Notes

A news scan found the headline 'SOL/USD C91106 trade attribution daily highvol'. This is a computed report from September 11 at 06:09 UTC. It is a post-hoc partition of a frozen baseline, not a filtered strategy return, for the years 2024 to 2025 for training and 2026 for holdout. The report credits fees of 25 basis points and a 10 basis points slip or trail. An unknown warmup is excluded only from labels, but all trades are retained. Compound refers to a subset payoff product. The report includes the following data: for the symbol SOL, with a daily time frame, in the train period, with the feature highvol, there are two sets of data. One set has a value of false, with 5 occurrences, a mean of -4.0819946601, a compound of -20.9324901622, 20 wins, a training rv median of 0.7496917283, and 0 unknown regime excluded. The other set has a value of true, with 4 occurrences, a mean of 14.660123137, a compound of 63.4468339794, 75 wins, the same training rv median, and also 0 unknown regime excluded. For the reused holdout period, there are also two sets of data. One set has a value of false, with 6 occurrences, a mean of -0.6385208045, a compound of -8.2163568127, 16.6666666667 wins, the same training rv median, and 0 unknown regime excluded. The other set has a value of true, with 0 occurrences, a mean of null, a compound of 0, null wins, the same training rv median, and 0 unknown regime excluded. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the trade attribution report for SOL/USD on September 11. The report was computed at 06:09 UTC. It is a post-hoc partition of the frozen baseline, which is a strategy return that is not filtered. The report used data from 2024 to 2025 for training and reused 2026 as a holdout. The fees were 25 basis points and the slip or trail was 10 basis points. The unknown warm-up was excluded only from labels, but all trades were retained. The compound is a subset payoff product. The report includes the following data: for the daily trend of SOL during the training period when the trend was false, the mean was null and the compound was 0. When the trend was true, the mean was 5.1010973474 and the compound was 32.6534928596. For the reused holdout period, when the trend was false, the mean was null and the compound was 0. When the trend was true, the mean was -0.6385208045 and the compound was -8.2163568127. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The next day's log returns are described, but do not include traded profit or loss. A signal was given before the close, which was higher than the 50-day Simple Moving Average, and the 20-day realized volatility was above the 2024-25 median. A regression was done using HAC, with a result of 7.2026, and the holdout was reused. The results are as follows: for SOL, during the reused holdout period, when the trend was 0 and volatility was not high, the mean next day log percentage change was 0.196407599, the median next day log percentage change was 0.0798035012, and the standard deviation of the next day percentage change was 3.1647690702. When the trend was 0 and volatility was high, the mean next day log percentage change was -0.4737013182, the median next day log percentage change was -0.680298901, and the standard deviation of the next day percentage change was 4.5916042325. When the trend was 1 and volatility was not high, the mean next day log percentage change was -0.1965354017, the median next day log percentage change was -0.2528037645, and the standard deviation of the next day percentage change was 2.8264531736. When the trend was 1 and volatility was high, the mean next day log percentage change was -0.4339448668, the median next day log percentage change was -0.4339448668, and the standard deviation of the next day percentage change was null. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The next day's log returns are described, without including any trading profits or losses. A signal was given before the close, which was higher than the 50-day Simple Moving Average, and the 20-day realized volatility was above the median for 2024-2025. A regression analysis was done using HAC 7, and the 2026 data was reused for testing. The results are as follows: for the SOL symbol, the training period had four scenarios, including trend and high volatility indicators, with the following statistics: for trend 0 and high volatility 0, the mean next day log percentage change was -0.2920207091, the median next day log percentage change was -0.3542176529, and the standard deviation of the next day percentage change was 4.2454390523; for trend 0 and high volatility 1, the mean was -0.2413303254, the median was -0.075885677, and the standard deviation was 5.0895613793; for trend 1 and high volatility 0, the mean was -0.1047212071, the median was -0.1881409722, and the standard deviation was 3.7928761054; and for trend 1 and high volatility 1, the mean was 0.4576812225, the median was 0.2549267481, and the standard deviation was 4.241040166. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Computed on September 11 at 06:09 UTC, this is a post-hoc partition of a frozen baseline, not a filtered strategy return, for the years 2024 and 2025 used for training and 2026 used for holdout, with 25 basis points credited for fees and 10 basis points for lip or trail, a term that refers to the difference between the expected and actual price of a trade. The unknown warmup, which is a period of time when a model is learning, is excluded only from labels, and all trades are retained. Compound refers to a subset payoff product, which is a way of calculating the return on investment. The data includes the following information for the symbol ETH, which is a type of cryptocurrency: a list of trade attributions with the following details: symbol, time frame, which is hourly, period, feature, which is slope, value, number of trades, mean, compound, win percentage, training return variance median, and whether the unknown regime is excluded. The list includes: a trade with a false value, 6 trades, a mean of 0.5718351736, a compound of -1.9967127934, a win percentage of 33.3333333333, a training return variance median of 0.6430540547, and the unknown regime excluded; a trade with a true value, 12 trades, a mean of 1.6890209276, a compound of 13.0557155153, a win percentage of 50, a training return variance median of 0.6430540547, and the unknown regime excluded; a trade with a false value in the reused holdout period, 3 trades, a mean of -0.2505320839, a compound of -1.0113592858, a win percentage of 33.3333333333, a training return variance median of 0.6430540547, and the unknown regime not excluded; and a trade with a true value in the reused holdout period, 3 trades, a mean of 6.0004219667, a compound of 18.4382753046, a win percentage of 100, a training return variance median of 0.6430540547, and the unknown regime not excluded. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline is not available, but a data line was found. The data is about a trade attribution for ETH/USD on September 11, computed at 06:09 UTC. This is a post-hoc partition of a frozen baseline, which is a previously established standard, and it is not filtered by a specific strategy return. The data is from 2024 to 2025 for training and 2026 for holdout, which means it is a period of time used to test a strategy. A 25 basis point, or 0.25%, fee was credited and a 10 basis point, or 0.10%, slip or trail was applied. A basis point is a unit of measurement for interest rates or fees. The unknown warmup, which refers to a period of time when a system is starting up, was excluded only from labels, which are categories or tags, and all trades were retained. The compound, which refers to a subset payoff product, is a calculation of the total return. The data includes the following information: a list of trade attributions for ETH, which is a symbol for Ethereum, with the following details: the time frame is hourly, the period is either train or reused holdout, the feature is highvol, which refers to high volatility, and the value is either true or false. The number of trades, n, ranges from 2 to 10, the mean return ranges from -0.3842640728 to 4.5045494485, the compound return ranges from -5.4242079408 to 40.8924117166, the win percentage ranges from 44.4444444444 to 75, and the training rv median, which refers to the median return of a strategy, is 0.6430540547. The unknown regime excluded is 0 for all trades. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91106 trade attribution hourly trend'. This is a data line showing the results of a trading strategy for Ethereum, a type of digital currency, against the US dollar. The data was computed on September 11 at 06:09 UTC. It is a post-hoc partition, which means it is a review of past data, of a frozen baseline, which is a standard for comparison, and it is not filtered, meaning all data is included. The strategy return, which is the profit or loss of the trades, is for the years 2024 and 2025 for training, and 2026 for testing, also known as the holdout period. The fees, which are the costs of the trades, are 25 basis points, which is 0.25%, and the slippage, which is the difference between the expected and actual price of a trade, is 10 basis points, which is 0.10%. The data is divided into two groups: