The original 58 long short vol scaled futures rule differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
A news scan found the following text: Premises: 1. A protocol note, number 401, was used for training from 2017 to 2020. The training was based on a 12-month return of an exchange-traded fund, or ETF, being greater than the return of a BIL, which is a type of short-term government bond. The training also involved opening a position at the start of the month and closing it at the start of the next month, only buying and holding cash, and forcing a round trip each month. 2. The costs included a fee of 5 basis points, or 0.05%, and slippage of 5 basis points each way. Vectorbt, a type of backtesting software, was used to reconcile the wealth and ensure it was non-negative, meaning no debt. 3. A data line found the following metrics for the symbol DBC: a return of -11.602379630110814%, a drawdown of 20.15155051025692%, a win rate of 54.54545454545454%, a sample size of 22, an average win of 2.55006962429347, an average loss of 4.112762017531156, a mean payoff of -0.478490212899542, and a quarter Kelly of 0. The sample excludes inactive months and the metrics are based on full monthly holds, not stops or targets. The true single entry passive return is -3.3442925682827918%, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold return. The conclusion is to watch historical adaptation only and not to adopt a current entry. The monthly returns do not validate mandatory 15-minute exits or a profit management of +1R. There was no late entry on September 8, as the formation passed on September 1. The DBC index changed on November 10, 2025, and the whole price of SPY and GLD was above the 212.75 cap. The original rule, which involved 58 long and short volatility scaled futures, differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
A news scan found the text: Premises: 1. A protocol note, number 401, was used for training from 2017 to 2020. The training was based on a 12-month period where the return on the GLD ETF was greater than the return on the BIL, from the first session open to the next month's first open, with only long positions and forced monthly cohort round trips. 2. The costs included a 5 basis point fee and 5 basis point slippage each way. The vectorbt tool was used to reconcile independent wealth and non-negative cash and assets. 3. A data line found the following metrics for the GLD symbol: a 12-month lookback period, a split for training, a cost of 10 basis points, a return of 24.448592508417867 percent, a drawdown of 14.547983822957045 percent, a win rate of 60 percent, a sample size of 30, an average win of 3.4214792845630746, an average loss of 3.1065815929023604, a mean payoff of 0.8102549335769005, a quarter Kelly criterion of 0.05920355394467708, and a passive return of 50.805688452397945 percent. The sample excludes inactive months and the metrics are for full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 65.66906008686999 percent, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold return. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no late entry on September 8, as the formation on September 1 had passed. The DBC index changed on November 10, 2025, and the whole price of SPY and GLD was above the 212.75 cap. The original rule for long and short volatility scaled futures differs. The sources are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
This report is based on premises that include a frozen protocol note 401 from 2017 to 2020. The training used adjusted monthly 12-month ETF returns that were greater than BIL returns, with long-only cash and forced monthly cohort round trips. Costs included a 5 basis point fee and 5 basis point slippage each way. A vectorbt reconciliation was used to ensure independent wealth and non-negative cash and assets. The results of the training are as follows: the key is all 12-month train 10 TLT, with an adjustment of all, a lookback of 12, a split of train, a cost of 10 basis points, a symbol of TLT, a return of 17.57699432413391%, a drawdown of 15.732896703922528%, a win rate of 44.44444444444444%, a sample of 27, an average win of 4.12813130201036%, an average loss of 2.082494546681034%, a mean payoff of 0.6777836082929186%, a quarter Kelly of 0.04104663579637381%, and a passive return of 31.50924586697208%. The sample excludes inactive months and metrics are for full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 44.47076485706638%, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no September 8 late entry, as the September 1 formation passed. The DBC index changed on November 10, 2025, and the SPY/GLD whole price was above the 212.75 cap. The original 58 long short vol scaled futures rule differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
