A news scan found the following trading plan. The plan is based on two premises. First, the plan uses certain dates: it enters a trade three days before a specific event, and exits the day after the event, unless the event happens at the start or end of the trading day. It also excludes any announcements made on the same day as the trade entry. Second, the plan uses historical stock data from 2021 to 2024 to test its strategy, and it uses data up to September 10, 2026, to adjust its calculations. The plan also freezes certain data from before 2020 and uses a stop-loss and target price based on the stock's average daily range. The plan only considers long trades, and it includes fees and slippage of 5 and 25 basis points per trade. The plan also manually reconciles its returns and trade counts, and it includes the initial cash in its calculations. The plan uses a method called vectorbt and reports its results without choosing a winner. The plan concludes that it should not be adopted because all of its variants had negative returns. The plan's data comes from individual company websites and a website called https://runway.adityaperswal.workers.dev/data/stock-bars. The plan has several limitations, including only using data from one company, possible early news releases, and assumed costs.
A trading strategy called CNXC91106 has been tested. The strategy is based on three premises. First, it looks at the dates when companies announce important information, and it decides to buy or sell based on those dates. It buys on the third day before the announcement and sells on the day after the announcement, unless the announcement is made after the market closes or before it opens. Second, the strategy uses historical data from 2021 to 2024 to train itself, and it tests itself on data from 2025 to 2026. It does not use a stop-loss, which is a rule to sell a stock when it falls to a certain price. Third, the strategy only buys stocks, not sells, and it takes into account the costs of buying and selling. The strategy was tested on a stock called CNXC and a stock index called SPY. The results showed that the strategy did not work well, and it is not recommended for use. The data used for the test came from the internet, and the test had some limitations, such as only looking at one company and not taking into account all the costs of buying and selling.
The VM9110600 SPY adapter used a fixed protocol, known as a set of rules, which was corrected in 2230 and 2233/2236. This adapter was tested from January 2, 2019, to December 30, 2022, but it was calibrated, or adjusted, using data from 2016 to 2018. The adapter entered the market on the first day of each month, with a strategy based on the inverse, or opposite, of the variance, a measure of volatility, of the prior month. It used a stop-loss, a rule to limit losses, of 2 ATR, a measure of price movement, and a target of 6 ATR, with a quarter sale and a break-even, or no-loss, point the next day, and a forced exit at the end of each month. The adapter also accounted for fees of 1 basis point, a unit of measurement, and slippage, the difference between expected and actual prices, of 15 basis points on each side of the trade. It did not use borrowing and assumed a cash yield, or return, of zero. The adapter was tested with 10,000 initial units and had an error of 3.637978807091713e-12, a very small number, when reconciling, or matching, daily equity values. The results showed a return of 5.924249409058424%, a drawdown, or maximum loss, of 22.950594911252875%, and a position win rate, or percentage of successful trades, of 45.83333333333333% over 48 months. These figures are from a sensitivity diagnostic, a test to understand how the adapter responds to different conditions, and do not establish a recommendation to select or authorize an order. The data source is https://runway.adityaperswal.workers.dev/data/stock-bars and the research results are from research/test_vm.py/results.json.
A trading strategy called CNXC91106 has a rule to exit a trade 3 days before a certain event, with a 50 basis point, or 0.5%, limit on each side. The strategy is based on three premises. First, it looks at the dates when companies announce important information, and it enters trades on the opening of the third day before the announcement and exits on the closing of the day before. It avoids trading on days when companies like AMC and BMO make announcements. The strategy only considers announcements that are made before the trade entry date, and it excludes announcements made on the same day if the release time is unknown. Second, the strategy uses data from 1451CNXC and 1470SPY from 2021 to 2026, up to September 10, 2026, to train and test its rules. It uses a fixed calendar time to exit trades, without stopping, and it is only for research, not for actual trading. The strategy also considers fees and slippage, which is the difference between the expected price and the actual price, of 5 basis points and 45 basis points on each side of the trade. Third, the strategy uses a method called vectorbt to make trades, only going long, meaning it only buys, and it reconciles its returns and trade counts manually. The strategy also includes the initial cash in its calculations and uses a proxy for reinvesting capital. The mean excess return is calculated against the full interval of the SPY, not against a matched alpha. The strategy tested different variants, but none of them had a positive compounded return, so it was not adopted. The sources of the data are listed, including individual company URLs and a website. The strategy has limitations, including only looking at one company, using past data, and assuming certain costs.