when the trend, which is the direction of the price, is false, meaning the price is not going up or down in a clear direction, and when it is true, meaning the price is going up or down in a clear direction. The results are as follows: when the trend is false, the mean return, which is the average profit or loss, is -6.14% and the compound return, which is the total profit or loss over time, is -23.20%. When the trend is true, the mean return is 3.45% and the compound return is 44.27%. The win rate, which is the percentage of profitable trades, is 25% when the trend is false and 50% when the trend is true. The training rv median, which is a measure of the risk of the trades, is 0.64. The unknown regime, which is a period of time when the market is not following a clear pattern, is excluded from the labels, which are the categories used to classify the data, but all trades are retained. The compound return is also calculated as a subset payoff product, which means it is the total profit or loss of a subset of the trades. The specific results are: for the training period, when the trend is false, there were 4 trades with a mean return of -6.14% and a compound return of -23.20%, and when the trend is true, there were 14 trades with a mean return of 3.45% and a compound return of 44.27%. For the holdout period, when the trend is false, there were 2 trades with a mean return of 1.97% and a compound return of 3.87%, and when the trend is true, there were 4 trades with a mean return of 3.33% and a compound return of 12.87%. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'ETH/USD C91106 trade attribution daily slope'. This is a report about a trading strategy for Ethereum, a type of cryptocurrency, traded against the US dollar. The report was computed on September 11 at 06:09 UTC. It looks at how well the strategy worked, using data from 2024 and 2025 to train the strategy and 2026 to test it. The strategy had fees of 25 basis points, which is 0.25 percent, and a 10 basis points slip, which is 0.10 percent. A basis point is a unit of measurement for interest rates and fees. The report also mentions a 'compound' payoff, which refers to the total return on investment over time, taking into account the effect of reinvesting earnings. The report lists the results of the strategy in different scenarios, including when the daily slope feature was used and when it was not. The slope feature refers to the rate of change of the price of Ethereum over time. The results are as follows: for the training period, when the slope feature was not used, the mean return was -1.8488518143 percent, and when it was used, the mean return was 0.3244966505 percent. For the test period, when the slope feature was not used, the mean return was -8.3317019201 percent, and when it was used, the mean return was 3.6928601266 percent. The report also includes other metrics, such as the number of trades, the win rate, and the median return. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the trade attribution report for ETH/USD on September 11. The report was computed at 06:09 UTC. It is a post-hoc partition of the frozen baseline, which is a strategy return that is not filtered. The report used data from 2024 to 2025 for training and reused 2026 as a holdout. A 25 basis point fee was credited and a 10 basis point slip or trail was applied. A basis point is a unit of measurement, one hundredth of a percent. The unknown warmup was excluded only from labels, but all trades were retained. Compound refers to a subset payoff product, which is a way of calculating returns. The report includes the following data: for ETH, the daily period, train feature highvol, which means high volatility, had a value of false, with 7 observations, a mean of -0.4739287355, a compound of -8.6968783326, a win rate of 42.8571428571 percent, a training rv median of 0.6430540547, and unknown regime excluded was 0. For ETH, the daily period, train feature highvol had a value of true, with 8 observations, a mean of 2.5312746489, a compound of 11.7477159529, a win rate of 37.5 percent, a training rv median of 0.6430540547, and unknown regime excluded was 0. For ETH, the daily period, reused holdout feature highvol had a value of false, with 3 observations, a mean of -5.037063053, a compound of -14.5234345745, a win rate of 0 percent, a training rv median of 0.6430540547, and unknown regime excluded was 0. For ETH, the daily period, reused holdout feature highvol had a value of true, with 2 observations, a mean of -1.2490981741, a compound of -3.6372909448, a win rate of 50 percent, a training rv median of 0.6430540547, and unknown regime excluded was 0. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following trade attribution data for ETH/USD on September 11, computed at 06:09 UTC. The data is based on a post-hoc partition of a frozen baseline, with a strategy return that is not filtered. The data is from 2024 to 2025 for training and 2026 for holdout, with credited fees of 25 basis points and a 10 basis points lip or trail. A basis point is a unit of measurement, equal to one hundredth of a percent. The unknown warmup is excluded only from labels, and all trades are retained. Compound refers to a subset payoff product. The data includes the following information: for the daily trend feature of ETH during the training period, when the trend value was false, the mean was -11.288723631, the compound was -11.288723631, the win was 0, the training rv median was 0.6430540547, and the unknown regime excluded was 1. When the trend value was true, the mean was 0.549098683, the compound was -5.0135207713, the win was 38.4615384615, the training rv median was 0.6430540547, and the unknown regime excluded was 1. For the reused holdout period, when the trend value was false, there was no data. When the trend value was true, the mean was -3.5218771014, the compound was -17.6324659487, the win was 20, the training rv median was 0.6430540547, and the unknown regime excluded was 0. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The next day's log returns are described, without including any traded profit or loss. A signal was given when the price before the close was higher than the 50-day Simple Moving Average, and the 20-day realized volatility was above the median for 2024-2025. A regression was done using HAC 7 for 2026, reusing the holdout data. The results are as follows: for ETH, during the reused holdout period, when the trend was 0 and volatility was 0, the mean next day log percentage change was -0.1574549453, the median next day log percentage change was -0.0375827968, and the standard deviation of the next day percentage change was 3.1039070376, based on 52 data points. When the trend was 0 and volatility was 1, the mean next day log percentage change was -0.5307911433, the median next day log percentage change was -0.260760704, and the standard deviation of the next day percentage change was 4.2653475468, based on 63 data points. When the trend was 1 and volatility was 0, the mean next day log percentage change was 0.1096703764, the median next day log percentage change was 0.0972674632, and the standard deviation of the next day percentage change was 2.8540954767, based on 106 data points. When the trend was 1 and volatility was 1, the mean next day log percentage change was 0.2965287123, the median next day log percentage change was 0.1771579996, and the standard deviation of the next day percentage change was 3.0373409041, based on 31 data points. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The results describe the next day's log returns, without including any trading profits or losses. A signal was given before the market closed, indicating that the price was higher than the 50-day Simple Moving Average, which is a calculation of the average price over the last 50 days, and the 20-day Relative Volatility, a measure of price fluctuations, was above the median for 2024 and 2025. A regression analysis, specifically a Heteroskedasticity-Adjusted Covariance, was used, with a value of 7.2026, and the results from 2026 were reused for testing. The following data was found for Ethereum, or ETH: when the trend was low and volatility was low, the average next day log percentage change was -0.2413823626, the median was 0.0520176651, and the standard deviation was 4.0020313672, based on 170 observations; when the trend was low and volatility was high, the average was -0.1257751485, the median was -0.0899421006, and the standard deviation was 3.7041705184, based on 205 observations; when the trend was high and volatility was low, the average was 0.1393187837, the median was 0.1482321681, and the standard deviation was 3.4304463397, based on 164 observations; and when the trend was high and volatility was high, the average was 0.3153091031, the median was 0.0360431008, and the standard deviation was 3.5907440259, based on 142 observations. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the trade attribution report for BTC/USD on September 11. The report was computed at 06:09 UTC. It is a post-hoc partition of the frozen baseline, which is not a filtered strategy return, for the period from 2024 to 2025 for training and 2026 for holdout. The fees were 25 basis points and the slip or trail was 10 basis points. The unknown warm-up was excluded only from labels, but all trades were retained. Compound refers to a subset payoff product. The report includes the following data: for BTC, with an hourly time frame, in the train period, the feature was slope, with a value of false, the mean was -0.9366771407, the compound was -9.4117978143, there were 25 wins, the training rv median was 0.4364544418, and 2 unknown regimes were excluded. For BTC, with an hourly time frame, in the train period, the feature was slope, with a value of true, the mean was -0.6027855486, the compound was -8.3132491664, there were 33.3333333333 percent wins, the training rv median was 0.4364544418, and 2 unknown regimes were excluded. For BTC, with an hourly time frame, in the reused holdout period, the feature was slope, with a value of false, the mean was -2.5358690131, the compound was -5.0075074081, there were 0 wins, the training rv median was 0.4364544418, and 0 unknown regimes were excluded. For BTC, with an hourly time frame, in the reused holdout period, the feature was slope, with a value of true, the mean was 3.3709189718, the compound was 13.3097470256, there were 75 percent wins, the training rv median was 0.4364544418, and 0 unknown regimes were excluded. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following trade attribution data for BTC/USD on September 11, computed at 06:09 UTC. This is a post-hoc partition of a frozen baseline, which is a previously established standard, using a strategy return that was not filtered. The data is from 2024 to 2025 for training and 2026 for holdout, which means it was used to test the strategy. A 25 basis point fee was credited and a 10 basis point slip or trail was applied. A basis point is a unit of measurement for interest rates or fees, equal to 0.01%. The unknown warmup, which refers to a period of time when a model is learning, was excluded only from labels, but all trades were retained. The compound, which refers to the result of combining multiple values, is a subset payoff product. The data includes the following information: a list of trade results for BTC, with a time frame of hourly, and a feature of highvol, which refers to high volume. The list includes the symbol, time frame, period, feature, value, number of trades, mean, compound, win rate, training return value median, and unknown regime excluded. The list contains four items: for the train period with highvol false, the mean is 4.8829110585, the compound is 45.5781713165, and the win rate is 44.4444444444; for the train period with highvol true, the mean is -1.5853159553, the compound is -17.1866405903, and the win rate is 30; for the reused holdout period with highvol false, the mean is 1.4019896435, the compound is 7.6357530493, and the win rate is 50; and for the reused holdout period with highvol true, there are no trades, so the mean and win rate are null. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91106 trade attribution hourly trend'. This is a data line. It shows results from September 11, computed at 06:09 UTC. The results are from a post-hoc partition of a frozen baseline, which is a set of data used as a standard for comparison, and a not filtered strategy return, which means the results have not been selected or changed. The data is from 2024 to 2025 for training and 2026 for holdout, which is a period of time when the system is not used to make decisions. There were 25 basis points, which is a unit of measurement for interest rates or fees, credited fees and 10 basis points lip or trail, which refers to the difference between the price at which a security is bought or sold and the price at which it is offered. The unknown warmup, which is a period of time when a system is starting up, was excluded only from labels, which are used to identify the data, and all trades were retained. The compound, which refers to the result of combining multiple things, is a subset payoff product, which means it is the result of multiplying multiple payoffs together. The data includes the following information: [{"symbol":"BTC","tf":"hourly","period":"train","feature":"trend","value":false,"n":4,"mean":-3.2875709471,"compound":-13.218430589,"win":25,"training_rv_median":0.4364544418,"unknown_regime_excluded":2},{"symbol":"BTC","tf":"hourly","period":"train","feature":"trend","value":true,"n":13,"mean":0.0178305173,"compound":-4.2914529128,"win":30.7692307692,"training_rv_median":0.4364544418,"unknown_regime_excluded":2},{"symbol":"BTC","tf":"hourly","period":"reused_holdout","feature":"trend","value":false,"n":1,"mean":-2.6228738134,"compound":-2.6228738134,"win":0,"training_rv_median":0.4364544418,"unknown_regime_excluded":0},{"symbol":"BTC","tf":"hourly","period":"reused_holdout","feature":"trend","value":true,"n":5,"mean":2.2069623349,"compound":10.5349451811,"win":60,"training_rv_median":0.4364544418,"unknown_regime_excluded":0}]. