All the same, to properly test in Amibroker, you need to be extremely savvy and attentive, since such inconsistencies in real trading can come out, from which other programs have protection. For sure, therefore, the testing speed due to this and exceeds other programs. Here I am, for example, I constantly check tests in Amibroker with Velslab and if it weren't for Velslab, then almost every time it would be guaranteed to incorrectly test in Amibroker, then one thing should be noted in the settings, then correct in the code . Who is the guru at Amibroker, that, of course, already knows how to do everything right. Но уверен, that newbies are definitely doing something wrong and because of this the system either produces results with a peek into the future, or with distorted results, especially in portfolio tests.
Here, for example, yesterday came up with a new system in Velslab. I tested it on current promotions and made sure that the results are quite attractive, at least on historical tests. Recoded to Amibroker in order to check on ALL existing and retired shares, tested and the test gave a uniform and confident drain. I began to understand what the problem is. It turned out the following:
Eight positions from the stock list can be opened at any given time. Positions are closed at the opening of the next session, if there is a signal to close them. If, for example, eight positions were opened and there are signals to close two positions, then six positions remain and we have the right to close two positions at the opening of the next session and simultaneously open two new positions (if there are signals to open, certainly). All this happens correctly in Amibroker.. But if some positions are closed by stop during the session, then actually by the end of the session we will have less than eight positions. And the Amibroker, closes positions by stop during the session and instead of them immediately opens new positions at the opening of the same session — that is, looking to the future.
Maybe, there are some tinctures in Amibroker to prevent this outrage, but a beginner is unlikely to notice this moment in the mass of transactions, especially if there are several tens of thousands — does not even think that this can happen. I wouldn't have Velslab, and I would not have noticed such a significant discrepancy in the test results, and would not even suspect a catch……
