A balance between QUALITY and QUANTITY - page 2

 
Prival:

The emphasis on points is not necessary, because the tester shows profit and drawdown in the currency, through which you can calculate the same points, and the ratio of profit to drawdown in the currency, will be the same as the profit to drawdown in points.

In short, I don't really understand the points...

 

One of the most important criteria is the maturity expectation. I would put it at the top of the list in general.

Note how many "grails" that earn a little bit in the tester lose on the real account :) Because of slippages, requotes, spread widening... And it's very unfortunate that the execution in the tester is "Instant".

Turning to the criteria, the second place is the drawdown. In third place - profit.

 

The question is interesting, especially since I have just started the tester version of the TS, at the same time I'll PR it :).

I start from the potential profitability of the TS. To estimate it I introduced the concept of "minimum deposit". This is none other than the maximum drawdown when testing with a fixed lot multiplied by the reserve coefficient, I take it equal to 1.5. Now having a minimal deposit we can estimate the maximal profitability of TS, just dividing the profit obtained in testing by this minimal deposit.

What should be the maximum profitability?

It is only obvious that it has to be higher than the bank's, otherwise there is no point in getting into such a high risk instrument. As for the upper bound, it is, imho, a matter of faith. I, for example, do not believe in grails, so even 50% per annum I am very suspicious. And all that has a yield higher than 100%, just throw away at once, as would throw away a perpetual motion project.

Fortunately, my TS fits within reasonable limits. Here is the balance for 11 years on a fixed lot, it turns out about 20% per annum.

Here's the same 11 years with MM

root extraction yields about the same 20%.

However, the problem is that the minimum deposit turns out to be about 20K, well, a little less. Moreover, despite all checks and no plausibility, it (like any other TS) will always be a high-risk investment, that is, the hypothetical investor these 20K should make only a small part of his portfolio.

 

Prival: ... 2. главный критерий после входа в сделку цена не должна идти против вас.


And what do you understand by "going against you" ? Going "with us" all the time is impossible for her. Or do you mean the very first moment of opening a trade?
 
sever30:

The emphasis on points is not necessary, because the tester shows profit and drawdown in the currency, through which you can calculate the same points, and the ratio of profit to drawdown in the currency, will be the same as the profit to drawdown in points.

I don't really understand the concept of pips...

Let me try to put it in other words. If you find such a TS (ideal by my (described above) criteria), it's the Holy Grail, in its purest form. You can enter the trade up to your tomatoes with the whole deposit.

You have to strive for the ideal, for risk-free trading.

 
Diamant: One of the most important criteria is the maturity expectation. I would put it at the top of the list in general.


Look at the results of test item 3 in my first post. Expectation is maximal, but there are only 4 deals. What kind of statistics is that? We are not even talking about statistics. This EA has been tested on a 5 year period. By the way, the testing period is also an important parameter.

There needs to be a single criterion for the quality of transactions.

 
Richie:

And what do you understand by "going against you" ? Going "with us" all the time is impossible for her. Or do you mean the very first moment of opening a trade?

Yes exactly the entry point. It should not go any lower. That is the single (the only one in my mind) criterion for the quality of trades.
 
Prival: yes exactly the entry point. It shouldn't go any lower than that


This is an interesting thought. If the price goes up so much at the entry point that it cannot even return to the entry point, it means that the train is "fast". If the train is "fast", it will not be able to stop immediately and even if the brake is fully applied it will move another "two hundred meters".

 
Richie:


Look at the results of the test p. 3 in my first post. Maximum expected payoff, but the transaction is only 4. What kind of statistics is that? We are not even talking about statistics. This EA has been tested on a 5 year period. By the way, the testing period is also an important parameter.

There needs to be a single criterion for the quality of transactions.


Of course, you are right, 4 trades in 5 years is not a statistic, but just a fluke. But look at the other 2 results. The 0.88 mathematical expectation may not be considered at all - don't you agree? Unless we are talking about trading 0.01 lots :) 46 and something is still there and there.

But don't you think that out of these 3 results ALWAYS none of them can be considered at least somewhat satisfactory?


It is not easy to formulate a single criterion. Probably, we should start a little bit with other things. First of all, we must evaluate the "physicality" of the TS.

That is, if on optimisation (disabling genetics) we observe stability of result on some range of parameters. This will allow to sift out random luck with parameters :) and to some extent eliminate fitting on history.

Then we will see what the result is. The system may gain points in a certain reporting period and this result is stable during the testing period (otherwise it may turn out that one year of profit and another year of loss). Well, in short, how stable it is. From this we must draw conclusions. So it is like this.

 

I also consider the maximum absolute drawdown to be the most important criterion. I estimate the drawdown distribution.

The second criterion is the statistical validity of the system test results, i.e. how much it is not fitting, how much the real drawdown will be higher than the test one.

These two criteria provide the basis for calculation of the working capital.

Profitability is a property of the strategy, profitability does not need to be optimized, profitability is formed naturally or we achieve it by improving the strategy.

I visually check the smoothness of the equity curve, it gives a general idea about the distribution of groups of trades, the distribution should be even, otherwise it is another fitting. This is enough for me for now, but of course I am thinking about how to optimally evaluate the evenness of trades, profits, losses in a mathematical way.