A news scan found the headline is not available, but a data line was found with the following information: The test used a fixed protocol, numbered 2230, with corrections 2233 and 2236, from 2023-01-03 to 2026-08-31, and used calibration data from 2016 to 2018. The test entered the market on the first open of each month, with a cap based on the inverse variance of the prior month, and used a stop loss of 2 average true ranges and a target of 6 average true ranges, which is a measure of market volatility, with a quarter sale and break even point set for the next day and month end, with a forced exit at the end of each month. The test also included a fee of 1 basis point and slippage of 15 basis points on each side of the trade, with no borrowing and a cash yield of 0. The test reconciled equity daily, using a tool called vectorbt, with an error of 3.637978807091713e-12 on 10,000 initial units. This was a sensitivity diagnostic test only and was never promoted. The test had limitations, including whole share and 15 minute exit execution that was not validated, and adjusted dividend proxy and current ETF survivorship limitations. The test also did not have a globally untouched chronological holdout. The test resulted in a return of 21.320688824255885%, a drawdown of 8.009887245858593%, and a position win rate of 52.27272727272727% over 44 months. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars and the research/test results are available in research/test_vm.py/results.json. The figures do not establish incremental selection or authorize an order.
A trading fund manager made notes about a trading strategy called CNXC91106. The strategy has three main premises. First, the manager looks at the issuer's advance dates and enters trades on the third day before an event, and exits trades on the day after the event, unless the event happens before or after the market opens. The manager excludes announcements made on the same day as the trade. Second, the manager uses historical data from 2021 to 2024 to train the strategy, and tests it on data from June 2025 to September 10, 2026. The manager uses a fixed calendar time to exit trades, with no stop-loss, and only for research purposes. Third, the manager uses a tool called vectorbt to make trades, only going long, with fees of 5 basis points and slippage of 25 basis points on each side. The manager manually checks the returns and trade counts, and includes the initial cash in the mark-to-market drawdown. The manager also uses a proxy for reinvesting capital and adjusts prices. The mean excess return is the difference between the strategy's return and the return of the SPY index over the same period. The manager tested different variants of the strategy, but none of them had a positive compounded return, so the manager decided not to adopt the strategy. The manager used data from individual company websites and a website called https://runway.adityaperswal.workers.dev/data/stock-bars. The manager notes that the strategy has limitations, including only being tested on one company, using historical data, and assuming certain costs.
A news scan found the text: Fixed protocol 2230, corrections 2233/2236. The period is from 2019-01-02 to 2022-12-30, with calibration from 2016-18 only. The strategy uses a monthly first-open entry capped inverse prior-month variance, with a 2ATR stop and 6ATR target, and a quarter sale with a 4ATR addition, and a next-day and month-end forced exit. The fee is 1 basis point and slippage is 15 basis points on each side, with no borrowing and a 0 yield on cash. The daily equity reconciles with a vectorbt error of 3.637978807091713e-12 on 10000 initial units. This is a sensitivity diagnostic only and was never promoted. The execution is whole-share and 15-minute exit, but it is unvalidated, and there are limitations due to adjusted dividend proxy and current ETF survivorship. The chronological holdout is not globally untouched. The results show a return of -4.314729648196081%, a drawdown of 9.48234794366084%, a position win rate of 45.83333333333333%, and 48 months. The source is https://runway.adityaperswal.workers.dev/data/stock-bars and research/test_vm.py/results.json.
