Notes

A news scan found the headline 'SOL/USD C91204 required CLI 980day breakout diagnostic'. The given data is about the SOL/USD trading pair. It has 748 bars of data and 36 trades. The return percentage is 72.15172735196003 percent. The buy and hold percentage is -35.31319554848966 percent. The maximum drawdown percentage is 18.795394376588376 percent. The win rate is 52.77777777777778 percent. The average trade percentage is 1.6935012342155107 percent. This data is based on a same close/prior close strategy, which means it looks at the closing price of the current and previous periods. There are no explicit fees or holdout periods considered. The data includes a partial final daily bar, but it does not follow a pullback policy or a deployable edge strategy, which refers to a strategy that can be used in real-world trading. node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback20 --days980 ; https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: a trading strategy for ETH/USD had 981 bars, 55 trades, a 40.808012459287 percent return, and a 12.116247945584746 percent buy and hold return. The maximum drawdown was 24.943256404302126 percent, the win rate was 30.909090909090907 percent, and the average trade was 0.7597218620104566 percent. This was based on a same close/prior close strategy, which is a method of making trading decisions, with no explicit fees or holdout, and includes a partial final daily bar, meaning it uses the last available daily data even if it is not complete. It does not use a pullback policy, which is a strategy of waiting for a price to drop before buying, or a deployable edge, which refers to a trading advantage that can be used in practice. node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback20 --days980 ; https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: {"symbol":"BTC/USD","bars":981,"trades":67,"returnPct":46.460751153240395,"buyAndHoldPct":75.66159304266115,"maxDrawdownPct":14.692428391826324,"winRate":34.32835820895522,"avgTradePct":0.6443234597984014}. This is a raw script output showing the results of a trading strategy for the BTC/USD symbol, which is a currency pair representing Bitcoin and the US dollar. The data includes the number of bars, or time periods, which is 981, and the number of trades, which is 67. The return percentage is 46.46%, and the buy and hold percentage, which is a strategy where an investment is held for a long time, is 75.66%. The maximum drawdown percentage, which is the maximum loss from a peak to a trough, is 14.69%. The win rate, or the percentage of successful trades, is 34.33%, and the average trade percentage is 0.64%. The strategy used is called same close/prior close, which means it looks at the closing price of the current and previous time periods. There are no explicit fees or holdout periods, and it includes a partial final daily bar. It does not use a pullback policy or a deployable edge, which are advanced trading techniques. node /app/agent/dist/backtest.js --symbols BTC/USD,ETH/USD,SOL/USD --strategy breakout --lookback20 --days980 ; https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'Complete 24 hour days only; absolute relative OHLC difference threshold 1bp'. The data shows that for the SOL/USD currency pair in the year 2026, there were 253 days with no difference of 1 basis point or more. The maximum difference percentage was 0, and the worst date was January 1, 2026, at 00:00:00+00:00, with a difference in the opening price, which was 124.5275. The hourly aggregation was also 124.5275. The cause of the difference is not identified, and it could be due to differing definitions or feed vintages, or a bad bar, but no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'Complete 24 hour days only; absolute relative OHLC difference threshold 1bp'. The data shows that for the SOL/USD pair in the year 2025, there were 363 days with a difference of 1 basis point or more, with a maximum field difference percentage of 10.14046426221924% on the worst date of 2025-12-02 00:00:00+00:00, affecting the low field, with a raw value of 138.327 and an hour aggregation of 124.3. The cause of the difference is not identified and may be due to differing definitions or feed vintages, or a bad bar, with no silent repair made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'Complete 24 hour days only; absolute relative OHLC difference threshold 1bp.' A data line was also found with the following information: the symbol is SOL, the year is 2024, there are 127 days, and 125 of them have a difference of 1 basis point, which is a unit of measurement, or 0.01%, for the difference between two prices. The maximum difference percentage is 0.11001863066208983%. The worst date was September 29, 2024, at 00:00:00, and the worst field was the close, or c, which is the price at the end of a trading period. The raw price was 158.6095 and the hourly aggregation was 158.435. The cause of the differences is not identified, and it could be due to differing definitions or feed vintages, which refers to the version of the data, or a bad bar, which is an error in the data. No silent repair was done, meaning that no automatic correction was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Complete 24 hour days only, with an absolute relative OHLC difference threshold of 1 basis point. The data shows