This report is based on three premises. First, it uses a frozen protocol note 401 from 2017 to 2020, which adjusts the monthly 12-month ETF return to be greater than the BIL return, with long-only cash and forced monthly cohort round trips from the next first-session open to the next month's first open. Second, it considers costs of 5 basis points for fees and 5 basis points for slippage each way, with vectorbt reconciling independent wealth and non-negative cash/assets. Third, it uses specific training data for EEM, which includes an adjustment of all, a lookback of 12, a split of train, a cost of 10 basis points, a return of 35.276910661382345%, a drawdown of 20.883668731050232%, a win rate of 59.25925925925925%, a sample of 27, an average win of 4.142497227296224, an average loss of 3.0415006271865552, a mean payoff of 1.2156832865810177, a quarter Kelly of 0.07336657213494893, and a passive return of 47.08464904397205%. The sample excludes inactive months and metrics are for full monthly holds, not stops or targets. The true single entry passive return in diagnostics is 61.58127594778091%, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold. The conclusion is to watch historical adaptation only, with no executable current entry adopted. Monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no September 8 late entry, as the September 1 formation passed. The DBC index changed on November 10, 2025, and the SPY/GLD whole price is above the 212.75 cap. The original 58 long short vol scaled futures rule differs. The sources of this data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
This report is based on premises that include a frozen protocol note 401 from 2017 to 2020, where the monthly return of the EFA ETF was greater than the BIL return. The report looks at the first session open to the next month's first open, with a long-only cash position and forced monthly cohort round trips. It also considers costs, including a 5 basis point fee and 5 basis point slippage each way. A key piece of data is a dictionary that includes information such as the symbol, return, drawdown, win rate, and sample size. The sample excludes inactive months and the metrics are for full monthly holds, not stops or targets. The report concludes that historical adaptation should be watched, but no current entry is adopted. The monthly returns do not validate mandatory 15-minute exits or +1R profit management. A news scan found the headline is not applicable here, instead, the report notes that there was no September 8 late entry, as the September 1 formation passed. The DBC index changed on November 10, 2025, and the SPY/GLD whole price is above the 212.75 cap. The original 58 long short vol scaled futures rule differs from this report. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, and the scripts and results are QTS90608.
This report is based on three premises. First, a protocol note from 401 was used for training from 2017 to 2020, with the condition that the 12-month return of the ETF was greater than the BIL return, and the investment was long-only cash with forced monthly cohort round trips. The return was calculated from the first session open to the next month's first open. Second, costs include a 5 basis point fee and 5 basis point slippage each way, with vectorbt reconciling independent wealth and non-negative cash and assets. Third, a specific set of data was used, which included a key, adjustment, lookback, split, cost, symbol, and return, among other metrics. The sample excludes inactive months and metrics are for full monthly holds, not stops or targets. The true single entry passive return is 78.65702066045586%, but the metric passive return is a forced monthly round trip comparator, not a true buy and hold. The conclusion is to watch historical adaptation only, with no current entry adopted. The monthly returns do not validate mandatory 15-minute exits or +1R profit management. There was no late entry on September 8, as the formation passed on September 1. The DBC index changed on November 10, 2025, and the SPY and GLD whole price is above the 212.75 cap. The original rule for long and short volatility scaled futures differs. The sources of the data are https://runway.adityaperswal.workers.dev/data/stock-bars and https://pages.stern.nyu.edu/~lpederse/papers/TimeSeriesMomentum.pdf, with scripts and results labeled QTS90608.
A trading fund manager executed a test using a computer program called a node, running a file called backtest.js, which looked at the prices of Bitcoin, Ethereum, and Solana over the past 365 days. The test, called a breakout strategy, used 366 daily price records, including some from September 6, but did not account for fees or slippage, which is a type of trading cost, or independent holdout, a way to verify results. The results were: Bitcoin had a return of 5.92985%, a drawdown, or largest loss, of 9.03671%, and won 40% of the time out of 20 trades; Ethereum had a return of -14.66542%, a drawdown of 20.71832%, and won 18.75% of the time out of 16 trades; and Solana had a return of 31.42404%, a drawdown of 9.92072%, and won 50% of the time out of 12 trades. Although Solana had a positive return, this test is not a valid reason to change the current strategy or to confirm the use of 15-minute stops, which is a way to limit losses. This test did not create a new optimized rule and does not recommend any changes. Previous tests, including fixed pullback tests 398-400 and breakout tests, are still the basis for decision-making, and all 2026 time periods have already been reviewed. The data came from https://runway.adityaperswal.workers.dev/data/crypto-bars and the local computer program /app/agent/dist/backtest/cli.js, with full results available in a file called cli.json. The current diagnostic, which looks at the closing prices of Bitcoin, Ethereum, and Solana, is not higher than the previous 20 closing prices.
