The research
Facts & stats
Considering that the bulk of the work to get results for 10 systems (20 including variants) started in April 2019 and lasted until September 2019, it took about 2564 man-hours. For the period we tested for (07/06/2015 – 27/07/2019) and for the 20 systems and the 7 assets, it took 30,240 single-week backtests to run on Metatrader 5 and 759,220 multi-week backtests to be extracted, for a total of 789,460 different tests. The data was collected in 20 SQL Server databases that need a total of 1.5TB of disk space. Their various backups also require about 1.5TB of disk space. Because a test has on average tens of thousands of different combinations, all databases have a total of more than 37,313,353,224 different test records.
For the needs of the d-Backtest PS method and the extraction of various statistics for each system, at least 75,806,500 additional records were required. The total number of records in the entire database exceeds 37,389,159,724. A total of 588 CPU cores were used for the running of the single-week tests, of which 456 were on machines of high computational capacity and consumed approximately 27,723 kWH of electricity.
Publications
7 publications in scientific journals
Automated trading systems’ evaluation using d-Backtest PS method and WM ranking in financial markets
Optimization of Backtesting Techniques in Automated High Frequency Trading Systems Using the d-Backtest PS Method
AdTurtle: An Advanced Turtle Trading System
Hedging and non-hedging trading strategies on commodities using the d-Backtest PS method. Optimized trading system hedging
Take Profit and Stop Loss Trading Strategies Comparison in Combination with an MACD Trading System
Performance Comparison of Three Automated Trading Systems (MACD, PIVOT and SMA) by Means of the d-Backtest PS Implementation
Profitability Edge by Dynamic Back Testing Optimal Period Selection for Technical Parameters Optimization, in Trading Systems with Forecasting