that for ETH in the year 2026, out of 253 days, there were 0 days with a difference of 1 basis point or more, with a maximum field difference percentage of 0, and the worst date was 2026-01-01 00:00:00+00:00, with the worst field being the opening price, and a raw value of 2966.39, and an hour aggregation of 2966.39. The differences may be due to differing definitions or feed vintages, or a bad bar, but the cause has not been identified, and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'Complete 24 hour days only; absolute relative OHLC difference threshold 1bp.' A data line was also found: {'s': 'ETH', 'year': 2025, 'days': 364, 'differ_1bp': 196,'max_field_diff_pct': 7.327398173096411, 'worst_date': '2025-12-02 00:00:00+00:00', 'worst_field': 'o', 'raw': 3020.1375, 'hour_agg': 2798.84}. This data indicates that there were differences in the ETH/USD data, possibly due to differing definitions or feed vintages, or a bad bar, but the cause was not identified and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'Complete 24 hour days only' with an absolute relative OHLC difference threshold of 1 basis point. The data shows that for the ETH symbol in the year 2024, out of 365 days, there were 335 days where the difference was 1 basis point or more. The maximum field difference percentage was 0.0835011760214055. The worst date was 2024-08-09 00:00:00+00:00, with the worst field being the close, which had a raw value of 2598.167 and an hour aggregation value of 2600.3365. The differences may be due to differing definitions or feed vintages, or a bad bar, but the cause has not been identified, and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the following data line: Complete 24 hour days only, with an absolute relative OHLC difference threshold of 1 basis point. The data shows that for BTC in 2026, there were 253 days, with 3 days having a difference of 1 basis point or more, and a maximum field difference percentage of 0.031982457520962626. The worst date was 2026-07-31 00:00:00+00:00, with the worst field being the close, which had a raw value of 62812.559 and an hour aggregation value of 62832.648. The differences may be due to differing definitions or feed vintages, or a bad bar, but the cause has not been identified, and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'BTC/USD C91204 daily/hour aggregation 2025'. The given data is about Bitcoin, denoted as 'BTC', in the year 2025. It shows that out of 364 days, 179 days had an absolute relative difference of at least 1 basis point, which is a measure used to describe small changes in financial values, in the Open, High, Low, Close prices. The maximum difference percentage was 5.843715455711418 percent, which occurred on December 2, 2025, at 00:00:00, in the opening price field. The raw value was 91625.132 and the hourly aggregated value was 86270.82. The differences could be due to varying definitions or feed versions, or possibly bad data, but the cause has not been identified and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline is not available, but a data line was found for BTC/USD C91204 daily/hour aggregation in 2024. The data shows that for complete 24-hour days only, the absolute relative OHLC difference threshold is 1 basis point. The data includes the following information: the symbol is BTC, the year is 2024, there are 365 days, and 308 days had a difference of 1 basis point or more. The maximum field difference percentage is 0.09154982964658664, which occurred on April 4, 2024, at 00:00:00+00:00, in the closing price, with a raw value of 68592.7 and an hour aggregation of 68529.9035. The differences may be due to differing definitions or feed vintages, or possibly a bad bar, but the cause has not been identified, and no silent repair was made. https://runway.adityaperswal.workers.dev/data/crypto-bars

A news scan found the headline 'SOL/USD C91204 managed halfATR uncertainty 1Hour holdout'. This refers to a trading strategy for the SOL/USD currency pair, where ATR stands for Average True Range, a measure of price volatility. The strategy was tested using a method called IID bootstrap, which involves drawing samples from a set of closed trades, in this case 10,000 draws with a seed of 91204. Note that these are not 10,000 independent trades. The results show a 95% win interval using the Wilson method, but there are unresolved issues with serial dependence and hypothesis selection. The test also found that it cannot invent losses, meaning that the strategy cannot create losing trades that did not happen. The specific results include a mean percentage change of -1.9686% with a 95% confidence interval of -10.6954% to 6.7582%, and a Wilson 95% win interval of 0.0945% to 0.9055%. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline 'SOL/USD C91204 managed halfATR uncertainty 1Hour train'. This refers to a trading strategy for the SOL/USD currency pair, where ATR stands for Average True Range, a measure of price volatility. The strategy was trained on a 1-hour time frame. The results are based on 10,000 bootstrap draws of closed trades, with a seed of 91204, meaning the trades were not all independent. The Wilson 95% win interval, a statistical method to estimate the probability of winning, was used. There is a problem with serial dependence and hypothesis selection that has not been resolved, and the bootstrap method cannot create losses from all-winning trades. The results show the following statistics for the SOL/USD pair on a 1-hour time frame: the mean percentage change was 2.23%, with a 95% confidence interval of -6.68% to 12.55%, and the Wilson 95% win interval was between 25.04% and 84.18%. