A news scan found the note: a trading strategy for Ethereum, or ETH, failed to work as expected. The strategy was to buy ETH when its price closed above its highest price in the last 7 days, but only if the price range in the last 7 days was smaller than the average price range in the last 60 days. The plan was to sell ETH when its price reached a target price or after 7 days. The strategy was tested using historical data from 2025 and the first 9 months of 2026, with the last 3 months of 2024 used to prepare the system. The results showed that the strategy did not work well, with losses of 9.5346% and 15.5393% in daily and hourly tests. Even when the strategy was adjusted, it still did not perform well. The conclusion is that this strategy does not have a positive effect and should be rejected. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and the test scripts are test.py and verify.py.
A news scan found the note: a trading strategy called C91008 for Bitcoin, or BTC, did not work as expected. The strategy was to buy when the price closed above the highest price of the last 7 days, but only if the price had not changed much in the last 60 days. The plan was to sell when the price went up by 1.5 times the average daily range of the last 20 days, or to sell after 7 days. The strategy was tested using data from 2025 for training, from October to December 2024 for warm-up, and from January 1 to September 9, 2026, for validation. The results were: a loss of 8.6832% and 14.9605% for daily and hourly trades, and a loss of 1.1218% for hourly training. Another test showed a loss of 0.5559% and 9.0164% for daily and hourly trades. The conclusion is to reject this strategy because it did not show any independent advantage. The test used fees of 25 basis points, or 0.25%, per trade, and a slippage of 10 basis points, or 0.10%, as a baseline, and 20 basis points, or 0.20%, as a stress test. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and the test scripts are test.py and verify.py.
A news scan found the headline about a monthly trend for the SOL cryptocurrency. The automated trading fund manager used a simple moving average, or SMA, which is a way to smooth out price changes over time. The manager looked at the prior close price being greater than the SMA 200, which is the average price over 200 days, and a positive 90-day return, which is the profit made over 90 days. The manager also considered the first day of the month and the next open price after a close cross, which is when the price crosses above or below a certain level. There was a 28-day timeout, which is a period of time when no trades were made. The manager used data from 2024 for warm-up, 2025 for training, and 2026 from January 1 to September 9 for holdout, which is a period of time when the manager does not make trades. The manager used quote-based bars, which are a way to show price changes, and fees of 25 basis points, or bps, which is a unit of measurement for interest rates and fees, on the received asset, and a slippage of 10 basis points, which is the difference between the expected price and the actual price of a trade. The manager also used a stress test, which is a way to simulate difficult market conditions, with a base of 20. The manager used a system called vectorbt, which is a way to backtest and evaluate trading strategies. The results of the training period showed a return of -0.5283% and a drawdown of 12.3606%, which is the maximum loss of value during a certain period. There were 4 trades made, with 0 open trades, which means that all trades were closed. The win rate was 50%, which means that half of the trades were profitable. The average win was 11.2285% and the average loss was -10.3158%. The results of the hourly training period showed a return of -2.0384% and a drawdown of 11.8227%. The holdout period showed a return of -1.8358% and a drawdown of 6.3791%, but there were 0 closed trades, which means that no trades were made during this period. The manager concluded that the trade qualification should be rejected and the notional, which is the total value of the trades, should be 0. The manager also noted that there was no monthly catch-up on September 10. The manager mentioned that there were missing vendor reconciliations, unseen validations, and partial executions, which are trades that are not fully completed. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline about a monthly trend for the Ethereum cryptocurrency, or ETH. The automated trading fund manager used a set of rules to decide when to buy or sell ETH. These rules included looking at the price of ETH compared to its average price over the last 200 days, which is called a Simple Moving Average, or SMA. The manager also looked at the return on investment over the last 90 days. The manager only considered buying ETH on the first day of each month. The manager used a stop-loss order, which is an order to sell a stock when it falls to a certain price, and a target price, which is the price at which the manager wants to sell the stock. The manager also set a time limit of 28 days for each trade. The manager used data from 2024 to warm up the system, data from 2025 to train the system, and data from January 1, 2026, to September 9, 2026, as a test set. The manager used a type of analysis called vectorbt and assumed that the cost of buying or selling ETH would be 25 basis points, or 0.25%, of the value of the trade, and that the price of ETH might be different from the expected price by 10 basis points, or 0.10%. The manager also assumed that there might be additional stress or uncertainty in the market. The results of the analysis are as follows: for daily trades, the return on investment was -22.91%, the maximum loss was 29.60%, and there were 6 trades, with 0 trades still open. The win rate was 33.33%, the average return per trade was -3.84%, the average return per winning trade was 8.31%, and the average loss per losing trade was -9.91%. For hourly trades, the return on investment was -14.58%, the maximum loss was 21.20%, and there were 6 trades, with 0 trades still open. The win rate was 33.33%, the average return per trade was -2.38%, the average return per winning trade was 6.63%, and the average loss per losing trade was -6.89%. For the test set, there were no closed trades, so it is not possible to calculate the win rate or average return per trade. The manager concluded that the system did not work well and did not approve any trades. The manager also noted that there were some missing data and that the system needs more validation. The data used for this analysis came from the following sources: https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline 'C91006 BTC monthly trend: zero closed holdout trades prevents edge calibration'. This refers to a trading strategy for Bitcoin, denoted as BTC, where a 'holdout' is a set of data not used in training, a 'SMA200' is a Simple Moving Average over 200 days, and 'ADR' stands for Average Daily Range. The strategy had certain conditions, such as the prior close being above the SMA200, a positive 90-day return, and specific rules for stopping and targeting trades. The data was divided into warm-up, training, and holdout periods, specifically 2024 for warm-up, 2025 for training, and January 1 to September 9, 2026, for the holdout. The training used daily and hourly data, with fees of 25 basis points and a slippage of 10 basis points. The results for the daily training included a return percentage of 11.0372%, a drawdown percentage of 10.6879%, and a win rate percentage of 57.1429%. For the hourly training, the return percentage was 1.2896%, the drawdown percentage was 12.2081%, and the win rate percentage was also 57.1429%. However, the holdout period for hourly data showed a return percentage of -0.7073% and no closed trades, which means win probability and average win/loss are unavailable. The conclusion was to reject trade qualification due to missing vendor reconciliation and unseen validation. The source data came from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline 'C91006 CLI momentum90 is a gross benchmark, not monthly strategy validation'. Required CLI was executed on September 10 at 06:03 UTC, with a momentum lookback of 90 days, using 617 and 618 bars. The results were as follows: Bitcoin had a return of -2.5390% with a drawdown of 25.7377% and 33.333% wins out of 3; Ethereum had a return of +17.8630% with a drawdown of 40.3609% and 33.333% wins out of 6; Solana had a return of -25.4192% with a drawdown of 43.3791% and 25% wins out of 12. The premises of this test were that the CLI continuous same-close momentum is not frozen, and it uses a 1830 monthly-entry rule, with no fees, slippage, or holdout, and includes the partial current day. The conclusion is that the positive Ethereum benchmark does not authorize entry, and it should be reported only as a harness diagnostic. The source of this data is https://runway.adityaperswal.workers.dev/data/crypto-bars, which is a data line.
A diagnostic test was done on a trading system called CLI, which looked at three cryptocurrency pairs: BTC/USD, ETH/USD, and SOL/USD. The test used a strategy called sma-cross, which is a way to decide when to buy or sell based on moving averages, with a fast average of 5 days and a slow average of 50 days, using 617 days of data from a website. The results were: BTC had a return of 24.768%, a maximum loss of 27.358%, and won 44.444% of the time, with 9 trades; ETH had a return of 98.249%, a maximum loss of 43.756%, and won 50% of the time, with 6 trades; and SOL had a return of -9.895%, a maximum loss of 56.787%, and won 20% of the time, with 10 trades. Each cryptocurrency had 618 days of data. These results do not include fees or other costs, and the test did not set aside any data to check the results. Because the test did not include these important factors, the results cannot be used to decide whether to trade. Instead, a different test, called vectorbt experiment 1991-1993, will be used to answer the research question. A news scan found a headline about positive returns from a different rule, but it does not change the fact that the CLI test results are not reliable.
