Every tick vs. Open Prices Only
Every tick vs. Open Prices Only
I've coded an EA that only looks at prices when a new bar appears, so to save backtesting time I decided to run my testing in Open Prices Only mode rather than Every Tick mode. Problem is that I get drastically different results between the two modes (0.6 profit factor in every tick mode vs. 2.7 profit factor in open prices only mode). I'm hoping someone might help me understand why I see such a drastic difference between the two backtest modes? Does Open Prices Only mode still use the downloaded tickstory data?
Re: Every tick vs. Open Prices Only
Hi Jt27,
The "Open" mode takes 1 tick per bar. If you are working on the 15-minute time-frame, for example, your EA would only see one-tick when testing when in reality there were potentially hundreds during that bar. This means your strategy is missing out listening to all the prices that occurred intra-bar where the price would have potentially moved markedly.
As a example, say your EA had a sell stop price of 20. If the open of the bar was at 10, then the strategy Tester will show your closed position at 10. In reality (and with tick data), you were more likely to have been filled at your specified stop price (or near to). This is why you are going to get quite different results between the two modelling methods.
Of course, some EAs/Strategies will be less impacted by testing with "Open price" mode, such as those that trade on a daily/weekly basis. In those cases, the "Open price" modelling method can give the tester a rough idea of how their strategy is performing (such as for optimization) until they are ready to do a test on a more accurate tick-by-tick basis.
The Tickstory data you exported will work for both 'Tick' and 'Open price' modelling modes.
Thanks.
The "Open" mode takes 1 tick per bar. If you are working on the 15-minute time-frame, for example, your EA would only see one-tick when testing when in reality there were potentially hundreds during that bar. This means your strategy is missing out listening to all the prices that occurred intra-bar where the price would have potentially moved markedly.
As a example, say your EA had a sell stop price of 20. If the open of the bar was at 10, then the strategy Tester will show your closed position at 10. In reality (and with tick data), you were more likely to have been filled at your specified stop price (or near to). This is why you are going to get quite different results between the two modelling methods.
Of course, some EAs/Strategies will be less impacted by testing with "Open price" mode, such as those that trade on a daily/weekly basis. In those cases, the "Open price" modelling method can give the tester a rough idea of how their strategy is performing (such as for optimization) until they are ready to do a test on a more accurate tick-by-tick basis.
The Tickstory data you exported will work for both 'Tick' and 'Open price' modelling modes.
Thanks.
Re: Every tick vs. Open Prices Only
OK, so is it safe to assume that the one tick that the tester sees is the first tick of the new bar?
Re: Every tick vs. Open Prices Only
That's the theory 