Possible bug when configuring in HalfPips or Pips

[pstrusi]

Hi Ninjas,

I want to explain a weird situation that I've realized doing different backtests. I work with IB, therefore my setting for Forex is in "HalfPip",data comes well and orders are sent normally.
But doing backtesting with the historical data something rare occurs, if I choose to set my NT in Pip, my backtesting are totally different, better and even more natural, but If I set in "HalfPip" the results not only aren't not good at all, but also it doesn't look realistic, cause numbers of trades, conditions....etc

Having in mind that is doing backtests working with historical data, it doesn't have much sense at all, so can you review what is this about?

Thanks

[NinjaTrader_Jason]

Hello pstrusi,

Data is different, so results can be different as well. When you use HalfPip, orders can execute on half pip prices. When set to Pip, orders will only only execute on whole pips. When you analyze the results, does this clarify the difference?

[pstrusi]

I think is not simple as that. For example the numbers of trades should be similar, instead I got anormally different numbers of trades, just this is meaningful enough to review that feature. Please consider to review that NT field

Thanks and please let me know of anything about it

[NinjaTrader_Jason]

Are you able to reproduce the issue if you test the SampleMACrossOver strategy on the same amount of data?

[pstrusi]

I'll try it, good suggestion to contrast. I´ll let u know when done

[pstrusi]

Hi Jason,

I did the backtesting with HalfPip and Pip working with "SampleMACrossOver". The results are different but not quite meaningful as in my code. However working with historical data it there shouldn't be any difference, don't u think?

I must debugging my code for this issue

Thanks

[NinjaTrader_Jason]

The differences you see with the SampleMACrossOver are likely caused by different executions. One tests only executes orders on whole pips, while the other test can execute orders at half pips.

Yes, please debug your strategy to check why it produces such different results.