A news scan found the note about a trading rule for the cryptocurrency SOL. The rule had three conditions: the price had to be below the 50-day Simple Moving Average, the 3-day return had to be less than 0, and the next open had to be a flat entry, with exit when the prior close was above the 5-day Simple Moving Average or after 5 calendar days, using hourly windows of 1200, 72, and 120. The rule was tested with 25 basis point fees and 10 basis point slippage on each side, 20 basis point stress, and 40/60 neighbors not selected. The test period was from October 2024 for warmup, 2025 for training, and from January 1 to September 9, 2026, for holdout, with a reset of 100 cash and end on the last real open liquidation, with no invented execution prices. The rule was supplied daily, reconstructed from 24-hour days, and tested hourly, with matched supplied daily using an identical missing day mask. The results of the training were: daily return of -22.8172% in 24 days, reconstructed return of -17.9771% in 23 days, and hourly return of -55.6408% in 152 hours. Out-of-sample results were: supplied daily return of -21.4602% in 12 days, reconstructed return of -18.9768% in 8 days, matched return of -18.9768% in 8 days, and hourly return of -50.1622% in 82 hours. The hourly whole capital mean was -0.8307%, with a win rate of 10.9756% and average win/loss of 0.5365/0.9992%. The conclusion was no entry. 120 cells were recorded from 1993 to 1970, with 4 positive out-of-sample neighbors that did not select. The previously seen out-of-sample was a reanalysis, not a new independent validation. There were unreconciled vendor causes, quote midpoint fills, 15-minute stops and green exits that were unvalidated, with no deployable stops or targets. Trade payoffs divided the vector backtest return by 1.0025, with a credit-asset fee approximation. The sources were https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A trading rule for Ethereum, called ETH, was tested. The rule had two main parts: when to enter a trade and when to exit. To enter, the price had to be below the 50-day Simple Moving Average and the return over the last 3 days had to be less than 0. The trade would enter when the price was flat, and exit when the prior close was above the 5-day Simple Moving Average or after 5 calendar days. The rule was tested on hourly data with windows of 1200, 72, and 120 hours. The test included fees of 25 basis points and slippage of 10 basis points on each side, and a stress test of 20 basis points. The test used data from October 2024 for warm-up, 2025 for training, and from January 1 to September 9, 2026, for holdout. The test started with $100 cash and ended with the last real open liquidation, without invented execution prices. The rule was tested on daily, reconstructed, and hourly data. The results were: daily return of 14.2986% with 19 trades, reconstructed return of 14.2986% with 19 trades, and hourly return of -52.5388% with 118 trades. Out-of-sample results were: daily return of -0.3949% with 21 trades, reconstructed return of -3.4189% with 14 trades, matched return of -3.4189% with 14 trades, and hourly return of -54.3360% with 111 trades. The mean return on whole capital was -0.6943%, with a win rate of 8.1081% and an average win/loss ratio of 0.6484/0.8128%. The conclusion was not to enter the trade. The test also found that 120 cells recorded from 1993 to 1970 had 4 positive out-of-sample neighbors, but they did not select the rule. The test was a reanalysis, not a new independent validation. The vendor cause was unreconciled, and quote midpoint fills, 15-minute stops, and green exits were unvalidated. There were no deployable stops or targets. Trade payoffs were divided by 1.0025, which is a credit-asset fee approximation. The data sources were https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A trading rule for Bitcoin was tested. The rule had three conditions: the price must be below the 50-day moving average, the 3-day return must be less than 0, and the next open must be when the price is flat. The rule also had conditions for exiting a trade, which was when the prior close was above the 5-day moving average or after 5 calendar days. The test used hourly data with 1200, 72, and 120 windows. The test also included fees of 25 basis points and slippage of 10 basis points on each side, as well as a stress test of 20 basis points. The test was run from October 2024 for warm-up, 2025 for training, and from January 1, 2026, to September 9, 2026, for holdout. The results showed returns of -9.67% for daily data, -4.33% for reconstructed data, and -55.64% for hourly data. Out-of-sample testing showed returns of -5.00% for daily data, -9.89% for reconstructed data, -10.70% for matched data, and -51.38% for hourly data. The hourly data also showed a mean return of -0.66%, a win rate of 9.17%, and an average win-to-loss ratio of 0.26 to 0.75. The conclusion was to not enter a trade. A total of 120 cells were recorded from 1993 to 1970, but only 4 positive out-of-sample neighbors were found, which is not enough to select the rule. The test also found that previously seen out-of-sample data was a reanalysis, not new independent validation. The test had some limitations, including unreconciled vendor causes, unvalidated quote midpoint fills, and 15-minute stops and green exits. The trade payoffs were divided by 1.0025 to approximate the credit and asset fees. The data sources were https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees.