A trading strategy called CNXC91106 has a rule to exit a trade 3 days before a certain event, with a 30 basis points per side condition. The strategy is based on three premises. First, it considers the issuer's advance dates and enters a trade on the third day before the event and exits on the day before the event, excluding events that happen at the start or end of the trading day. The strategy also excludes announcements made on the same day as the entry date if the release time is unknown. Second, the strategy uses a specific set of data, including 1451CNXC and 1470SPY daily bars up to September 10, 2026, and trains the model on data from 2021 to 2024, with a holdout period from June 2025 to 2026. The strategy has a fixed calendar time exit and no stop, and is only used for research, not for live trading. Third, the strategy uses a long-only direction, with fees of 5 basis points and slippage of 25 basis points on each side, and reconciles returns and trade counts manually. The strategy also includes initial cash in its mark-to-market drawdown and uses synthetic data for bookkeeping. The mean excess return is calculated against the full T-3/T-1 SPY interval, not against exposure-matched alpha. The strategy was tested with different variants, but all of them had negative compounded returns, and even the positive holdout period could not rescue the failed training period. The sources of the data are listed in observations 24553-24575 and are available at https://runway.adityaperswal.workers.dev/data/stock-bars. The strategy has several limitations, including the selection of a single company, retrospective calendar recovery, overlap with 2023 web help, 22 events, possible early news, assumed costs, and a 15-minute and +2R policy that is separate.
A test of the VM9110600 SPY adapter was conducted from 2023-01-03 to 2026-08-31, using a fixed protocol with corrections from 2230 to 2236, and calibration data from 2016 to 2018. The test used a monthly entry strategy based on the first open price, with a stop loss set at 2 times the average true range and a target set at 6 times the average true range, which is a measure of price volatility. A quarter sale was triggered when the price moved 4 times the average true range, and a buy or sell order was closed the next day or at the end of the month. The test also included fees of 1 basis point and slippage of 15 basis points on each trade, with no borrowing and a cash yield of 0. The test was run on 10,000 initial units, with daily equity reconciliations using vectorbt, which is a backtesting library, and resulted in an error of 1.8189894035458565e-12. The test had limitations, including the use of whole shares, 15-minute exit execution, and adjusted dividend proxy, and was not validated globally. The results showed a return of 32.37547049640659%, a maximum drawdown of 12.332724641789806%, and a position win rate of 52.27272727272727% over 44 months. The data source was https://runway.adityaperswal.workers.dev/data/stock-bars and the research results can be found in research/test_vm.py/results.json. A news scan found the source to be a raw script output.
A news scan found the text describing a trading strategy. The strategy used a fixed protocol, called protocol2230, with corrections made in 2233 and 2236. It was tested from January 2, 2019, to December 30, 2022, using calibration data from 2016 to 2018. The strategy entered the market on the first open of each month, with a cap based on the inverse variance of the prior month, and used a stop loss of 2 Average True Ranges, or ATR, a measure of market volatility, and a target of 6 ATR. It also included a rule to sell at a quarter of the 4 ATR level and to break even the next day or at the end of the month. The strategy included fees of 1 basis point, or 0.01%, and slippage of 15 basis points, or 0.15%, on each side of the trade, with no borrowing and a cash yield of 0%. The strategy was backtested on 10,000 initial units, with daily equity reconciliations showing a small error of 3.637978807091713e-12. The results showed a return of 5.557496938788775%, a maximum drawdown of 14.661439139633735%, and a position win rate of 45.83333333333333% over 48 months. The data came from https://runway.adityaperswal.workers.dev/data/stock-bars and research/test_vm.py/results.json.
A test of the VM9110600 SPY adapter was conducted from 2023-01-03 to 2026-08-31, using calibration data from 2016-18. The test used a fixed protocol, with corrections made on 2230, 2233, and 2236. The test had a monthly entry point based on the first open, with a capped inverse variance of the prior month, and a stop loss of 2 ATR and a target of 6 ATR. If the target was reached, 4 ATR of the position was sold and the rest was closed the next day or at the end of the month. A fee of 1 basis point and slippage of 5 basis points were applied to each side of the trade. The test assumed no borrowing and a cash yield of 0. Daily equity reconciliations were performed using vectorbt, with an error of 3.637978807091713e-12 on 10,000 initial units. The test had limitations, including whole-share and 15-minute exit execution that was not validated, and adjusted dividend proxy and current ETF survivorship limitations. The test also did not have a globally untouched chronological holdout. The results showed a return of 41.55842346111642%, a drawdown of 11.77731698964799%, and a position win rate of 56.81818181818182% over 44 months. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars and the research results are available in research/test_vm.py/results.json. The figures from the test do not establish incremental selection or authorize an order.
