Making a Python trading system for MT. - page 7
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
Of course you're right. Just in case, I'll check this branch in 6 months to see what I can guess. Good luck,
Thank you. I don't think that will be necessary. In 6 months this branch will be stalled deep in the basement.) Not Destiny.
The results of the first Python test were published yesterday. For your convenience, so you don't have to flip through the pages, I will repeat the graph again without any changes.
Perhaps to many people the results will not seem well. However, let me remind you that as of today the GO on Sbera futures is 3583.94 р. That is, the system has earned about 1700 r for 3 months, which amounts to 47.4% of GO (or, in your terminology, the deposit). In one month, it will be - 15.8%. In Forex terms it may not roll, but for Forts it is quite normal profit.
And this is an absolutely crude system without any tuning, with the most primitive algorithms of tracking and even position opening. The system already now, in this form, can be put on the market, and the results will be approximately similar to the test.
However, we are not going to release the system to the market yet, but to find out how promising the strategy is for further improvement, i.e. whether it has a potential for further development. For this purpose, we initially performed simple entries and primitive deal tracking to see how much can be squeezed out of the strategy. For the same purpose, we wrote in the Strategy Tester report not only open-close deals and profit in deals, but also the maximum profit in each deal.
Just add these maximal profits and we will see what we can ideally get from the strategy. We got the following one - 11220 points. I.e. we currently use a bit more than 1/10 of the strategy possibilities in terms of profit. By improving the strategy it is real to get another 30-40% of the maximum.
What can be done immediately, and without even thinking about it.
1. Close all deals through the night (it is an intraday strategy).
2. To avoid gaps, start trading 5-10 minutes after the market opening. Depending on the situation, of course.
3. Prohibit trading at low trading volumes. This is usually at the evening session.
This by itself will give results. And all of this must be done in the first place, and only after that one begins to improve the algorithms of opening and tracking of trades. All this is not taken into consideration in the strategy so far, and it is indiscriminately fed with the entire flow of quotes.
That is exactly what I will be doing for now.
Just yesterday I said that the system could be launched on the market right now, and I guess I was not mistaken.
I managed to perform only a part of measures mentioned in the previous post, and at the same time I decided to test the system with new data. The previous chart was on the futures - SBRF-12.17, the next one is on the futures - SBRF-06.18. No changes in algorithms and parameters of direct opening and maintaining of trades were introduced.
And here is the picture itself:
As before, x is the trade number, y is the total profit in pips. Trade with a fixed lot - 1 futures SBRF-06.18. Time frame - 1 min. Testing period - 3.5 months.
We still see the initial entry section where the futures have low trading volume before the previous futures close and the initial period after the previous futures close.
yep, and again the old graphs are not from matplotlib
No, the graphs are new, just from under the chicken. But they are not built in Python, but from CSV reports. For fast plotting in Python not everything is ready yet.
And, if you read the thread carefully, I wrote that the old tried and tested system with some innovations will be ported to python.
But in python, yes, will be exactly from matplotlib.
No, the graphs are new, just from under the chicken. But they are not built in Python, but from CSV reports. To build graphs quickly in Python is not all ready yet.
And, if you read the thread carefully, I wrote that the old tried and tested system with some innovations will be ported to python.
quantopian.com has a tester if you're interested. And they also fund successful strategies
the tester can be plugged in as a libu or directly on the website
quantopian.com has a tester if you are interested. And they also fund successful strategies
Your own tester is more interesting, because you can fully control the process, which is absolutely necessary for testing. And the tester itself is a very simple construction.
I don't see the point of funding anything, because I do it only for myself.
Having made such a profit -13000 points instead of 1700 points in the previous test and the predicted limit for the strategy of 11220 points, of which I planned to take 30-40% more, I was really puzzled - where did the money come from? All right, some activities should give something, but not that much.
The solution was simple. The market has changed. With the same other system parameters, the standard deviation on SBRF-12.17 was 47 points. On SBRF-06.18 it was 100 points. Plus, the trading volume has increased, which is a more complete market and less dynamic (in terms of noise) and more predictable price that reduces the entry error. All in all, no miracles.
Having made such a profit -13000 points instead of 1700 points in the previous test and the predicted limit of 11220 points for the strategy, of which I planned to take 30-40% more, I was really puzzled - where did the money come from? All right, some activities should give something, but not that much.
The solution was simple. The market has changed. With the same other system parameters, the standard deviation at 12.17 was 47 pips. At 06.18 it was already 100 points. Moreover, the trading volume has increased, it is a more complete pack and less dynamic, in terms of noisiness, and more predictable price that reduces the entrance error. All in all, no miracles.
So, where do we enter? Down, or up?
So where do we go in? Down or up?
We enter so as to make a profit to the right of the current price)