A news scan found the headline 'UNI frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for the cryptocurrency UNI. The strategy had two conditions to enter a trade: the closing price was higher than the 50-day Simple Moving Average, which is a measure of the average price over 50 days, and the 3-day return was less than 0. The strategy would exit a trade when the previous closing price was higher than the 5-day Simple Moving Average or after 5 calendar days. The strategy used a 1-hour and daily time frame, with specific rules for when to open and close trades. The strategy was tested using historical data from 2025, with a holdout period from January 1, 2026, to September 9, 2024, and a warm-up period in October 2024. The results showed a return of -58.0976% for the hourly time frame and -17.5305% for the daily time frame. The strategy also had a drawdown, which is the maximum loss from a peak to a trough, of 58.8265% for the hourly time frame and 19.0732% for the daily time frame. The win percentage, which is the percentage of trades that were profitable, was 33.0189% for the hourly time frame and 63.6364% for the daily time frame. The strategy was not validated for 15-minute exits or stops and is not deployable. The data was sourced from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'LTC frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for the cryptocurrency Litecoin (LTC) that was tested and rejected. The strategy was based on two premises: the closing price being higher than the 50-day Simple Moving Average (SMA), and a 3-day return less than 0. The strategy entered the market at the next available bar open when the conditions were met, and exited when the previous close was higher than the 5-day SMA or after 5 calendar days. The hourly window for trading was set at 1200, 72, and 120.2. The strategy used a 25 basis point fee and 10 basis point slippage on each side, with a 20 basis point stress test. The training data was from 2025, with a holdout period from January 1 to September 9, 2024, and an October warm-up period. The results showed a daily return of -19.69% and a hourly return of -49.68%. The daily win rate was 35.71%, while the hourly win rate was 9.64%. The strategy was not validated for 15-minute venue exits or stops, and is not deployable. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'HYPE frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy called HYPE that was tested and found not to work. The strategy was based on two premises: when the price of a stock was higher than its average price over the last 50 days and the return over the last 3 days was less than 0, the strategy would enter the market at the next available opening price. The strategy would exit the market when the previous closing price was higher than the average price over the last 5 days or after 5 calendar days. The strategy used a technique called vector backtesting with 25 basis points fees and 10 basis points slippage on each side, and 20 basis points stress. The strategy was trained on data from 2025 and tested on data from 2026 from January 1 to September 9, with a warm-up period in October 2024. The strategy used a cash reset of 100 and would liquidate at the last available opening price. The strategy excluded missing bars and did not fill them for execution or signals. The strategy had distinct daily and hourly patterns. The training data had limits on inference from 1937 and did not include pre-2026 training data for HYPE. The strategy did not validate 15-minute venue exits or stops and was not deployable. The results of the strategy were as follows: for the hourly strategy, the return was -41.19%, the drawdown was 41.61%, and the win percentage was 36.84%. For the daily strategy, the return was 17.98%, the drawdown was 13.42%, and the win percentage was 62.5%. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'LINK frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for the cryptocurrency LINK. The strategy has two main rules: it enters the market when the closing price is higher than the 50-day Simple Moving Average and the 3-day return is less than 0, and it exits when the previous closing price is higher than the 5-day Simple Moving Average or after 5 calendar days. The strategy uses a 1-hour and daily time frame, with a 1.2% fee and 0.1% slippage on each trade. The training data is from 2025, and the holdout period is from January 1, 2026, to September 9, 2024, with a warm-up period in October 2024. The results show a return of -56.20% for the hourly time frame and -12.67% for the daily time frame, with a drawdown of 56.57% and 23.29% respectively. The win percentage is 14.39% for the hourly time frame and 47.37% for the daily time frame. The strategy uses a concept called 'quarter Kelly', which is a measure of the optimal fraction of the portfolio to invest in a trade. The results also show that the capital return is not exactly credited to the asset ledger due to fees and trade returns. The source of the data is from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'XRP frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for the cryptocurrency XRP. The strategy had two conditions to enter a trade: the closing price had