From practice to theory and back to practice

 
And so, there are many theories and practice of them alas.
But everyone has their own practice, and I'm sure that everyone here has practiced more than one system. I'm even sure that almost every system was profitable for a while. But as soon as the system broke down, it could tolerate changes, but eventually it went into the trash.
 
So why not figure out what exactly happens to the systems, and when they profit and when they fail.
Running a couple of months ahead of time I can say that the results, tables and codes will not appear until two months from now due to lack of time. But before I start coding and testing it is interesting to hear the opinions of those present. I wonder who and how researched their systems. Was it just tests or were they real online trading reports, maybe someone studied the signals of their systems and could see something.
 
Why did I bring up the topic of research practices? It's quite simple. When we run tests, we see the Wediv, for example, and do not consider it further. But, sometimes before it goes down the balance increases. So relying only on the test chart we may lose the potential that was expected when we created this grail.
 
Let's try to take apart the most ordinary flip-flop system on some ordinary indicator. Let it be a two-wheeled system. And the signals for action will be crossing the slow one with the fast one.
 
Anatolii Zainchkovskii:
Let's try to take apart the most ordinary flip-flop system on some ordinary indicator. Let it be a two-wheeled system. And the signals for action will be the slow one crossed by the fast one.
What can happen to the systems? They always work the same way and invariably according to the algorithm.
 
I think many have tested such a system, and many have selected parameters for optimisation, some have probably implemented stoplots and takeprofits.
 
Vitali Kadel:
What can happen to the systems? They always work the same way and invariably according to the algorithm.
I agree, according to the algorithm. But here is an example of one deal, the system signaled a buy, the algorithm performed it and put stops, and here is the most interesting part. The deal seemed to be profitable, but then it turned around and went in the red. And most likely, it lost on the stop loss.
 
Few people probably dwell on what happened with a trade, everyone is usually interested in what the system will show next, because we all understand that not all trades can be profitable, and we just want the balance to grow.
 
But let's try to analyze each signal, how much after entering the trade could be the maximum profit and how much could be the maximum drawdown. So we need to determine the values for each signal of the system, getting the time series of signals we can build two charts first for profit the second for drawdown.I think we will see how these charts wave, in theory it will be two oscillators. But let's go back, usually the system has fixed stops, their charts are just straight lines.
 
So, continuing the theme, after we display charts of unrealized profits and losses of each signal, we can at least see where it was reasonable to put a stop, that's one. The second point, after seeing these charts someone will probably say: I would not have entered in this trade, but I would have entered in that one.
 
So, I think that having such data you can filter out those deals that led your and my systems to losses. But please understand that systems that are losing on a martin will not work here. There is a different error there.