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91106 trade attribution daily slope'. This is a data line showing information about a trading strategy for Bitcoin, denoted by the symbol 'BTC', in US dollars, 'USD'. The data was computed on September 11 at 06:09 UTC. It is a post-hoc partition of a frozen baseline, which means it is a review of past data, not filtered by strategy return. The data is from 2024 to 2025 for training and 2026 for holdout, which is a period of time used to test the strategy. A 25 basis point fee was credited and a 10 basis point slip or trail was applied. A basis point is a unit of measurement, one hundredth of a percent. The unknown warm-up, which is a period of time at the start of the data, was excluded only from labels, but all trades were retained. The compound refers to a subset payoff product, which is a way of calculating the return on investment. The data shows the following results for the slope feature: for the train period with a false value, the mean was -3.8738487402, the compound was -21.75676328, the win rate was 16.6666666667 percent, the training return volatility median was 0.4364544418, and the unknown regime excluded was 1. For the train period with a true value, the mean was 1.0680561584, the compound was 6.2262298661, the win rate was 54.5454545455 percent, the training return volatility median was 0.4364544418, and the unknown regime excluded was 1. For the reused holdout period with a false value, the mean was -5.7320116811, the compound was -16.3642976478, the win rate was 0 percent, the training return volatility median was 0.4364544418, and the unknown regime excluded was 0. For the reused holdout period with a true value, the mean was 5.2260021194, the compound was 15.554398783, the win rate was 66.6666666667 percent, the training return volatility median was 0.4364544418, and the unknown regime excluded was 0. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91106 trade attribution daily highvol'. This is a report about a trading strategy for Bitcoin, denoted by the symbol BTC, in US dollars, denoted by USD. The report was computed on September 11 at 06:09 UTC. It looks at the performance of a trading strategy, which is a set of rules for buying and selling, from 2024 to 2025, and then applies those rules to 2026 without changing them. The strategy's return, or profit, is calculated after subtracting 25 basis points, or 0.25%, in fees and 10 basis points, or 0.10%, in costs for buying and selling. A basis point is a unit of measurement for interest rates and fees, equal to 0.01%. The report excludes a period known as the 'unknown warmup', which is a time when the strategy is learning and not making actual trades. The compound return is a measure of how much the strategy's returns have multiplied together over time. The report lists the results for different scenarios, including when the feature 'highvol', or high volatility, is true or false. Volatility refers to how much the price of Bitcoin changes over time. The results are as follows: for the training period from 2024 to 2025, when highvol was false, the strategy had a mean return of 1.2171556572%, and when highvol was true, the mean return was -0.0467036914%. For the reused holdout period in 2026, when highvol was false, the mean return was 0.9330969289%, and when highvol was true, the mean return was -2.6252082004%. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91106 trade attribution daily trend'. This is a computed report from September 11 at 06:09 UTC. It is a post-hoc partition of a frozen baseline, which is a set of data used as a reference point, and it is not filtered by a specific strategy return. The report covers the period from 2024 to 2025 for training and 2026 for holdout, which means it is testing the strategy on a separate set of data. The report credits fees of 25 basis points, which is a measure of the cost of trading, and it also considers a 10 basis points lip or trail, which refers to the difference between the expected and actual price of a trade. The unknown warm-up period, which is the initial period of a trading strategy, is excluded only from the labels, but all trades are retained. The compound refers to a subset payoff product, which means it is a measure of the overall return of a set of trades. The report includes the following data: for the daily trend feature of Bitcoin, the value is false, with 0 occurrences, a mean of null, a compound of 0, and a win rate of null, during the training period. For the same feature, the value is true, with 17 occurrences, a mean of -0.6761455705, a compound of -16.8851595071, and a win rate of 41.1764705882, during the training period. For the reused holdout period, the value is false, with 0 occurrences, a mean of null, a compound of 0, and a win rate of null. For the same feature, the value is true, with 6 occurrences, a mean of -0.2530047809, a compound of -3.355266979, and a win rate of 33.3333333333, during the reused holdout period. