Can we add an advanced option for Strategy evolver utilize granular limit/stop processing with a particular scale like for normal strategy backtesting added to the evolver tab to check and select granular processing scale?
I have found numerous strategy evolver strategies that have been evolved for days look very promising to then have significant lower results after i add granular processing to backtest the strategy that utilized limits or stops. I believe it if given the option the evolver could find more robust strategies albeit, it would require more time per run and adjustment for timing out each evolver run due to the increased data processing that would be required at the user's discretion.
I have found numerous strategy evolver strategies that have been evolved for days look very promising to then have significant lower results after i add granular processing to backtest the strategy that utilized limits or stops. I believe it if given the option the evolver could find more robust strategies albeit, it would require more time per run and adjustment for timing out each evolver run due to the increased data processing that would be required at the user's discretion.
Rename
Such option would make the overall processing time significantly longer.
Granular processing just gives you another, different result - sometimes it will be worse, sometimes better. And, granular processing isn't available for delisted symbols used in Wealth-Data backtests.
Something you could try is dig in to the Evolver Preferences and reduce the Position sizing to prevent too many NSF positions. Use 1% sizing for a DataSet with 100 symbols. The overall return will be far less, but it will be more of a raw profit backtest since almost all trades will be included. Later you can take the result and run it with a more realistic size and put it through a Monte Carlo simulation to see the probabilities.
Something you could try is dig in to the Evolver Preferences and reduce the Position sizing to prevent too many NSF positions. Use 1% sizing for a DataSet with 100 symbols. The overall return will be far less, but it will be more of a raw profit backtest since almost all trades will be included. Later you can take the result and run it with a more realistic size and put it through a Monte Carlo simulation to see the probabilities.
Your Response
Post
Edit Post
Login is required