Auto or manual - page 14
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
Trying it by gut feeling is a dead end! This is what the "league" shows...
If this way worked, a good result would have already been found in the meantime... And so: today this adviser ..., tomorrow another ..., then the third ...
And mind you, even groups of advisors don't work... EVERY year, year after year.
That's the thing about "not by gut feeling", isn't it?
It has been correctly stated here - "rubbish TCs". The League experience shows me that the 80/20 rule works in full force - only 20% of the systems show something, the rest are crap. There are three divisions in the League, well, more than half of the experts have never made it to the top division. That is, their profit periods are much shorter than their loss periods, and on all variants of the input parameters. In other words, some types of TS do not fit some symbols. And this allows us to not make a rough choice of Expert Advisors, but at least to reject such TCs that, no matter how much they are overoptimized, will never reach the leaders.
That's the thing... In testing - the system seems to show good results, corresponding to the Middle Division. But when I set it on demo trade, it even stays in the middle division for a short time, first declining to the lower one, and then it shows inadmissible behavior and is directed to overoptimization. Once again the situation changes as if during optimization the results were not very bad, but after being submitted for demo trade the programmer quickly goes in the red.
At the same time good TS work in a different way. On the demo-period they, as a rule, show excellent results, and when set in the higher division they may start to go down, sometimes even go down to the lower one, however, after that they still go up bringing them to the higher one again. And such TCs are over-optimised much less frequently. If one manages to switch off these systems for the duration of the loss, one can be constantly in profit.
Well, that's the point, it's "not by trial and error".
It was correctly stated here - "rubbish TCs". The League experience shows me that the 80/20 rule works in full force - only 20% of the systems show something, the rest are crap. There are three divisions in the League, well, more than half of the experts have never made it to the top division. That is, their profit periods are much shorter than their loss periods, and on all variants of input parameters. In other words, some types of TS do not fit some symbols. And this allows us to not make a rough choice of Expert Advisors, but at least to reject such TCs that, no matter how much they are overoptimized, will never reach the leaders.
That's the thing... In testing - the system seems to show good results, corresponding to the Middle Division. But when I set it on demo trade, it even stays in the middle division for a short time, first declining to the lower one, and then it shows inadmissible behavior and is directed to overoptimization. Once again the situation changes as if during optimization the results were not very bad, but after being submitted for demo trade the programmer quickly goes in the red.
At the same time good TS work in a different way. On the demo-period they, as a rule, show excellent results, and when set in the higher division they may start to go down, sometimes even go down to the lower one, however, after that they still go up bringing them to the higher one again. And such TCs are over-optimised much less frequently. If one manages to switch off these systems for the duration of the loss, one can be constantly in profit.
Look, go to the rubbish factory with the rubbish. You don't know how to trade on principle.
Yeah. I can't, that's the thing... That's why I'm selecting robots that can.
Yeah. I can't, that's the thing... That's why I'm selecting robots that can.
It was correctly stated here - 'rubbish TCs'...
That's the thing... In testing - the system seems to be performing well, corresponding to the Middle Division. But when I set it on demo trade, it even stays in the middle division for a short time, first declining to the lower one, and then it shows inadmissible behavior and is directed to overoptimization. Once again the situation changes as if during optimization the results were not very bad, but after being submitted for demo trade the programmer quickly goes in the red.
At the same time good TS work in a different way. On the demo-period they, as a rule, show excellent results, and when set in the higher division they may start to go down, sometimes even go down to the lower one, however, after that they go up anyway, reaching the higher one again. And such TCs are over-optimised much less frequently. If one manages to turn off these systems for the duration of the loss - then one can be constantly in profit.
You have to think/experiment with optimization criteria. Profit, balance, DD as criteria lead to over-optimization/fitting. For training/evolution purposes I managed to synthesize criteria that help to get robust TS (or TS parameters). That allows to have a replacement/replacement before some TC, as you say, from Higher Division. By the way, think about it, maybe you should not wait for drawdowns, but really replace them earlier, although your criteria are not sharpened for this.
That's the thing about "not by rote".
Well, what do you call it? The selection of EAs is based on indicators, which are themselves TRUE data...
There is no SYSTEM that at the time of EA development, gives the planned results, i.e. on the PERIMENT data...
SELECTING secondary indicators doesn't do anything ... You can get scared and get stuck on the right Expert Advisor, and that's the "rule of thumb" ...
Well, what do you call it? The selection of EAs is based on indicators, which are themselves TRUE data...
There is no SYSTEM which, at the time of EA development, gives the planned results, i.e. on the PERIMENT data...
SELECTING secondary indicators doesn't do anything ... You can get scared and get stuck on the right Expert Advisor, and this is a "gut feeling" method ...
So there is no data at all, except for tick prices, which are secondary as well.
Yes, you're right! Tick data is not up to us....
But everything else should be in a SYSTEM... An EA is a SYSTEM product, which depends on primary data laid down by a trader.
You don't go over primary data using gut feeling ... Without a systematic approach to developing an EA, you won't get anything good out of it.