Non-fitting system - main features - page 2
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
Makes sense - seconded. In general, the optimisation for non-different - is an alchemical flask. They make a TS out of a wild bunch of poorly understood indicators, input their parameters into the external tool, and it's a shuffle. They will see how it works. But what if it is a grail? Perhaps, the alchemists were searching for the Philosopher's Stone.
Naturally, there is the probability (I forgot - it is a known figure) that a monkey, accidentally pressing the keys, will play "War and Peace".
===
Maybe we should look at the problem from this perspective: what is the purpose of optimization? And what isn't.
Well that goes without saying, no alchemy necessary.
A sliding window of optimisation... although this does not give a complete guarantee either. But the robustness (in the sense of low sensitivity of the system to changes in the input parameters, and the price is the same input parameter) still increases. Well, yes, the statistical importance of a window should be at least something.
Nobody is talking about profit guarantees on forwards. We are still talking about culling known unprofitable TS.
This is due to the fact that the larger the profit and the smaller the drawdown, it means that the TS has learned the history too well, respectively, in the future it is likely to work badly, because the market anyway will be different ......
No one is talking about forward profit guarantees. It's about culling known unrobust TCs for now.
Well if you don't require guarantees, then - what I wrote about. It eliminates a lot of things. But there is one problem, it is more convenient to do it on self-written tester, emtesh is not very suitable for this. And considering the results is not in the form of a single graph, but in the form of a field of results.
From my observations, if most (more than 80%) of any random combinations of variables in an EA give a sustained profit, then the probability of not fitting increases significantly.
No part of any combination will ever give a sustainable profit. Markets are non-stationary by nature. So all the mythical terminology about supposed "stability" and so on. "guarantees" are irrelevant in this context.
Another amateurish thought.
If we take the parameters by which the TS becomes robust to the outside, it is even worse than a flawed system.
It is definitely a flawed system.
I would refer those data that appears in the news to external parameters.
It's strange, but the market sometimes reacts to them.
For example, I haven't seen an EA that reacts to a change in the discount rate of an instrument yet.
And this is not a big deal. ;)
A variant of amateurish performance:
We choose a short period for optimization (say, 3 weeks) 2-3 weeks before today and optimize it. We run the best results through all available history and look at them. If the nature of the curve before and after optimization is similar to that of the optimization period, MTS has a right to refine it. I give more importance to the drawdown than to the profit factor when assessing it.
There is a non-amateur version...... truth the effect is the same :)
- optimization, on period N
- recording results in a file
- analysis in excel
- writing the parameters of acceptable results in another file
- testing of Expert Advisor with these parameters on a tank and forward N/2
- selection of results comparable to results obtained during the optimization period (essentially the same "nature of the curve") ....
- selection of a single one and its final optimization......
I'm tired of writing :) .... But it's a problem with the result, although everything seems to be accounted for and random results are rigidly cut off.....
the width and smoothness of the optimum zone. The same PF should not jump too much near the optimum value. But it is not suitable for all parameters.
If there is a filter that filters out some of the trades, then if we tighten it, the quality of the system (PF as an indicator) should increase while the number of trades decreases.
Well, that's self-explanatory, no alchemy necessary.
OK. Then I ask (everyone) to answer me the retarded question: what is the purpose of optimisation?
If we are looking for optimal parameters, then what is this if not a fitting? Or are we using the optimiser to explore the TS itself? A test drive like that.
Then, maybe focus on the very criteria of TC research by means of optimization?