Hello all,
With the notice that I learned long ago that I should never blame something else but myself (dig deeper and the blame is on you), this time I can’t manage to blame myself. And so the question whether anyone has ever encountered something like this :
- The in Live running system does not behave according its source code, which can be retrieved from PRT itself. See first picture for what I mean. It makes profit all right, but it also makes losses I don’t recognize and via a strange kind of “trailing” which I could have made myself (I recognize it from other strategies), but which is not active in this strategy. In itself this is not important for you. The only importance is that I can’t mimic its behavior with backtesting in retrospect. Not by any means. The backtesting, however, shows how the strategy was intended.
To picture for you what happened in order to get to this situation, I will describe this for you :
- Back in December, I took one of my existing Systems (strategies) after it has ran for a year. Relatively to what I saw happening I did not much more than changing a setting available already in the code. Backtested that, re-optimized a little and gave it to ProOrder again. Of course I gad given the System a new name.
- For the months since, I saw strange exits, caused by a Limit above the price, which also is active when in loss. This latter is typically not me, but bugs always luring.
Today I finally sat down for looking for the culprit, took the .itf involved, and saw nothing of the kind what happened in Live many times.
- I opened the source as given to PRT (see first picture), I compared Parameters (which are there with fixed values) , searched for that setting in the code I changed back in December, and saw all was fine.
- Especially with the knowledge that I could not resemble the Live happenings with now backtesting in retrospect, I copied the source as given to PRT (1st picture) to a new strategy. Again backtesting was as expected. As an additional step I copied the existing .itf now including the Parameters, and pasted the source as given to PRT. No difference again.
I can only conclude that in ProOrder something else is running, and I can not see it.
In picture 2 and 3 you see that trailing at the wrong side of the price (look at the 2.83 contracts order), which *is* a technique I apply here and there, and undoubtedly is something from myself indeed. But not in this System and not intended to work when in Loss, but assumed this is something old and which worked wrongly in some past, it will just be my code. But again, not the code I gave to ProOrder in December, and this also not the code which I can retrieve today from ProOrder.
Did anyone ever ran into something which can only be explained by old code running or something, or at least something else running than intended ?
.
