Hi
my EA backtests most favourably on H4 charts. Its a shame MT4 doesn't do H6, HN8 & H12 as i think these would work well also.......
My EA uses 2 of the same indicator and a maximum stop-loss only.
I have been backtesting/optimising my EA on the last 6 months of data, which is typically 70-110 trades.
I then plan to trade/run the optimum 6 month settings for a month, and then re-backtest/optimise the last 6 months at the end of each month, updating the 2 parameters of the indicator/s and the maximum stop-loss size if needs be.
My reasoning for this method of backtest/optimisation is that - yes i could backtest more data history eg. 1-2 years of historical H4 data, or samples of data from 2004 etc. but the market dynamics change over time, so who cares what the markets were doing 2-3 years ago?
The fact that i only use an indicator (*2) and a maximum stop-loss shows that my EA is very simple. Therefore i believe that my backtest results are not curve-fitted. If i had used an SL, TSL & profit target, and optimised those parameters, i could see how curve fitting would be much more likely.
The fact that my Backtests generate a similar number of historical trades before and after optimisation, also suggest that curve-fitting is minimal/non-existent. If each round of optimisation was producing fewer & fewer historical trades, this would be a tell tale sign of curve-fitting. But this has NOT happened with my tests.
What do you think about my 6 month backtest/optimisation and my plan to review/refresh the optimum parameters at the end of each month.
Honest and frank opinions wanted please.
Thanks a lot.
my EA backtests most favourably on H4 charts. Its a shame MT4 doesn't do H6, HN8 & H12 as i think these would work well also.......
My EA uses 2 of the same indicator and a maximum stop-loss only.
I have been backtesting/optimising my EA on the last 6 months of data, which is typically 70-110 trades.
I then plan to trade/run the optimum 6 month settings for a month, and then re-backtest/optimise the last 6 months at the end of each month, updating the 2 parameters of the indicator/s and the maximum stop-loss size if needs be.
My reasoning for this method of backtest/optimisation is that - yes i could backtest more data history eg. 1-2 years of historical H4 data, or samples of data from 2004 etc. but the market dynamics change over time, so who cares what the markets were doing 2-3 years ago?
The fact that i only use an indicator (*2) and a maximum stop-loss shows that my EA is very simple. Therefore i believe that my backtest results are not curve-fitted. If i had used an SL, TSL & profit target, and optimised those parameters, i could see how curve fitting would be much more likely.
The fact that my Backtests generate a similar number of historical trades before and after optimisation, also suggest that curve-fitting is minimal/non-existent. If each round of optimisation was producing fewer & fewer historical trades, this would be a tell tale sign of curve-fitting. But this has NOT happened with my tests.
What do you think about my 6 month backtest/optimisation and my plan to review/refresh the optimum parameters at the end of each month.
Honest and frank opinions wanted please.
Thanks a lot.