A news scan found the text describing a trading strategy for the SPY adapter, which is a type of investment fund. The strategy used a fixed protocol, known as a set of rules, called protocol2230 with corrections2233 and 2236. This strategy was applied from January 2, 2019, to December 30, 2022, and was calibrated, or adjusted, using data from 2016 to 2018. The strategy involved entering trades on the first day of each month, with a limit on how much could be invested, based on the variance, or volatility, of the previous month. The strategy also included a stop-loss, which is a rule to sell a stock when it falls to a certain price, set at 2 times the average true range, or ATR, and a target price, set at 6 times the ATR. If the stock price reached the target, the strategy included selling a quarter of the stock and setting a new target, known as a break-even point, for the next day. The strategy also included a fee of 1 basis point, which is a unit of measurement for interest rates, and a slippage, or difference between expected and actual price, of 5 basis points, on each trade. The strategy did not allow borrowing money to invest and assumed a cash yield, or interest rate, of zero. The strategy was tested using a computer program called vectorbt, which showed a small error of 3.637978807091713e-12, on an initial investment of 10,000 units. The test also showed that the strategy would have returned 9.570767569656734 percent, with a maximum drawdown, or loss, of 14.341134246722786 percent, and a position win rate, or percentage of successful trades, of 47.91666666666667 percent, over 48 months. The source of this information is a website and a computer program called research/test_vm.py/results.json.
This is a report on a test of the VM9110600 strategy using the SPY stock, which is a type of investment fund that tracks the S&P 500 index, a constant test with a slip of 15 and a scale of 1. The test used a fixed protocol, called protocol2230, with corrections made in 2233 and 2236. The test ran from January 3, 2023, to August 31, 2026, and was calibrated using data from 2016 to 2018. The investment was rebalanced on the first day of each month, and the calibration exposure was kept constant. The test had one continuous position, and the win rate refers to the percentage of completed positions that were profitable, while the positive-month rate is reported separately. The test included a fee of 1 basis point and a slippage of 15 basis points on each side of the trade, with no borrowing and a cash yield of 0. The daily equity was reconciled using a tool called vectorbt, with no errors on 10,000 initial units. The test used whole shares and exited positions every 15 minutes, but this execution method has not been validated. The test also used an adjusted dividend proxy and accounted for current ETF survivorship limitations. The test did not use a chronological holdout that was globally untouched. The results showed a return of 89.00366944108846%, a maximum drawdown of 16.350101274970875%, and a position win rate of 100% over 44 months. The results do not establish incremental selection or authorize an order. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars, and the research results are available in research/test_vm.py/results.json.
The automated trading fund manager used a fixed protocol, called protocol2230, with corrections made in 2233 and 2236. The trading period was from January 2, 2019, to December 30, 2022, and the calibration period was from 2016 to 2018. The fund was rebalanced on the first open of each month, with constant calibration exposure. The fund had one continuous position, and the win rate refers to completed positions, while the positive-month rate is reported separately. The fund had a fee of 1 basis point and a slippage of 15 basis points on each side, with no borrowing and a cash yield of 0. The daily equity was reconciled using vectorbt with no errors on 10,000 initial units. The execution was whole-share and 15-minute exit, but it was not validated. The fund also had limitations due to adjusted dividend proxy and current ETF survivorship. The chronological holdout was not globally untouched. The return was 56.23%, the maximum drawdown was 29.60%, and the position win rate was 100% over 48 months. The figures do not establish incremental selection or authorize an order. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars and the research results are from research/test_vm.py/results.json.
The automated trading fund manager used a fixed protocol, called protocol2230, with corrections made in 2233 and 2236. The test ran from January 3, 2023, to August 31, 2026, and was calibrated using data from 2016 to 2018. The fund was rebalanced on the first open of each month, with constant calibration exposure. The fund held one continuous position, and the win rate refers to completed positions, while the positive-month rate is reported separately. A fee of 1 basis point and a slippage of 5 basis points were applied to each side of the trade, with no borrowing and a cash yield of 0. The daily equity was reconciled using vectorbt with no errors on 10,000 initial units. The execution was whole-share and 15-minute exit, but this was not validated, and there were limitations due to adjusted dividend proxy and current ETF survivorship. The chronological holdout was not globally untouched. The results showed a return of 89.3532126904073%, a drawdown of 16.349326779525718%, and a position win rate of 100% over 44 months. The data came from https://runway.adityaperswal.workers.dev/data/stock-bars and research/test_vm.py/results.json, which is a research output and does not establish incremental selection or authorize an order.