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The results are about the next day's log returns, which are descriptive and not about actual trades or profits. A signal was given before the market closed, which was higher than the 50-day Simple Moving Average, and the 20-day Relative Volatility was above the median for 2024-2025. A regression analysis was done using HAC 7 for 2026, reusing a holdout set. The results are as follows: for BTC, during the reused holdout period, when the trend was 0 and volatility was 0, the mean next day log percentage change was -0.3349343318 and the median was -0.0371878756, based on 61 days, with a standard deviation of 2.276267162. When the trend was 0 and volatility was 1, the mean was -0.0590420751 and the median was -0.2716527659, based on 63 days, with a standard deviation of 3.3316305541. When the trend was 1 and volatility was 0, the mean was 0.2090561172 and the median was 0.0465000164, based on 98 days, with a standard deviation of 1.9675392969. When the trend was 1 and volatility was 1, the mean was -0.3716946555 and the median was -0.4291939307, based on 30 days, with a standard deviation of 1.8900469043. https://runway.adityaperswal.workers.dev/data/crypto-bars

A computation was made on September 11 at 06:09 UTC. The results describe the next day's log returns, without including filled orders or traded profits and losses. A signal was given before the close, which was higher than the 50-day Simple Moving Average, and the 20-day Relative Volatility was above the median for 2024 and 2025. A regression analysis was done using HAC, with a holdout from 2026 that was reused. The results are as follows: for BTC, during the training period, when the trend was 0 and volatility was low, the mean next day log percentage change was -0.1199999101 and the median was -0.0057064763, with a standard deviation of 2.398598903. When the trend was 0 and volatility was high, the mean was 0.0793610817 and the median was 0.0024956304, with a standard deviation of 2.7675649671. When the trend was 1 and volatility was low, the mean was 0.1221119431 and the median was 0.006760381, with a standard deviation of 2.062201902. When the trend was 1 and volatility was high, the mean was 0.1582606844 and the median was 0.1578482225, with a standard deviation of 2.7882405982. https://runway.adityaperswal.workers.dev/data/crypto-bars

A test of the GLD adapter model from January 3, 2023, to August 31, 2026, showed a return of 37.26 percent, with a maximum drawdown of 9.46 percent. The compound annual growth rate, or CAGR, which is a measure of an investment's annual growth rate, was 9.08 percent. The Sharpe ratio, which is a measure of risk-adjusted return, was 1.19. The annualized volatility, or ann_vol, which is a measure of the investment's risk, was 7.56 percent. There were 918 trading sessions and 44 months in the test period. The win rate, which is the percentage of winning trades, was 52.27 percent. A total of 44 trades were made, with 52.27 percent of months being positive. The average trade return was 1.16 percent, with an average win of 4.65 percent and an average loss of 2.66 percent. The mean exposure, which is the average amount of capital invested, was 0.33. The test also showed a vectorbt max error, which is a measure of the model's accuracy, of 3.64e-12. A total of 104 orders were made. The test started on January 3, 2023, and ended on August 31, 2026. This was a measurement taken on September 11, using a frozen 2230 SIP, with a fee of 1 basis point, and fractional normalization, with daily proxy. https://runway.adityaperswal.workers.dev/data/stock-bars