to be higher than the 50-day Simple Moving Average, and the 3-day return had to be less than 0. The strategy would exit a trade when the previous closing price was higher than the 5-day Simple Moving Average or after 5 calendar days. A Simple Moving Average is a way to smooth out price data over a certain period. The strategy used a 25 basis point fee and 10 basis points of slippage on each side of a trade, with a 20 basis point stress test. A basis point is a unit of measurement for interest rates and fees. The strategy was trained on data from 2025 and tested on data from January 1 to September 9, 2024, with a warm-up period in October 2024. The results showed a daily return of -10.12% and a hourly return of -37.99%. The strategy also had a daily drawdown of 16.52% and a hourly drawdown of 39.38%. A drawdown is the percentage decrease in value from a peak to a trough. The strategy was not validated for 15-minute exits or stops and is not deployable. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'SOL frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for the cryptocurrency SOL. The strategy had two conditions to enter a trade: the closing price had to be higher than the 50-day Simple Moving Average, which is a way to smooth out price data over time, and the 3-day return had to be less than 0. The strategy would enter a trade at the next available opening price when these conditions were met and the previous day's closing price was higher than the 5-day Simple Moving Average, or after 5 calendar days. The strategy used hourly time frames with specific settings, including a 25 basis point fee and 10 basis point slippage, which refers to the difference between expected and actual trade prices. The strategy was trained on data from 2025 and tested on data from 2026, with a warm-up period in October 2024. The results showed a daily return of -21.46% and a hourly return of -50.16%, with a drawdown of 28.08% and 51.02% respectively. The strategy was not validated for 15-minute exits or stops and is not deployable. The data was sourced from https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'ETH frozen trend-pullback daily/hourly holdout rejected'. This refers to a type of investment strategy that uses a trend, or a pattern of price movement, and a pullback, or a brief reversal of that trend. The strategy was applied to Ethereum, a type of cryptocurrency, and was rejected. The strategy had two main rules: it would enter the market when the closing price was higher than the 50-day simple moving average, which is a way to smooth out price fluctuations, and the 3-day return was less than 0. It would exit the market when the previous closing price was higher than the 5-day simple moving average or after 5 calendar days. The strategy was tested on hourly and daily time frames, which are periods of time used to analyze price movements. The test used a type of simulation called a vector backtest, which is a way to test an investment strategy using historical data, and included fees of 25 basis points, which is a way to measure a small percentage, and slippage of 10 basis points, which is the difference between the expected price and the actual price of a trade. The test also included a stress test, which is a way to simulate the performance of an investment strategy during difficult market conditions, of 20 basis points. The strategy was trained on data from 2025 and tested on data from 2026, and the results showed that the strategy did not perform well. The daily time frame had a return of -0.3949 and a drawdown, which is the maximum amount of money lost, of 14.3784. The hourly time frame had a return of -54.3359 and a drawdown of 54.6961. The results also showed that the strategy had a low win percentage, which is the percentage of trades that were profitable, and a high average loss, which is the average amount of money lost per trade. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-trading.
A news scan found the headline 'BTC frozen trend-pullback daily/hourly holdout rejected'. This refers to a trading strategy for Bitcoin that was tested and rejected. The strategy had two main rules: it would enter a trade when the closing price was higher than the 50-day Simple Moving Average, which is a measure of the average price over the last 50 days, and the 3-day return was less than 0. It would exit a trade when the previous closing price was higher than the 5-day Simple Moving Average or after 5 calendar days. The strategy was tested on hourly and daily time frames, with specific rules for each. The testing used historical data from 2025 for training and 2026 from January 1 to September 9 for testing, with a warm-up period in October 2024. The results showed that the strategy did not perform well, with negative returns and high drawdowns, which is the difference between the highest and lowest prices during a given period. The hourly strategy had a return of -51.38% and a drawdown of 51.38%, while the daily strategy had a return of -5.00% and a drawdown of 12.38%. The testing also used a method called bootstrapping to simulate different scenarios and estimate the strategy's performance. The results were not optimistic, and the strategy was rejected.