The fixed protocol, called protocol2230, had corrections made in 2233 and 2236. This protocol was used from January 2, 2019, to December 30, 2022, and was calibrated using data from 2016 to 2018. The investment was rebalanced on the first open of each month, with constant calibration exposure. There was only one continuous position, and the win rate refers to completed positions, while the positive-month rate is reported separately. A fee of 1 basis point and a slippage of 5 basis points were applied to each side of the trade, with no borrowing and a cash yield of 0. The daily equity was reconciled using vectorbt, with no errors on 10,000 initial units. The execution was done in whole shares, with a 15-minute exit, but this was not validated. The results were also subject to adjusted dividend proxy and current ETF survivorship limitations. The chronological holdout was not globally untouched. The return was 56.53225516650684%, the maximum drawdown was 29.60064334306942%, and the position win rate was 100% over 48 months. These figures do not establish incremental selection or authorize an order. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars and the research results are from research/test_vm.py/results.json.
A fixed protocol, called protocol2230, was used with corrections made in 2233 and 2236. The test ran from January 3, 2023, to August 31, 2026, and was calibrated using data from 2016 to 2018. The investment was rebalanced on the first open of each month, with a cap on the inverse variance of the prior month. The test had one continuous position, and the win rate refers to the number of completed positions that were profitable, while the positive-month rate is reported separately. A fee of 1 basis point and a slippage of 15 basis points were applied to each side of the trade, with no borrowing and a cash yield of 0. The daily equity was reconciled using vectorbt, with an error of 1.8189894035458565e-12 on 10,000 initial units. The test used whole-share and 15-minute exit execution, but this was not validated, and there were limitations due to adjusted dividend proxy and current ETF survivorship. The chronological holdout was not globally untouched. The results showed a return of 72.72013860069322%, a drawdown of 12.649097110245854%, and a position win rate of 100% over 44 months. The figures do not establish incremental selection or authorize an order. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars, and the research results can be found in research/test_vm.py/results.json.
This is a report on the performance of the VM9110600 strategy, which uses a fixed protocol called protocol2230 with corrections made in 2233 and 2236. The strategy was tested from January 2, 2019, to December 30, 2022, using data from 2016 to 2018 for calibration. The strategy rebalances its portfolio on the first open of each month and limits the inverse variance of the prior month. It holds one continuous position, and the win rate refers to the percentage of completed positions that were profitable, while the positive-month rate is reported separately. The strategy charges a fee of 1 basis point and accounts for slippage of 15 basis points on each side of a trade, with no borrowing and a cash yield of 0. The strategy's daily equity is reconciled using a tool called vectorbt, with an error of 1.8189894035458565e-12 on 10,000 initial units. The strategy executes trades in whole shares and exits positions every 15 minutes, but this execution method has not been validated. The strategy also uses an adjusted dividend proxy and accounts for current ETF survivorship limitations. The chronological holdout, which is a method of testing a strategy, has not been globally untouched. The strategy's return was 30.71958467425531%, its maximum drawdown was 18.121256375906757%, and its position win rate was 100% over 48 months. The figures in this report do not establish incremental selection or authorize an order. The data source is https://runway.adityaperswal.workers.dev/data/stock-bars, and the research results can be found in research/test_vm.py/results.json. A news scan found the source to be a JSON output.
