Hello,
this is a bug I wanted to report for a long time, just didn´t have the time to do so as it is a somewhat complicated one. I noticed it from Multicharts 5.0 and up.
This bug applies for almost ANY strategy and ANY timeframe, so you can easily test it by yourself.
My setup is this: a custom strategy which I unfortunately can´t publish (but as I said it happens almost for any, even basic, strategy anyway), on the 1 Minute chart, no Intrabar Order Generation (though it doesn´t matter as well, happens with and without Intrabar Order Generation enabled).
To recreate this bug, just setup any chart and any timeframe with one of your favorite strategies. Now set the "Days Back" setting of your chart to 3 days, just as an example, or so far back that you get at least 1 page full of trades in the "List of Trades" in the "Strategy Performance Report".
Write down those trades or better, take a screenshot.
Now ONLY change the "Days Back" (NOTHING ELSE) of your chart to like 50 days. Now again open the "List of Trades" and compare the trades from the screenshot you´ve previously taken with the "2 Days Back" setting to the last trades of the "50 Days Back" setting.
And you will see, the trades don´t match up exactly. Sometimes more, sometimes less, but in 90% of the cases they don´t match up for the time the trade was entered, the profit and so on, just simply there are completely different trades appearing (while many are similiar though), even there was no settings changed on the strategy itself. And that actually shouldn´t happen since I need accurate back-testing results.
Hope this bug gets fixed.
Attached are some screenshots to illustrate the problem.
[Bug Report] Different Trades when changing Days Back
-
- Posts: 100
- Joined: Jul 17 2009
- Location: Germany
- Attachments
-
- missingtradescompare.GIF
- (114.28 KiB) Downloaded 454 times
-
- missingtrades2.GIF
- (105.13 KiB) Downloaded 453 times
-
- missingtrades1.GIF
- (124.04 KiB) Downloaded 455 times
-
- Posts: 100
- Joined: Jul 17 2009
- Location: Germany
P.S.: The last trade on the "missingtradescompare.gif" can be ignored since it of course is a open trade.
- Marina Pashkova
- Posts: 2758
- Joined: Jul 27 2007
Hi geektrader,
The behavior you are reporting is not a bug. Whether it occurs depends on your strategy. It's possible that increasing the number of days back caused the strategy to be slightly different.
For example, you had an entry on bar 1 of the original dataseries. Then you increased the number of bars back and the starting point for the strategy changed. On the new data series, you might have an entry on a bar just preceding what was bar 1 on the original chart. Thus, entry on that bar will no longer be possible.
So it really depends on your strategy. Just to show you that trades can easily be identical, please see the attached screenshots. They were created using the following code (you can also check that with, Bollinger Bands LE and SE, for instance or many other strategies for that matter):
---------------------------------------------
if date<>date[1] then buy 1 contract this bar on close;
setexitonclose;
----------------------------------------------
The behavior you are reporting is not a bug. Whether it occurs depends on your strategy. It's possible that increasing the number of days back caused the strategy to be slightly different.
For example, you had an entry on bar 1 of the original dataseries. Then you increased the number of bars back and the starting point for the strategy changed. On the new data series, you might have an entry on a bar just preceding what was bar 1 on the original chart. Thus, entry on that bar will no longer be possible.
So it really depends on your strategy. Just to show you that trades can easily be identical, please see the attached screenshots. They were created using the following code (you can also check that with, Bollinger Bands LE and SE, for instance or many other strategies for that matter):
---------------------------------------------
if date<>date[1] then buy 1 contract this bar on close;
setexitonclose;
----------------------------------------------
- Attachments
-
- daysback(30).png
- (41.93 KiB) Downloaded 449 times
-
- daysback(50).png
- (42.64 KiB) Downloaded 449 times
-
- Posts: 100
- Joined: Jul 17 2009
- Location: Germany
Marina,
I see what you are saying, but apart from Trades entered on Bar 1 and on the last Bar of the test-range, the inbetween trade should be identical. Look at my 2 screenshots, there are trades that are somewhere in-between for both "days back" values and should hence be identical, which they are not.
But OK, if you say this is correct then I believe you.
However, as I mentioned earlier, I did test the exact same strategy with same parameters in TS 8.6 and there the in-between trades are ALWAYS identical. So I still would say this is a bug in Multicharts.
I see what you are saying, but apart from Trades entered on Bar 1 and on the last Bar of the test-range, the inbetween trade should be identical. Look at my 2 screenshots, there are trades that are somewhere in-between for both "days back" values and should hence be identical, which they are not.
But OK, if you say this is correct then I believe you.
However, as I mentioned earlier, I did test the exact same strategy with same parameters in TS 8.6 and there the in-between trades are ALWAYS identical. So I still would say this is a bug in Multicharts.
- Andrew Kirillov
- Posts: 1589
- Joined: Jul 28 2005
- Has thanked: 2 times
- Been thanked: 31 times
- Contact:
Geektrader,
We tried more and we must admit that we can’t reproduce the mentioned behavior.
Could you send a strategy and workspace that demonstrates the issue?
We tried more and we must admit that we can’t reproduce the mentioned behavior.
Could you send a strategy and workspace that demonstrates the issue?
- TJ
- Posts: 7774
- Joined: Aug 29 2006
- Location: Global Citizen
- Has thanked: 1036 times
- Been thanked: 2233 times
My setup is this: a custom strategy which I unfortunately can´t publish ....
maybe you can whip up a simple dummy strategy to illustrate the bug?
I would be interested to understand the problem as well.