A news scan found the body of a note from a trading fund manager that said the following: the manager ran a mean-reversion screen using the required command line interface with a lookback of 20 days, entry of 2, exit of 0, and 617 days of data. The manager's premises engine uses current closes for signals and fills, with no fees, slippage, or holdout. Seven of the eight gross returns were positive, with the largest being LINK at 69.1614% and a drawdown of 11.9156%, a win rate of 73.6842%, and 19 trades. Other results included LTC at 45.3542% and a drawdown of 12.1012%, a win rate of 80%, and 20 trades, and HYPE at 14.61285% with 2 trades and no history from 2020 to 2025. This is not a frozen trend-pullback rule, so the result is not a replication or parameter selection. Other results included BTC at -3.257646%, a drawdown of 16.639125%, a win rate of 61.1111%, and 18 trades, ETH at 5.419817% with 15 trades, SOL at 15.553973% with 17 trades, XRP at 36.165798% with 19 trades, and UNI at 19.214556% with 22 trades. The average gross loss may be far worse than the win rate implies. The manager rejected deployment from this diagnostic and required a next open independent daily and hourly holdout. All figures are from observation 19211, and the latest 618 bar includes a potential partial data point from September 10, which includes 206 HYPE data points. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars, and the local source is /app/agent/src/backtest/engine.ts and strategies.ts.
A news scan found the headline is not available, but a diagnostic report of a momentum screen was found. The report ran a test on three cryptocurrency pairs, BTC/USD, ETH/USD, and SOL/USD, using a momentum strategy with a lookback period of 3 days and 617 days of data. The results showed a return of -10.948474% for BTC, -35.643586% for ETH, and -39.976039% for SOL, with drawdowns of 33.437946%, 62.105551%, and 60.750072% respectively. The test also reported win rates of 32.58427%, 28.888889%, and 26.373626%, with 89, 90, and 91 trades respectively. The test used the same close signals and fills, without considering fees or slippage. The results include the latest bars from days 618, 617, and 618, which may be partial and from the current day. Therefore, the diagnostic test cannot validate any entry, and even with optimistic gross results, the screen loses. A separate analysis using vectorbt, which includes data from 2025 for training and 2026 for holdout, will supersede the current execution and evaluation method. The data source is https://runway.adityaperswal.workers.dev/data/crypto-bars, and all figures are from observations recorded without any selection from the command line interface.
A news scan found the headline 'SOL fixed trend momentum daily/hourly test: reject'. This refers to a test of a trading rule for the cryptocurrency SOL. The rule was based on two premises: the closing price being higher than the 20-day Simple Moving Average, and a positive momentum over 3 days. If these conditions were met, the rule would enter a trade at the next open, and exit at the next open if the conditions failed. The test used hourly data with a 480/72 lookback period, and accounted for fees of 0.25% on each side, plus a 0.10% slippage baseline and 0.20% stress. The test was run from 2025, with an untuned holdout period from January 1 to September 9, 2026, and a warm-up period in November 2024. The initial cash was reset to 100 at each period, and the final open was liquidated. The test also used a neighbors parameter of 15/2 and 30/5 for robustness, but not for holdout selection. A complete slow window was required, and missing bars were not filled. The test did not include stop or target prices in the diagnostic, and failed before the 15-minute exit gate, so it cannot be deployed. The daily and hourly rules were evaluated separately, not just as changes in execution frequency. The baseline hourly results were: symbol SOL, time frame 1 hour, slow window 20, lookback 3, period out of sample, slippage 0.001, return -40.24%, drawdown 58.60%, number of trades 91, win percentage 16.48%, average win 4.78, average loss 1.51, mean -0.47, confidence interval -1.24 to 0.67, Kelly criterion 0. The baseline daily results were: symbol SOL, time frame 1 day, slow window 20, lookback 3, period out of sample, slippage 0.001, return 4.67%, drawdown 35.77%, number of trades 21, win percentage 28.57%, average win 11.05, average loss 3.63, mean 0.56, confidence interval -2.65 to 4.82, Kelly criterion 0.01. All 24 cells were saved in two preceding observations, including training and stress. The test also used a bootstrap diagnostic with 10,000 trades to infer serial dependence and selection limit. The test rejected the family of momentum rules, not just this specific rule. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The test was run using local study and test rule scripts, and prior failed rules 1639-41 were separate.