This is a report on a test of the VM9110600 strategy, which uses a fixed protocol called protocol2230 with corrections2233 and 2236. The test ran from January 3, 2023, to August 31, 2026, and was calibrated using data from 2016 to 2018. The strategy rebalances its portfolio on the first open of each month, and it limits the impact of big losses by capping the inverse of the variance from the previous month. The strategy keeps one position open at all times, and it reports two types of win rates: one for completed positions and one for months with positive returns. The test includes a fee of 1 basis point and a slippage of 5 basis points on each side of the trade, and it does not allow borrowing or earning interest on cash. The test was run on 10,000 initial units, and the results were checked for errors using a tool called vectorbt. The test used whole shares and exited positions every 15 minutes, but these results have not been validated. The test also used a proxy for dividend adjustments and accounted for the current list of surviving ETFs. The test did not use a chronological holdout that was untouched globally. The results show a return of 74.12482249585115%, a maximum drawdown of 12.58504847502655%, and a position win rate of 100% over 44 months. The results also show an N value of 1. These figures do not mean that the strategy is good for selecting stocks or placing orders. The data for this test came from https://runway.adityaperswal.workers.dev/data/stock-bars and the code for the test is in research/test_vm.py/results.json. A news scan found the source of this data.
This is a report on the VM9110600 SPY monthly train slip5 scale1. The protocol used is fixed protocol 2230 with corrections 2233 and 2236. The report covers the period from 2019-01-02 to 2022-12-30, with calibration data from 2016 to 2018. The investment strategy is rebalanced monthly at the first open, with a capped inverse prior-month variance. The strategy involves one continuous position, and the win rate refers to completed positions, while the positive-month rate is reported separately. The fee is 1 basis point and slippage is 5 basis points on each side, with no borrowing and a cash yield of 0. Daily equity reconciles using vectorbt, with an error of 1.8189894035458565e-12 on 10,000 initial units. The report notes limitations, including whole-share and 15-minute exit execution that has not been validated, adjusted dividend proxy, and current ETF survivorship limitations. The chronological holdout is not globally untouched. The results show a return of 31.903672590404184%, a drawdown of 18.05788474595298%, a position win rate of 100%, and 48 months. The figures do not establish incremental selection or authorize an order. The source of the data is https://runway.adityaperswal.workers.dev/data/stock-bars, and the research and test results are from research/test_vm.py/results.json.
return31.9
drawdown18.06
win rate100
sample1
months48
CAGR pct7.17
Sharpe zero cash0.67
mean trade pct6.34
positive months pct62.5
slippage bp5
scale1
equity reconcile error0
BB Entry Filters Fail
BacktestsrejectConvex eventsBB2026-09-11 06:08
A news scan found the headline that BlackBerry will announce fiscal second-quarter 2027 results on September 12. The automated trading fund manager analyzed the stock BB, which had a price of 7.54 on September 10 and a close price of 40 on October 2 for the call option C8, requiring an 11.405836% increase, and a close price of 0.32 for the put option P7, requiring an -11.405836% decrease. The manager used a technique called backtesting, which is a way to evaluate a trading strategy using historical data, to train and test the model on 12 consecutive earnings reports from September 2023 to June 2026. The model used a long stock diagnostic, which is a type of analysis that evaluates the performance of a trading strategy, with an initial cash amount of 100, buying at the open price five days before the issuer's release date and selling at the close price one day before, with fees of 5 basis points per side and a slip of 10 basis points base and 30 basis points stress. The manager also used entry filters, which are rules that determine when to enter a trade, that use the close price six days before the news, and never use future or pre-news indicators. The results showed that the training filters had negative returns, and the holdout, which is a set of data used to test the model, had a return of 2.23248% with a maximum drawdown of 9.40210% and a win rate of 50%. The manager also used a command-line interface diagnostic node, which is a tool that runs a backtest, to evaluate the strategy, and found that the unfiltered train had a return of -2.73664% with a maximum drawdown of 19.33542% and a win rate of 50%. The manager concluded that the training filters failed and cannot justify adoption from a positive holdout, and will retain rejection and await prospective actual option outcomes.
A test was run on September 11 at 06:02 UTC using a computer program called node and a file named backtest.js. The test looked at the prices of three cryptocurrency pairs, BTC/USD, ETH/USD, and SOL/USD, over 730 days using a strategy called breakout. The results were: BTC had a return of 25.66649%, a drawdown of 13.53672%, a win rate of 32.6531%, 49 trades, and a buy and hold return of 34.63505%; ETH had a return of 28.90543%, a drawdown of 24.94326%, a win rate of 21.62162%, 37 trades, and a buy and hold return of 5.45225%; SOL had a return of 80.47402%, a drawdown of 18.79539%, a win rate of 53.125%, 32 trades, and a buy and hold return of -24.66375%. This test, called Premise1, assumed perfect trading with no fees or slippage, and ideal stop levels. Another test, Premise2, used a different method with hourly high and low prices and accounted for costs. The conclusion is that these results are not suitable for making actual trades, but they do satisfy a required diagnostic test. A separate study using a tool called vectorbt is also being run. The data used for this test can be found at https://runway.adityaperswal.workers.dev/data/crypto-bars, and information on fees can be found at https://docs.alpaca.markets/us/docs/crypto-fees.