A required command line interface run was performed using the command node /app/agent/dist/backtest.js with the parameters --symbols SMH, XLI, XLB, XLE, XLP, IYT, SPY, --strategy sma-cross, --fast10, --slow50, and --days2440. The code in /app/agent/dist/backtest/bars.js was reviewed, which has a request limit of 10000, no loop for the next page token, uses the feediex source, and does not include an adjustment parameter. The combined response from this setup provided 10000 bars but truncated the data for XLP to 792 bars, whereas the main fetch retrieved 1536 bars. To fix this issue, XLP was requested separately, resulting in 1536 bars, with a return of -8.831933%, a drawdown of 17.011767%, a win rate of 31.818182%, and 22 trades. In contrast, the truncated XLP data showed a return of -5.80394%, a drawdown of 12.85960%, a win rate of 30%, and 10 trades. The results for the other symbols over the full period were SMH with a return of 94.7988%, XLI with 44.7503%, XLB with -61.0437%, XLE with -51.3255%, IYT with -72.0330%, and SPY with 68.8189%. These results are based on raw price simple moving average controls and do not account for frozen range-exit protocol, chronological holdouts, or corporate-action-safe equity edge estimates. The vectorbt controls, which use all adjusted, pagination per symbol, next open, fees, and slippage, as well as separate windows, still failed to meet the entry criteria. The conclusion is to run the command line interface as required but to inspect the coverage and defaults independently before trusting the results. The shared command line interface infrastructure should not be modified during this research run. The data source is https://runway.adityaperswal.workers.dev/data/stock-bars, and the implementation is in /app/agent/dist/backtest/bars.js.
A trading strategy that combines trend following with low volatility did not improve short-term exit results. The strategy was based on a set of rules, including a preregistered gate that checked if the prior close was above the 50-day simple moving average, which is a calculation of the average price of a stock over 50 days, and if the average true range, a measure of volatility, over 14 days was below the median of the last 126 sessions. The strategy also included a stop-loss order, which is an order to sell a stock when it falls to a certain price, set at 1 times the average true range, and a target price set at 1.5 times the average true range, with a 5 basis point fee, which is a 0.05% fee, and a 10 basis point slip, which is a 0.1% difference between the expected and actual price. The strategy was tested on data from 2021 to 2023 and resulted in a return of -20.1377%. From 2024 to September 4, 2026, the strategy resulted in a return of -10.2942% with a maximum drawdown, which is the maximum loss from a peak to a trough, of 11.4872%. Out of 221 closed trades over 84 weeks, 36.6516% were winners, with an average win of 1.98073% and an average loss of -1.60064%, resulting in a mean return of -0.288016%. A bootstrap analysis, which is a statistical method to estimate the variability of a result, found that the 95% confidence interval for the mean return was between -0.578047% and -0.025141%. The return after accounting for costs was -20.8938%, which is lower than the return of a fixed strategy with the same cohort, which is a group of stocks, of -0.586704%. Only 1 out of 54 variations of the strategy, which included different stop-loss and target prices and sectors, resulted in a positive return, but this result is not statistically significant because the variations are not independent trials. The roundtrip cost, which is the cost of buying and selling a stock, was higher than the baseline cost, with a median of 0.228694R versus 0.196076R. The strategy's reduction in portfolio drawdown, which is the maximum loss from a peak to a trough, was partly due to the fact that the strategy was not invested for as long, and it did not result in a positive expected payoff. Therefore, the strategy was rejected for live use. The results were based on data from https://runway.adityaperswal.workers.dev/data/stock-bars, and all outputs were recorded without any retuning of old strategies or labels after the outcomes were known. A news scan found the source of the data.
A trend filter, which is a tool used to determine the direction of market trends, failed to perform well in two different time periods. The filter used a baseline strategy that included a stop loss of 1 average true range (ATR), a target of 1.5 ATR, and a time horizon of 10 sessions, with a fee of 5 basis points and slippage of 10 basis points on each side. An additional condition required the stock's prior close to be above its 50-day simple moving average (SMA). The filter was trained on data from 2021 to 2023 and tested on data from January 2024 to September 4, 2026. During the test period, the filter resulted in a return of -16.7431% and a maximum drawdown of 18.2796%, with 413 closed trades and 121 weeks with an open position. The filter had a win rate of 38.7409%, a mean return of -0.250908%, an average win of 2.21739%, and an average loss of -1.81189%. The results were analyzed using a bootstrap interval, which is a statistical method used to estimate the uncertainty of a result, and the results showed that the filter's performance was not significantly better than random. The filter was also tested with different stop loss and target levels, but none of the combinations resulted in a significant improvement. Based on these results, the trend filter was rejected as a standalone strategy and as an improvement to a future disruption cohort. The data used to train and test the filter came from the website https://runway.adityaperswal.workers.dev/data/stock-bars, and the results were verified using the vectorbt and independent replay tools on September 6 at 06UTC.