A trading model called an adapter for the symbol GLD had the following results: it made 8.594624456479094 percent return, with a drawdown of 12.863200967894684 percent, and a compound annual growth rate, or CAGR, of 2.0826844150378054 percent. The Sharpe ratio, which is a measure of risk, was 0.2938675447944952, and the annual volatility was 8.14649785275333 percent. Over 1008 sessions and 48 months, the model had a win rate of 47.91666666666667 percent, with 48 trades, and 47.91666666666667 percent of months being positive. The mean trade was 0.39022782352469954 percent, with an average win of 3.394105559086327 percent and an average loss of 2.373339693191998 percent. The model had a mean exposure of 0.3765074530558196, and a vectorbt max error of 3.637978807091713e-12, with 108 orders. The trading period was from 2019-01-02 to 2022-12-30. This was measured on Sep 11, with a frozen value of 2230, using a system called SIP, with a fee of 1 basis point, or 1/100th of a percent, and fractional normalization, with daily proxy data. https://runway.adityaperswal.workers.dev/data/stock-bars

A test of the GLD adapter model, which is a type of computer program that helps make investment decisions, found the following results: the symbol being traded was GLD, the model being used was an adapter, and the sample being tested was a test sample. The test had a slippage of 5 basis points, which is a measure of the difference between the expected price of a trade and the actual price, and a c multiplier of 1, which is a factor that affects the model's decisions. The return on investment was 43.29%, the drawdown, which is the largest decline in value, was 8.48%, and the compound annual growth rate, or CAGR, was 10.38%. The Sharpe ratio, which is a measure of risk-adjusted return, was 1.36, and the annual volatility, which is a measure of how much the value of the investment changes, was 7.48%. The test ran for 918 sessions, or periods of trading, over 44 months, and had a win rate of 52.27%. There were 44 trades made, and 52.27% of months had a positive return. The average return per trade was 1.33%, the average win was 4.80%, and the average loss was 2.47%. The mean exposure, which is the average amount of money invested, was 0.33. The test also found a maximum error of 3.64e-12, which is a very small number, and there were 104 orders, or trades, made. The test started on January 3, 2023, and ended on August 31, 2026. This test was measured on September 11, and used a frozen dataset from 2230, with a system of SIP, or systematic investment plan, and a fee of 1 basis point, with fractional and normalized daily proxy data. https://runway.adityaperswal.workers.dev/data/stock-bars

A performance report for the GLD symbol using an adapter model on a train sample shows the following results: the return was 13.220439532381523 percent, the drawdown was 13.176694279509316 percent, and the Compound Annual Growth Rate, or CAGR, was 3.1528446739564897 percent. The Sharpe ratio, a measure of risk-adjusted return, was 0.420648287937519. The annualized volatility, or ann_vol, was 8.176226203185141 percent. Over 1008 sessions and 48 months, the win rate was 47.91666666666667 percent, with 48 trades made. The percentage of positive months was also 47.91666666666667 percent. The average trade return was 0.5172105268009635 percent, with an average win of 3.5788922277903708 percent and an average loss of 2.2995366381092914 percent. The mean exposure, or the average amount of capital used, was 0.3899464018699038. There were 109 orders placed, starting on January 2, 2019, and ending on December 30, 2022. A news scan found the headline 'Sep11measurement' and other notes mentioned 'frozen2230', 'SIP/all', 'fee1bp', 'fractionalnormalized', and 'dailyproxy'. https://runway.adityaperswal.workers.dev/data/stock-bars

A trading fund manager's notes for the symbol GLD show a constant model test with the following results: the return was 117.6517761730966, the drawdown was 23.90681734597957, and the Compound Annual Growth Rate, or CAGR, which is a measure of investment return, was 23.799542450330158 percent. The Sharpe ratio, which is a measure of risk, was 1.2823909760535954, and the annual volatility, or ann_vol, which is a measure of investment risk, was 17.911838952174026 percent. There were 918 trading sessions, or N_sessions, and 44 months, or N_months, with a win rate of 100 percent and one trade. The average monthly return was positive 65.9090909090909 percent of the time. The mean trade return was 120.50830707094494 percent, and the average win return was also 120.50830707094494 percent. The average loss return was not applicable, or null. The mean exposure, which is the average amount of capital used, was 0.8917665178284222. The vectorbt_max_error, which is a measure of the difference between expected and actual results, was 0. There were 45 orders, and the test started on 2023-01-03 and ended on 2026-08-31. This was a Sep 11 measurement with a frozen 2230, SIP, or Systematic Investment Plan, all with a fee of 1 basis point, or 1bp, and fractional normalized and daily proxy. https://runway.adityaperswal.workers.dev/data/stock-bars