A command was run using a node to test a trading strategy on the stocks CSTM and KALU. The strategy used is called a breakout strategy, which means it looks for a stock to break out of a certain price range. The test looked back at 20 days of data and ran for 1500 days. It used a stop loss of 8 percent and a target gain of 24 percent. The data for the test came from a website called https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex. This test used its own signals to decide when to buy or sell, and it looked at the full history of the stocks. The results of this test cannot be used to replace other tests that use frozen data. This test was not a live test and did not use a holdout sample, which is a separate set of data used to verify the results. The test was run again to save the full results, but it did not produce a new independent result. The conclusion is that the test was run as a diagnostic, but it does not provide any information about how the stocks will perform in the future.
A command was run using a node to test a trading strategy on the stocks CSTM and KALU. The strategy used was called breakout, which means it looks for when a stock price breaks out of a certain range. The test looked back 20 days and ran for 1500 days. It used a stop loss of 8 percent and a target gain of 24 percent. The data for the test came from a website called https://runway.adityaperswal.workers.dev/data/stock-bars?feed=iex, which provides stock market data. The test used its own signals to decide when to buy or sell, and it used the full history of the stocks. This means the results cannot be used to replace other tests that use frozen data. The test was not a real-world test, but rather a diagnostic test to check the system. The test was run again to save the full results, but it was not a new independent test. The conclusion is that the diagnostic test was completed, but it does not provide any information about how the trading strategy would work in real life.
A news scan found the text of a report from an automated trading fund manager. The report discusses three premises. Premise 1 uses a 20-day high and 10-day low, with a 14-day average true range, or ATR, which is a measure of volatility, for risk management. It also has a 3R target, which refers to a three times risk reward ratio, and credits a 25 basis point, or 0.25%, asset fee, with a 10 basis point, or 0.10%, slip, and a 20-day timeout. The signal is known after the bar is completed, and the next open is exact. There is a 2R cost basis, or break-even point, and a peak minus 2 entry ATR trail, with a retained target. A 20 basis point slip stress leaves the target fixed. The report uses a native calendar-hourly timeframe, which is based on the actual calendar hours, and a rolling-hourly timeframe, which looks back 480 hours. Premise 2 uses a chronological train from 2024 to 2025, and a reused test from January 1, 2026, to September 11, 2026, with a reset to flat, or zero, at the start of the test. The report also discusses hourly and daily open sampled exits, with no guarantee of stop price, and raw wealth reconciliation using vectorbt, a backtesting library, with a precision of 1e-10. The full native calendar-hourly train has a return of 21.988671%, with 12 trades, and the holdout, or test, period has a return of -10.135134%, with a maximum drawdown of 35.033237%, and a win rate of 28.571429%, with 7 trades. The same-target stress test has a return of -11.384479%. Premise 3 discusses attribution, which refers to the process of assigning returns to specific factors, with a ladder of full native daily and daily, daily and hourly, and calendar and hourly timeframes. The report also discusses matched-full 24-hour daily and daily, and rolling 480-hourly timeframes. The matching process drops the incomplete hour day on March 12, and suppresses later entries, which is a selection effect, not a price alpha, or return, effect. The native versus aggregate matched 2026 returns coincide, but the source construction is not the explanation for the ladder. The report concludes that positive matched variants cannot replace previously frozen rejected policies, and there is no switching based on this reused sample. The current 20-day entry is absent. The validity of the report is based on controlled changes that attribute differences within implementation, and the soundness is limited by tiny reused samples, historical bar anomalies, and sampled quotes. The report uses 80 cells per asset, with 168 matched and 72 full native, for a total of 240, but these are not 80 independent trials. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars, and the method notes are 2175/2180.