Machine learning in trading: theory, models, practice and algo-trading - page 1500
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
It looks like some kind of grail is coming to me. Testing was done on 28 pairs and 4 timeframes (5-15-30-60). The final result of testing (profit in pips) is given below. However, I have tested it not in a tester, but programmatically, and without the spread. I have created a multi-currency Expert Advisor for it. I trade it on the demo. If it starts moving in the plus direction I will post the signal for review.
I have also invented two new predictions for the Expert Advisor:
1. Calculate the profit of crossovers of two bars for a certain period and submit it for input.
2. input the crossing prices or the difference (ratio) of the current and crossing prices
It looks like some kind of grail is coming to me. Testing was done on 28 pairs and 4 timeframes (5-15-30-60). The final result is given below. However, I have tested it not in the tester, but programmatically and without the spread. I have created an Expert Advisor multi-currency Expert Advisor. I set it on a demo. If it will start fluctuating in the plus direction I will add the signal for reviewing.
It is always like that on the learning one. But nobody has it on the test one.
It looks like some kind of grail is coming to me. Testing was done on 28 pairs and 4 timeframes (5-15-30-60). The final result is given below. However, I have tested it not in a tester, but programmatically and without the spread. I have created a multi-currency Expert Advisor for it. I have tested it on demo. If it will start fluctuating in the plus direction I will post the signal.
I have the same grail if "viterbi" is considered as a ready function from the package and it is one euras m5 instrument.
But if I simulate real trading where the model receives only the last data, then it fails ((
I obtained 100% similarity with results of the "Viterbi" batch function if I train the model with data in the sliding window, but then there is a delay in the signals + - for the size of the window and the model does not make money
So I highly recommend to simulate some kind of "real trading" when data feed into the model is exactly the same as in reality
Is this on the training section or on the test section (forward)?
On the training one, it is always like that. But no one has it on the test one.
Have you tried to do something with the indicator on the machines that I downloaded here?
Well, it turns out that in R the "grail" button has already been invented, while in mql it hasn't been rewritten yet.
golden words! I have written about a hundred times in various threads that the most valuable thing in MT4 / MT5 is the strategy tester, it quickly destroys any graphic assumption or TS
))))
SZS: there is a grail in mql, it is enough to apply your articles ;) I think it is a pity that MQL-developers reduced all their current work to optimizing and speeding up the terminal and MQL5, while the standard libraries have not been replenished with new MQL and NS packages for many years
It's a pity that the MQgL developers reduced all their current work to optimizing and speeding up the terminal and MQL5, while the standard libraries have not been replenished with new MO and NS packages for years
ZS: there is a grail in mql, enough ..........
You can take a look at???? ;)
The question is rhetorical...