perspective on strategy tester

[Nathan]

Hi all, I've observed different results on the strategy tested and was simply seeking a little perspective.

I originally worked on a couple of EA's that used the same skeleton code but each operated strictly in a specific timeframe.  I've attempted to incorporate them together to run simultaneously in a single EA.  I specify which timeframe to use in each function call.  When I reference Open/High/Low/Close I use iOpen.... etc.  I am pretty sure that each signal is correctly coded to work on its desired timeframe regardless of the timeframe of the chart the EA is in.

In strategy tester I get different behavior depending on the timeframe selected.  In the 1M and 5M the results are the same.  It is different in the 15M and 4H timeframes where I see more trades.  The signals each use one of the 5M, 15M, 1H, or 4H timeframe.

In addition to the below, are there any other suggestions or perspective on the strategy tester I might want to be aware of?  When I comment out all but a single signal it behaves more as expected.  I also set a fixed spread.

https://www.mql5.com/en/articles/1512

Zero bar of another timeframe for the same symbol under test is modeled approximately

Open = correct Open, Close = correct Close, Low = min (Open,Close), High = max (Open,Close), Volume = final Volume (false)

Best regards,

Nathan

[[Deleted]]

I doubt someone can help you without seeing the code.

Are the results reproducible ? For example on M5 you always have the same results ?

[William Roeder]

niqmadu: https://www.mql5.com/en/articles/1512 Zero bar of another timeframe for the same symbol under test is modeled approximately Open = correct Open, Close = correct Close, Low = min (Open,Close), High = max (Open,Close), Volume = final Volume (false)	No longer true, see 'Testing Features and Limits in MetaTrader 4' - MQL4 Articles

[Nathan]

zirkoner:

I doubt someone can help you without seeing the code.

Are the results reproducible ? For example on M5 you always have the same results ?

Yes, if I set the tester to the same timeframe I get the same results each test.  It is almost as if in the lower timeframe test the higher timeframe signal is not executed or not executed identically to how it would execute by itself in its timeframe.

I'll keep hammering at the code as I would expect the fault to lie there.  Before I ask any further insight I'll do detailed testing including writing the results to a log file and comparing the times and number of trades to the individual tests.

Thank you to both yourself and WHR for taking a moment to respond.

N