Many thanks !!
> The report's information about ticks and bars modelled is merely a depiction of how many bars and ticks that are in the FXT file
So understanding this correctly: the report from backtest will always depict how many bars and ticks that are in the FXT file, regardless of chosen tested date range. This is how the product functions and there is no export settings in tickstory to change it.
However, the backtest will use the actual ticks during the specified date range which will be represented as 99,9% modelling quality (but I will not see how many actual ticks the EA is using during the specified date range). Same with bars in test, it is not "same" bars tested, but superficially reporting "equal" nr of bars tested (variation depending on chosen time frame when backtesting but still equal with different date range).
> I think the main issue here is how is understanding how this is impacting your back-testing
Precisely, not knowing what source to trust creates a dilemma. Does 99,9% modelling quality mean that I successfully exported the tick data into Metatrader with 99,9% precision or, does it mean that it is using the correct tick data during the specified date range I want to backtest with 99,9% precision.
From my understanding, the visual mode often skips ticks to speed up the backtest, hence the design appears to be for visualizing the EA's behavior, not for precise backtesting.
Rgds / A
Same nr of bars in test and ticks modelled for every year
Re: Same nr of bars in test and ticks modelled for every year
Hi Anders,
In response to your points, you are correct in what you're saying. Additionally:
> Precisely, not knowing what source to trust creates a dilemma. Does 99,9% modelling quality mean that I successfully exported the tick data into Metatrader with 99,9% precision or, does it mean that it is using the correct tick data during the specified date range I want to backtest with 99,9% precision.
99.9% modelling quality means that you have correctly exported and utilised the tick data you have exported. The 99.9% is also a superficial number that is within the FXT file itself. We use this number to determine if the user has correctly used the tick data that Tickstory has exported. In usual circumstances, MT4 will create an FXT file and do a calculation to determine the "modelling quality" based on the input data. This is usually <90% due to the fact it's using interpolated/guessed data. This number is determined before the test even begins. Therefore, this number is only used to determine the quality of the data being used. When a user has gotten 99.9% modelling quality, we know they have correctly used Tickstory's exported tick data for their test.
> From my understanding, the visual mode often skips ticks to speed up the backtest, hence the design appears to be for visualizing the EA's behavior, not for precise backtesting.
I'm not aware of this occurring during "Visual mode" or even normal mode on the Strategy Tester. "Every tick" should mean just that. I do believe this sort of aggregation might happen in a live environment where if multiple ticks are coming in fast, some ticks might not be sent to your EA.
To summarise, generally speaking you can refer to your report for your EA's performance and of course trade results to see how your EA performed during the test period you selected. As long as you have 99.9% modelling quality on your back-test, you can be sure that this test used Tickstory's exported tick data.
Hope this helps.
In response to your points, you are correct in what you're saying. Additionally:
> Precisely, not knowing what source to trust creates a dilemma. Does 99,9% modelling quality mean that I successfully exported the tick data into Metatrader with 99,9% precision or, does it mean that it is using the correct tick data during the specified date range I want to backtest with 99,9% precision.
99.9% modelling quality means that you have correctly exported and utilised the tick data you have exported. The 99.9% is also a superficial number that is within the FXT file itself. We use this number to determine if the user has correctly used the tick data that Tickstory has exported. In usual circumstances, MT4 will create an FXT file and do a calculation to determine the "modelling quality" based on the input data. This is usually <90% due to the fact it's using interpolated/guessed data. This number is determined before the test even begins. Therefore, this number is only used to determine the quality of the data being used. When a user has gotten 99.9% modelling quality, we know they have correctly used Tickstory's exported tick data for their test.
> From my understanding, the visual mode often skips ticks to speed up the backtest, hence the design appears to be for visualizing the EA's behavior, not for precise backtesting.
I'm not aware of this occurring during "Visual mode" or even normal mode on the Strategy Tester. "Every tick" should mean just that. I do believe this sort of aggregation might happen in a live environment where if multiple ticks are coming in fast, some ticks might not be sent to your EA.
To summarise, generally speaking you can refer to your report for your EA's performance and of course trade results to see how your EA performed during the test period you selected. As long as you have 99.9% modelling quality on your back-test, you can be sure that this test used Tickstory's exported tick data.
Hope this helps.