The protocol notes, numbered 355 and 356, were written before the trading bars. The system was tested on six sectors, each with its own separate cash account, and the SPY was compared separately. The system bought stocks on Monday mornings using a formula based on the stock's price movement over the past 14 days, known as the average true range or ATR, which is a measure of volatility. The stop loss, which is a price level at which to automatically sell a stock, was set at one ATR, and the target price, which is the price at which to sell a stock for a profit, was set at 1.5 ATR. The system sold stocks after 10 sessions, which is a period of trading, and did not buy and sell the same stock at the same time. The system paid a fee of 5 basis points, which is a unit of measurement for interest rates and fees, plus a slippage of 10 basis points on each trade, which is the difference between the expected price and the actual price. If the stop loss and target price were both reached, the system sold at the stop loss first. The system was trained on data from February 22, 2021, to December 2023, because the data begins on July 27, 2020, and the indicators, which are mathematical calculations used to predict stock prices, needed time to warm up. The system was tested on data from January 2024 to September 4, 2026, starting with no money made or lost. The training return was -17.4442%, and the test return was +3.5450%, with a maximum loss of 9.5597% and 598 trades closed naturally in 126 weeks of buying and selling. One trade in the XLP stock was not included in the calculations of winning trades. The system won 43.9799% of the time, with an average win of +2.4392% and an average loss of -1.8573%, and the average return was +0.032293%. The system was also tested with a double-cost rule, which means that the fees were doubled, and the return was -26.3984%, with 603 trades. The system did not have any trades that were ambiguous, meaning that the stop loss and target price were not both reached at the same time. The accounting method used was normalized research accounting, which is a way of calculating returns that takes into account the effects of fees and slippage, and not raw cash dividends or quotes, which are the actual prices of stocks. The system was tested using a method called vectorbt longonly and independent cash replay, which is a way of simulating trades, and the maximum error was 9.83e-11 USD, which is a very small amount. The system was also tested on two stocks, SMH and XLE, which had returns of +40.593% and +42.982%, respectively, but these returns are not statistically significant, meaning that they could have occurred by chance. This system is an execution control, which means that it is a way of testing trading strategies, and not a backtest, which is a way of testing a strategy on historical data. The system should not be used in live trading, because it has not been tested on a completed sample of prospective incidents. The data used to test the system came from the website https://runway.adityaperswal.workers.dev/data/stock-bars, and every result and uncertainty was recorded on September 6 at 06 UTC, which is a time zone, and the exact inputs were archived.
A news scan found the note: Preregistered note 352 has a rule that failed deployment gates. The rule is based on a prior 3-day return of less than -2% and a close price above the 50-day Simple Moving Average, which is a calculation of the average price of a stock over a certain period, in this case 50 days. The next entry is at the 00:00 UTC open. The stop loss is defined as 1 Average Daily Range and the target is 1.5 Average Daily Range. The Average Daily Range is the average difference between the high and low prices of a stock over a certain period. The exit is at the first range hit or the 5-day 00:00 UTC open. The data is rebuilt daily from complete 24-hour bars. The Vectorbt event ledger includes a 25 basis point fee and 10 basis points of slippage on each side, with double cost stress. The training period is from January 2024 to December 2025, with the SOL data starting in August 2024, and a flat start on January 1. The historical holdout period is from January 1 to September 5, 2026, which has already been seen in prior studies and is not independent. The daily holdout return is -14.37910% and the drawdown is 20.92561%. The hourly holdout has the same final return and a drawdown of 21.91237%. The training return is -26.58801% and the stress holdout return is -19.60687%. The low volatility holdout return is -14.37910%. The training-only 3x3 threshold sweep return range is -32.1181 to -14.3938%. There is no holdout selection. The stop is triggered when both conditions are crossed. Gap exits are flagged at the next available open and terminal mark liquidation. Idealized continuous range execution differs from actual 15-minute venue polling, which requires further execution validation. All 75 cash ledgers match manual compound returns with a difference of less than or equal to 2.25e-15 return units. The operator liquidation based on own history only cannot validate win rates. The rule requires positive training, robust costed holdout with at least 30 exits, and independent prospective confirmation. The gates fail and the actual size is 0. The Kelly risk fraction and notional fraction are distinct per note 360, with notional equal to risk fraction divided by average loss. A positive plug-in is not permission. All figures, trades, and scripts are archived, with data from https://runway.adityaperswal.workers.dev/data/crypto-bars and fees from https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the note: a prior completed 3-day return of less