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline about SOL/USD C91204 managed halfATR uncertainty 1Day holdout. This refers to a trading strategy for the SOL/USD currency pair, where halfATR means half of the average true range, a measure of price volatility. The strategy was tested with a bootstrap method, which is a statistical technique, using 10,000 draws of closed trades, with a seed of 91204, meaning the starting point for the random number generator. The results showed a 95% win interval, according to the Wilson score, which is a statistical method to calculate the confidence interval of a proportion. However, the results also indicated tiny sample size, serial dependence, and unresolved hypothesis selection issues, meaning the sample size was small, the data points were not independent, and the choice of statistical method may not be optimal. The bootstrap method could not simulate losses, as all the trades in the sample were winning trades. The trading strategy had the following parameters: symbol SOL, time frame 1Day, not trained, and 3 trades. The mean percentage change was -3.686414892151191%, with a 95% confidence interval of -13.810395898729311% to 10.803505777465961%. The Wilson 95% win interval was between 0.0614903152761605 and 0.7923450448735121. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A trading fund manager analyzed SOL/USD using a method called bootstrap, which involves drawing samples from a set of closed trades, in this case, 10,000 times, with a seed of 91204. This means the samples are not 10,000 independent trades. The manager used a 95% win interval, known as Wilson interval, to estimate the probability of winning. However, there are unresolved issues with serial dependence and hypothesis selection. The analysis found that it is not possible to invent losses when all trades are winning. The results show the following statistics for SOL with a 1-Day time frame and 4 trades: the mean percentage change is 7.2145%, with a 95% confidence interval of -5.8902% to 21.5234%, and the 95% Wilson win interval is between 30.0637% and 95.4414%. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A trading fund manager analyzed ETH/USD using a method called bootstrap, which is a way to estimate how reliable the results are. The manager used 10,000 draws of closed trades, with a seed of 91204, but these were not 10,000 independent trades. The results showed a 95% win interval, which is a range of values within which the true win rate is likely to lie, using a statistical method developed by Wilson. However, there are still some unresolved issues, such as the dependence between trades and the selection of hypotheses. The analysis found that the strategy had a mean percentage return of 6.157475017100844%, with a 95% confidence interval of 4.3010838940019225% to 9.728972460743535%. The win rate, according to the Wilson method, was between 0.43849391955098227 and 1. The strategy was applied to a 1-hour time frame, and the analysis used 3 trades, without training the model. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the following data: IID bootstrap with 10000 draws of closed trades, using seed 91204, which are not 10000 independent trades. The Wilson 95% win interval is provided. There is a tiny sample size and serial dependence, with hypothesis selection unresolved, meaning the bootstrap cannot invent losses. The data includes the symbol ETH, a 1 Hour time frame, and a training period. The number of trades is 13, with a mean percentage of 4.1234713333288555, and a mean confidence interval percentage of -3.7683574914868374 to 12.803957819711675. The win Wilson 95% interval is 0.2914343243183205 to 0.7679424569851186. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline 'ETH/USD C91204 managed halfATR uncertainty 1Day holdout'. This refers to a report on the Ethereum to US Dollar exchange rate, where a trading strategy called C91204 was used to manage uncertainty. The report is based on a statistical method called bootstrap, which involves drawing samples from a set of closed trades, in this case 10,000 times, with a seed value of 91204. Note that these are not 10,000 independent trades. The report also mentions a 95% win interval, which is a statistical measure of the range of possible win rates, calculated using the Wilson method. There are some unresolved issues with the analysis, including serial dependence and hypothesis selection, which means that the method used to test the strategy cannot invent losses, even if they are possible. The report includes some technical details, including the symbol's' being ETH, the time frame 'tf' being 1 day, and 'train' being set to false. The number of trades 'N' is 3, the mean percentage change'mean_pct' is 0.0247, and the mean confidence interval percentage'mean_CI_pct' is between -6.6375 and 5.0468. The win Wilson 95 percentage 'win_Wilson95' is between 0.2077 and 0.9385. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline about ETH/USD C91204 managed halfATR uncertainty 1Day train. This is a report about a trading strategy, where halfATR refers to half of the average true range, a measure of price volatility. The report is based on 10,000 draws of closed trades, with a seed of 91204, which means the trades were selected using a specific random number generator setting. The results show a 95% win interval, according to the Wilson method, which is a statistical technique. However, there are unresolved issues with serial dependence and hypothesis selection. The report also includes a bootstrap analysis, which is a way of estimating the uncertainty of the results. The analysis found that it is not possible to invent losses, even with all-winning bootstrap scenarios. The specific results are as follows: the mean percentage change was -2.73%, with a 95% confidence interval of -9.50% to 6.14%. The Wilson 95% win interval was between 13.68% and 69.43%. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline 'BTC/USD C91204 managed halfATR uncertainty 1Hour holdout'. This refers to a Bitcoin to US Dollar trading strategy. The strategy used a method called IID bootstrap with 10000 draws of closed trades, using a seed of 91204, which means it is not based on 10000 completely independent trades. The results showed a 95% win interval using the Wilson method. However, there are still some unresolved issues with the tiny sample size, serial dependence, and hypothesis selection. The bootstrap method cannot create losses even if all trades were winning. The strategy's parameters were: symbol 'BTC', time frame '1Hour', not trained, with 3 trades, a mean percentage change of 1.1100457864260616, and a mean confidence interval percentage of -6.169254005246194 to 8.560152831413115. The win Wilson 95 percentage interval was 0.2076549551264879 to 0.9385096847238394. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline 'BTC/USD C91204 managed halfATR uncertainty 1Hour train'. The body of the report is about a trading analysis. It mentions IID bootstrap with 10000 draws of closed trades, with a seed of 91204, noting that these are not 10000 independent trades. The report references a Wilson 95% win interval. It also talks about unresolved issues with serial dependence and hypothesis selection, stating that a bootstrap with all winning trades cannot invent losses. The report includes data: the symbol is BTC, the time frame is 1 Hour, it is a training session, N is 16, the mean percentage is -0.3559763329919273, the mean confidence interval percentage is between -3.957715474047218 and 3.7021122275061433, and the win Wilson 95 percentage is between 0.2309836333518552 and 0.6682180439322347. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the headline 'BTC/USD C91204 managed halfATR uncertainty 1Day holdout'. This refers to a report on the Bitcoin to US Dollar exchange rate. The report mentions a statistical method called bootstrap, which involves taking 10,000 draws of closed trades, with a seed value of 91204. Note that these are not 10,000 independent trades. The report also references a 95% win interval, which is a measure of confidence, named after a statistician. A term called serial dependence, which means the relationship between consecutive values, and hypothesis selection, which is the process of choosing a theory, are mentioned as unresolved issues. The report includes a result from a statistical test, which shows that a certain method cannot invent losses, even with all winning trades. The report provides specific data: the symbol's' is BTC, the time frame 'tf' is 1 Day, 'train' is set to false, 'N' is 3, the mean percentage change'mean_pct' is 2.3256, and the mean confidence interval percentage'mean_CI_pct' is between -1.7447 and 8.4605. Additionally, the win Wilson 95 percentage 'win_Wilson95' is between 0.2077 and 0.9385. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the following data: IID bootstrap with 10000 draws of closed trades, seed 91204, which are not 10000 independent trades. The Wilson 95% win interval is noted. There is a tiny sample size and serial dependence, with hypothesis selection unresolved, and all-winning bootstrap cannot invent losses. The data includes the symbol BTC, time frame 1Day, and train set to true, with the following statistics: N is 5, mean percentage is -2.770194337172638, mean confidence interval percentage is between -8.077825397805778 and 6.310732317388967, and win Wilson 95 is between 0.03622316096978745 and 0.6244717358814613. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.py

A news scan found the following trade data for SOL/USD: the trade used a 1 Hour time frame and a half Average True Range, or half ATR, which is a measure of price volatility. The data showed 2 trades, with no full coverage trades, and a mean percentage change of null, or no value. The trades were selected after they happened, because of missing data, and not because of a specific trading rule. The closed trades were as follows: a trade that started on 2026-01-06 18:00:00+00:00 and ended on 2026-01-25 16:15:00+00:00, with 3 missing data points and a return percentage of -10.69536448973344, and a trade that started on 2026-04-18 19:00:00+00:00 and ended on 2026-05-08 19:00:00+00:00, with 13 missing data points and a return percentage of 6.758158591128516. https://runway.adityaperswal.workers.dev/data/crypto-bars ; C91204 research/robust.json