A strategy for every bar.
- This topic has 27 replies, 6 voices, and was last updated 4 years ago by
.
-
Is the 100 limitation the same in v11? Am I correct in thinking you have the free Daily version of v11?
Sorry GraHal – I missed your questions due to my own ramblings and musings! I have been doing my testing on a combination of v11 and v10.3 and my latest tests were all on v10.3. The v11 EOD has 5 day weeks where the v10.3 via IG has 6 day weeks so I have tended to stick with that on the last few tests just to keep some continuity. Also I struggle to use meaningful spreads in v11 EOD on DJI because it only allows one decimal point and seems to have different pricing /pipsize. So that is a long way of saying that yes perhaps I would be better off testing on v11 in some ways but in other ways not.
1 user thanked author for this post.
Is the 100 limitation the same in v11?
I just did a quick test on my v11 EOD and I’m afraid the 100 limit is still there.
1 user thanked author for this post.
You could either run 31 strategies or just start a strategy every day that only opens trades tomorrow. If it doesn’t open a trade tomorrow because conditions are not met then delete it and set a new strategy to run for the new tomorrow. Alternatively set 6 strategies running, one for each trading day of the week then once a week set a new strategy running for any days of a week that still have a trade open. This way you will never miss a single trade because you are already on the market.
You can simply add:
123456dn = 1 //Change from 0 to 5 for Sunday to Fridaytradeon = 0if opendayofweek = dn thentradeon = 1endifThen use TRADEON as a condition for entry.
I do use tick by tick mode if the strategy I’m testing has any way to exit mid candle with pending orders or stop and target orders. If the strategy only decides whether to close a trade or not at the end of a candle then I don’t need tick by tick.
1 user thanked author for this post.
SEB testing for me is just a much better robustness test than my VRT because the VRT just used the idea of shaking up and randomising the bars we traded on and then giving us some sort of monte carlo results for us to analyse whereas SEB testing shows us almost 100% of all possible trades if not 100% of all trades. The downside to SEB testing is that due to the need to use something like Excel to analyse the results it is really only practical on daily,weekly or monthly strategies.
I would absolutely love it if PRT added a tool that just added together the results of every possible trade on every bar and gave us statistics and an equity curve of those results. It would be a very powerful tool and much more powerful than our current back tests that miss trades because we are already on the market with a position and so fool us into thinking a strategy is good when in fact it is just curve fitted luck.
Your test of 45% more trades giving 29% more gain is a good example of a strategy that looked better in a standard back test than reality might actual give us. The good thing is that it was still profitable even if not quite as good as it first appeared. Our expectations of its potential have at least been brought a little back down to earth.
At the moment I don’t have any live strategies running because I am out sailing around Greece for the summer with limited gigabytes and not so much time for looking at screens due to the demands of navigation and swimming off the back of the boat and firing up the BBQ whilst sinking a glass of cold wine! It is not all beer and skittles though as I have spent the last few days sweating like a pig in 35C whilst fixing the anchor windlass!
1 user thanked author for this post.
‘dn’ and ‘wn’ are the variables that you set or optimise to open trades on certain days of certain weeks.
dn = day number (0 to 5)
wn = week number (1 to 53)
Setting dn to -1 carries out a normal back test to use for comparison.
1 user thanked author for this post.
I’ve had a daily strategy on forward test now for a couple of month with five identical strategies running – one for each day of the week – but I noticed that some trades are still open for over a week meaning that I missed possible trades. The ideal solution to this is that I have 53 strategies running (one for every possible week in the year) but IG won’t let me do this in demo so I adjusted the code to at least let me have one strategy for each day in even numbered weeks and one for each day in odd numbered weeks. So 12 possible entry days and a trade will only be missed when a trade is open for over two weeks.
Here is the SEB code:
123456789101112131415161718dn = 1 //0 to 5 for each day of the weekwn = 1 //1 or 2 for odd or even weeksonce thisweek = 1if opendayofweek < opendayofweek[1] thenif thisweek = 1 thenthisweek = 2elsethisweek = 1endifendifif opendayofweek = dn and thisweek = wn then(buy code here)endif(exit code here)The attached images show the results of the 12 strategies running the same code but just trading on different days in odd and even weeks. There are also images for the most profitable and least profitable equity curves plus one for what a normal back test result would show us. Lots of robustness information there!
1 user thanked author for this post.
Find exclusive trading pro-tools on 