A news scan found the headline 'ETH fixed trend momentum daily/hourly test: reject'. This refers to a test of a trading rule for Ethereum, which is a type of digital currency. The rule, called Premises1, is based on two conditions: the closing price is higher than the average price over the last 20 days, and the momentum, or rate of change, over the last 3 days is positive. A momentum is a measure of how fast the price is changing. If both conditions are met, the rule enters a trade at the next open, and exits at the next open if the conditions are no longer met. The test used hourly and daily time frames, with a lookback period of 480 and 72 hours, respectively. A lookback period is the amount of time used to calculate the conditions. The test also included fees of 0.25% on each side of the trade, plus a slippage baseline of 0.10% and a stress test of 0.20%. Slippage refers to the difference between the expected and actual prices of a trade. The test was run from January 1, 2025, to September 9, 2026, with a warm-up period in November 2024. The initial cash was reset to $100 at the start of each period, and the final open position was liquidated. The test used 15 and 2 neighbors for robustness, but not for holdout selection. A holdout selection is a method of evaluating a trading rule by setting aside a portion of the data. The results showed that the rule failed, with a return of -46.58% and a maximum drawdown of 53.99% for the hourly time frame, and a return of -11.42% and a maximum drawdown of 37.74% for the daily time frame. A drawdown is the maximum loss from a peak to a trough. The test also included a diagnostic to evaluate the performance of the rule, but it failed before the 15-minute exit gate, which means it cannot be deployed. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees. The test was run using local study and test rule scripts, and prior failed rules 1639-41 were separate.
A news scan found the headline 'BTC fixed trend momentum daily/hourly test: reject'. This refers to a test of a trading rule for Bitcoin. The rule is based on two conditions: the closing price is higher than the average price over the last 20 days, and the momentum, or rate of change, over the last 3 days is positive. If both conditions are met, the rule enters a trade at the next open, and exits at the next open if the conditions are no longer met. The test used hourly data with a lookback period of 480 and 72 hours, and accounted for fees of 0.25% on each side of the trade, as well as slippage of 0.10% and stress of 0.20%. The test was run from January 1, 2026, to September 9, 2026, with a warm-up period in November 2024, and used an initial cash balance of $100 that was reset at the start of each period. The test also used a final open liquidation, meaning that any open trades were closed at the end of the test. The results of the test showed that the rule did not perform well, with a return of -43.15% and a maximum drawdown of 52.33% over 91 trades on an hourly basis, and a return of -6.24% and a maximum drawdown of 27.17% over 24 trades on a daily basis. The test also showed that the rule had a low win percentage, with only 9.89% of trades being winners on an hourly basis, and 29.17% of trades being winners on a daily basis. The test used a bootstrap method to estimate the confidence interval of the results, and found that the rule did not perform well enough to be deployed. The source of the data is https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees, and the test was run using local study and test rule scripts.
