|
Difference in result when back-testing with 1min vs ticks |
whiletrue
|
Post subject: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Fri 13 Apr, 2018, 10:26
|
|
User rating: 0
Joined: Fri 17 May, 2013, 06:58 Posts: 76 Location: SwedenSweden
|
Hi,
I back-test with EUR/SEK and only act on 1h bars. I've recently found out that I get a very different result if I back-test with 1min bars or ticks. The result is consequently around +18 pips in difference for the 1min bars.
Any ideas on why I get this strange result?
Thanks.
|
|
|
|
|
mtnfx
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Tue 17 Apr, 2018, 11:01
|
|
User rating: 18
Joined: Thu 20 Apr, 2017, 22:42 Posts: 165 Location: Russian Federation,
|
I think result strongly depend on the logic of your strategy, for example - if you are using limit orders - they will be processed 'once per minute'. That means all min/max prices during the minute will most probably be ignored - i think orders will be processed using close price of 1-minute-bar.
PS: these are only my thoughts, I do not know exactly how it works ... but I was also getting rather different result when testing with 'tick' and '1 minute' accuracy.
|
|
|
|
|
whiletrue
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Wed 18 Apr, 2018, 10:31
|
|
User rating: 0
Joined: Fri 17 May, 2013, 06:58 Posts: 76 Location: SwedenSweden
|
True. But if I use PLACE_BID / PLACE_OFFER, shouldn't I get exact entry prices then?
|
|
|
|
|
whiletrue
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Sat 21 Apr, 2018, 17:29
|
|
User rating: 0
Joined: Fri 17 May, 2013, 06:58 Posts: 76 Location: SwedenSweden
|
Ok, so it seems that its a jforex issues:
This is one of my long order:
Order submitted with place_bid @ 10.29971 Order open price: 10.29954 3 last historical ticks: shift 0: 10.29934/10.29954 shift 1: 10.29956/10.29976 shift 2: 10.29972/10.29992
I would expect my open price to be my place_bid price, but jforex seems to use the first ask price that's below my place_bid price.
=(
|
|
|
|
|
mtnfx
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Sat 21 Apr, 2018, 22:27
|
|
User rating: 18
Joined: Thu 20 Apr, 2017, 22:42 Posts: 165 Location: Russian Federation,
|
I think PLACE_BID and PLACE_OFFER is just not possible to "reliably" test on history - because of nature of these orders - there are no actually BUYers and SELLers for your orders. Any "approximation" provided by tester will be "good enough".
|
|
|
|
|
whiletrue
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Mon 07 May, 2018, 14:57
|
|
User rating: 0
Joined: Fri 17 May, 2013, 06:58 Posts: 76 Location: SwedenSweden
|
What type of orders are you using in order to get the same open and close price regardless of backtesting with 1min or ticks?
|
|
|
|
|
mtnfx
|
Post subject: Re: Difference in result when back-testing with 1min vs ticks |
Post rating: 0
|
Posted: Tue 08 May, 2018, 21:53
|
|
User rating: 18
Joined: Thu 20 Apr, 2017, 22:42 Posts: 165 Location: Russian Federation,
|
I think there is no way to backtest on "1min" with the same accuracy as on "ticks", just because granularity of "1min" is different from "ticks". And this means - getting "same open and close prices" for both is not possible.
|
|
|
|
|
|
Pages: [
1
]
|
|
|
|
|