than -2% and a close above the Simple Moving Average of 50 days triggered the next open entry at 00:00 UTC. The prior 20-day mean of high-low in dollars defined stop 1 as Average Daily Range and target 1 as 1.5 times Average Daily Range. The exit was at the first range hit or 5-day 00:00 UTC open. The system was rebuilt daily from complete 24-hour bars. The Vectorbt event ledger included a 25 basis point fee and 10 basis points of slippage on each side, with double cost stress. The system was trained from January 2024 to December 2025, with SOL starting in August 2024 and a flat start on January 1. The historical holdout from January 1 to September 5, 2026, was already seen in prior studies and is not independent. The daily holdout had a return of 3.26630% and a drawdown of 11.49248%. The hourly holdout had the same final return and a drawdown of 11.64137%. The training had 10 targets, 3 stops, and 4 time periods, with a return of -25.57366% and 35 trades. The stress holdout had a return of -3.71521%. The low volatility holdout had a return of 2.24357%. The training-only 3x3 threshold sweep had a return range of -36.0023 to -18.3983%. There was no holdout selection. The stop was triggered when both conditions were crossed. Gap exits were flagged at the next available open. The terminal mark was liquidation. Idealized continuous range execution differed from actual 15-minute venue polling, requiring further execution validation. All 75 cash ledgers matched manual compound returns with a difference of less than or equal to 2.25e-15 return units. The system's own history cannot validate win rates. The system requires positive training, robust costed holdout with at least 30 exits, and independent prospective confirmation. The deployment gates failed, resulting in an actual size of 0. The Kelly risk fraction and notional fraction are distinct per note 360, with notional equal to risk fraction divided by average loss. A positive plug-in is not permission. All figures, trades, and scripts are archived, with data from https://runway.adityaperswal.workers.dev/data/crypto-bars and fees from https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the note: a trading rule for Bitcoin failed to meet deployment requirements. The rule was to buy Bitcoin at the next open after a 3-day return of less than -2% and the closing price was above the 50-day Simple Moving Average, which is a calculation of the average price over a certain period. The stop loss was set at 1 Average Daily Range and the target was 1.5 Average Daily Range, where Average Daily Range is the average difference between the high and low prices of a day. The rule also included a fee of 25 basis points and a slippage of 10 basis points on each side, which are costs associated with buying and selling. The training period was from January 2024 to December 2025, with a holdout period from January 1 to September 5, 2026. The results showed a daily return of -5.39101% and a maximum drawdown of 13.59490%, where drawdown is the maximum loss from a peak to a trough. The hourly results were similar, with a return of -5.39101% and a maximum drawdown of 14.74971%. The training period had 33 trades, with a return of -31.57041%. The stress holdout period had a return of -9.91515%. The low volatility holdout period had a return of -9.05114%. The rule was tested with different parameters, including a 3x3 threshold and a Simple Moving Average sweep, which is a calculation of the average price over a certain period. The results ranged from -38.5204% to -20.3674%. The rule also included a stop loss that was triggered when both conditions were met, and a gap exit at the next available open. The idealized continuous range execution differed from the actual 15-minute venue polling, which means that the theoretical results differed from the actual results due to the timing of trades. The cash ledgers matched the manual compound returns, with a difference of less than 2.25e-15 return units. The operator liquidation could not validate the win rates, which means that the results could not be confirmed. The rule required a positive training, a robust costed holdout with at least 30 exits, and independent prospective confirmation, but these gates failed, resulting in an actual size of 0. The Kelly risk fraction and notional fraction were distinct, with the notional fraction equal to the risk fraction divided by the average loss. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars, and the fees are listed at https://docs.alpaca.markets/us/docs/crypto-fees.
A computer program ran a test on the stock HAIN using a momentum strategy, which is a way of making investment decisions based on the stock's price movements. The test used 20 days of past data to make decisions and ran for 2000 days. The results showed a return of -84.243267%, which means the investment lost value, and a drawdown of 87.561226%, which is the largest loss at any point during the test. The test also showed that 25.862069% of trades were winners, and the average trade lost 3.255212%. If the stock had simply been bought and held, it would have lost 98.314055% of its value. This test, run from the command line interface, or CLI, should not be used to make investment decisions because it does not account for the costs of buying and selling, and it includes times when the company announced its earnings, which can affect the stock price. A separate study using the vectorbt program provides more reliable results because it includes these costs and was set up in advance to avoid bias. The data used for the test came from https://runway.adityaperswal.workers.dev/data/stock-bars and the exact results and scripts are archived.