A news scan found the headline about an unconditional long every 3 days at 00 UTC starting from January 1, 2025, without using RSI or trend filters and not making new entry recommendations. The strategy was trained from 2025 and tested from January 1, 2026, to September 8, 2026, excluding boundary crossings and incomplete 72-hour paths from all cadences. The strategy used a prior completed native day Wilder ATR 14, stop 1 ATR, and target 1.5 ATR from the same 15-minute open reference. The first day, hour, and 15-minute completed close crossing executed at the observed close plus 5 basis points or 15 basis points slip, or 72-hour time exit, assuming optimistic zero quote and latency delay with no intra-bar ideal fills. Cash fees were 0.0025 for entry and 0.0025 for exit. The long-only vector backtest used a 15-minute mark grid with 36 compounded payoff reconciliations passing 1e-6 percent. The baseline out-of-sample results were: for 1 Day, return -36.93073427468296, drawdown 59.91421940730469, number of trades 70, percentage of winning trades 47.14285714285714, average win 4.242198589980639, average loss 4.7067872293689135, and mean -0.4879796288184106; for 1 Hour, return -24.153715262172938, drawdown 48.868742831016796, number of trades 70, percentage of winning trades 47.14285714285714, average win 4.411310293275279, average loss 4.4049829528313404, and mean -0.24873042252393443; for 15 Minutes, return -18.732950311684434, drawdown 45.59516535029439, number of trades 70, percentage of winning trades 47.14285714285714, average win 4.496454563326082, average loss 4.30118505132661, and mean -0.15372637584748414. The training returns for day, hour, and 15 minutes were -51.5944, -73.1472, and -67.0913, respectively. Every out-of-sample baseline and stress test was negative, with a raw quarter Kelly of 0. The native hour matches 4 times 15-minute closes exactly, and the native daily close differs by less than or equal to approximately 12 basis points in full history, with entry using a common 15-minute open to avoid larger daily open mismatch. Complete case omission may bias the regime, and the calendar was reused from prior research and is not independent confirmation. Stop overshoot worsens loss above nominal 1R, and real 15-minute venue checks do not guarantee matched bar close fill. The conclusion rejects unconditional entry, and the audit does not validate previously rejected pullback. The sources are https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees, with all rows observations from 18056 to 18091 and local audit.py.
A news scan found the headline about an unconditional long every 3 days at 00 UTC starting from January 1, 2025, without using RSI or trend filters and not making new entry recommendations. The training period was from 2025 to September 8, 2026, and the out-of-sample period was from January 1 to September 8, 2026, omitting boundary crossings and incomplete 72-hour paths from all cadences. The prior completed native day Wilder ATR 14, stop 1 ATR, target 1.5 ATR from the same 15-minute open reference was used. The first 1-day, 1-hour, and 15-minute completed close crossing executes at the observed close plus 5 basis points or 15 basis points slip, or 72-hour time exit, with optimistic zero quote and latency delay, and no intrabar ideal fills. Cash fees for entry were 0.0025 and for exit were 0.9975. The long-only vector backtest used a 15-minute mark grid, and 36 compounded payoff reconciliations passed 1e-6 percent. The baseline out-of-sample results were as follows: for 1-day time frame, the return was -49.95067278071193, drawdown was 61.77394490365453, number of trades was 61, percentage of profitable trades was 44.26229508196721, average win was 3.878693206992178, average loss was 4.826088557776246, and mean return was -0.9731523668131724. For 1-hour time frame, the return was -37.418698372687395, drawdown was 50.49805926238281, number of trades was 61, percentage of profitable trades was 44.26229508196721, average win was 3.995623078004095, average loss was 4.320962944734015, and mean return was -0.6398510986040323. For 15-minute time frame, the return was -32.963717240492755, drawdown was 46.209085516003576, number of trades was 61, percentage of profitable trades was 44.26229508196721, average win was 4.139849602814085, average loss was 4.236197399024663, and mean return was -0.5287667588665288. The training returns for daily, hourly, and 15-minute time frames were -61.1674, -72.3326, and -70.2655, respectively. Every out-of-sample baseline and stress test was negative, with a raw quarter Kelly of 0. The native hour matches 4 times 15-minute closes exactly, and the native daily close differs by less than or equal to approximately 12 basis points in full history, with entry using a common 15-minute open to avoid a larger daily open mismatch. Complete case omission may bias the regime, and the calendar was reused by prior research and is not independent confirmation. Stop overshoot worsens loss above nominal 1R, and real 15-minute venue checks do not guarantee matched bar close fill. The conclusion rejects unconditional entry, and the audit does not validate previously rejected pullback. The sources of the data are https://runway.adityaperswal.workers.dev/data/crypto-bars and https://docs.alpaca.markets/us/docs/crypto-fees, with all rows of observations from 18056 to 18091 and local audit.py.