MT5 Strategy Tester says 100% History Quality means results are GOOD?
- Modelling quality
- Problems with MT5 Strategy Tester
- Would you subscribe to a signal with deposit $5000, average monthly profit of 10% and max 15 year historichal DD of of 20%? (If yes, how long would the signal history have to be, if no why?)
No.
Not at all.
No.
Not at all.
Does that mean these results on MT5 Strategy tester are meaningless? Can there be at least some correlation between Tester results and results on DEMO account? If Tester results do not correlate or not same as what we get on DEMO in the case what purpose Strategy Tester in MT5 serves? Do you have any info on this?
To test if the programmed logic fires properly and to test for things like runtime errors.
You test on historic data it is not possible to test on the future, so results can not be the same.
Or do you think that it's possible to predict the future ?
To test if the programmed logic fires properly and to test for things like runtime errors.
You test on historic data it is not possible to test on the future, so results can not be the same.
Or do you think that it's possible to predict the future ?
I think my question is NOT about predicting future. I think I did not put my question correctly. My question is assuming that we run DEMO or LIVE(with no slippage assumption) on exactly similar data as Strategy Tester has run or putting it another way if in the future if somehow DEMO has identical data that of Strategy Tester is run, in such scenario will that strategy on Demo account produce similar results or at least there will be some correlation. I know it is just impossible to have future data identical to past data but what I am trying to understand here is that how valid Strategy Tester results are. In MT4 ST used to give quite erroneous results. Is there any improvement in MT5 as far as tick modeling is considered?
Yes.
But it's meant to be a closer approximation, because some people believe that these results should be exactly the same i don't know why.
Yes.
But it's meant to be a closer approximation, because some people believe that these results should be exactly the same i don't know why.
Yes, that is what I expected when it shows 100% History quality of data it must mean something. There must be some correlation between the demo live run on similar data on which ST has run. If I am getting %profit on ST then I should get some %profit (not exactly the same) when strategy is run on demo live.
I don't think so, but you can try.
Why do you think there can not be some profit(not exactly same) when run on demo account (assumption here is same data as it was used in ST)? Since it is demo account assume that there would not be any slippage or any other delay in execution. In such scenario, I think the results on ST will correlate with demo account run on identical data. I know in MT4 it was not the case particularly if EA is working on Tick movement in such cases MT4 ST will show huge profit but when it is run on other testers like TDS where history data quality is 99.9% it will show huge losses. Now that in MT5 History data quality is 100% and the test result show profit then it would show similar (not exactly the same) results on demo forward testing run on exactly same data. The problem with MT4 ST is its Tick movement is not modeled correctly and thus ends up in History data quality to 25%. But I think in MT5 this problem is fixed and that is the reason it shows history data quality 100%. What is your thoughts on this?
Because it is not only a data issue. Sure inferior data means inferior results.
But tester is not 100% simulation. For example. Swap is not calculated historically. It takes today's swap rates and applies on history, not how it should be in 100% simulation. So for example if you have long holding times, the profit or loss might be way off. This is not the only issue, there are a lot of other peculiarities, skewing results.
Use tester to debug/test your code, not to determine profitability.
Because it is not only a data issue. Sure inferior data means inferior results.
But tester is not 100% simulation. For example. Swap is not calculated historically. It takes today's swap rates and applies on history, not how it should be in 100% simulation. So for example if you have long holding times, the profit or loss might be way off. This is not the only issue, there are a lot of other peculiarities, skewing results.
Use tester to debug/test your code, not to determine profitability.
I agree with you up to certain extent but I have already used commercially available tester with 99.9% data quality showing profitable results for an EA (in MQL4) and when that EA is run on Demo it shows the profit similar(not exactly the same) to what is seen on tester. This prompted me to ask question about MT5 ST since it claims that history data quality is 100% I was curious to know if profit results obtained on MT5 ST are valid just as they are for commercially available tester.
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
You agree to website policy and terms of use