Yoghurt systems and canned systems or The relationship between trading tactics and the reliability of historical test results - page 18
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
to Prival
I enjoy reading your posts. With one "but", as usual :)))
We are not in a battle where you have to win - it is more important not to lose.
I don't think speed of decision-making is the most important thing - it's a projection on the WWE, but that's not what Forex is about at all.
Look at an elementary (any) zigzag in hindsight - is speed the most important thing? :)
Not trying to change your mind..... just my point of view. And, I too am a military man who has had to operate many of your (not personally, of course) instruments, some of which I was far from thrilled with :)
Glad to see the military, I think you'll find this one very clear and bilsko 'Random Flow Theory and FOREX' my post there starts in red. I've done a lot of TS over the last three years only 1 of them was loss making, but so far I'm not out for the fight yet. So it's not a 'but' but yes. I think our regiment has arrived. Good luck. Promised to have a new topic by Monday, I need to think about drawing. So I'm off.
Speed is important, first come, first served.)
A Pipswitch is an Expert Advisor which has a profit of 1-5 pips.
>> Thank you.
I would like to propose for discussion the issue of the relationship between trading tactics and the reliability of the test results on history.
When creating a trading system, we rely on some regularities depending on the input variables. We often optimize the system according to these parameters at the testing stage and rejoice when we get good results. And then it turns out that the system starts to fail after a week. Some systems only start to fail after a few months.
For optimisation = fitting
I suggest and discuss the factors that affect the "shelf life" of systems. Why some systems are yoghurt, others canned tuna,
For different optimisation parameters = fit
and is it possible to create a perpetuum mobile or at least come close to it?
- Tell us please, haven't you found any ancient manuscripts or writings during excavations?
- Yes, you know what's interesting, we've even found ancient manuscripts or writings in excavations...
(c) COMEDY CLUBI think you can create a perpetuum mobilee, or at least get close to it, but through optimization = tweaking you'll only get close to yoghurts, tuna at most.
All IMHO of course, but for me personally - the further you go, the harder IMHO it is.
Blindness a priori = profitable parameters from one TF are unsuitable for another TF.
There are completely blind systems)), but there are no completely sighted ones yet, because
-either the comp must have consciousness (for this complete sightedness),
-or the rules of the game must be clearly formalized (e.g. chess).
If the computer has no consciousness, then the maximum that can be obtained is a slightly sighted "canned".
From this we get the following:
optimization != fitting=TRUE
There are blind and there are sighted systems.
.........................
If the computer has no consciousness, the maximum you can get is a slightly sighted "canned".
From this we get the following:
optimization != fitting=TRUE
Alexander, I'm sorry, I didn't catch your point. How did you manage to derive the formula from the vision of the system:
optimization != fitting
Alexander, I'm sorry, I didn't catch your point. How did you manage to derive the formula from the vision of the system:
optimization != fitting
Please Igor!
a) -Indicators built on averaging are low frequency filters.
The intersection of two averaging indicators is a difference circuit or a differential, i.e. a "high frequency" filter (in quotes, because the high frequency filter is obtained indirectly), the TC with inputs and outputs at the intersection of averaging is in fact a mid-pass filter.
b) -So we cannot hope that the frequency, and bandwidth, Q-factor of the filter selected for the tested section will continue in the future.
Optimization of such midrange filter will not be a 100% fit, but only one third))), because the TC still "sees" in its range both input and output.
Here we do not interfere with the decisions of the TC, but optimize the width of the capture and the catch location, i.e. do not tweak the automatic action of the system, do not adjust for the result.
c) - Fitting starts when we start to break the "vision" of the TS, in particular by setting a fixed take.
On the one hand the TS will go blind due to the definition of a hard exit, but on the other hand the profits may increase, and that's the real fit,
Because the size of a profitable hard profit is typical of history, but not for the future, but then again in the future the EA won't adjust the take by itself,
and such a "broken fit" EA is a landmine for the client = a perfect fit.
This is where I disagree. There is "over-optimisation" - this is when the parameters of the TS during optimisation are really adjusted to the historical data so that the TS is not capable of working in the future. Or rather, of course it can, but it will not bring profit. This is when the optimization is completed in order not to become overoptimized or what option to take out of those ones provided by the MT4 optimizer in order for the TS not to be overoptimized. I think many have noticed that if you take the best variant of optimization, it usually does not work well in the future.
to LeoV
If there are no blind - fit parameters in the TC, over-optimisation is unnoticeable.
to LeoV
If there are no blind - adjustable parameters in TC, over-optimisation is unnoticeable.
Blind - what does that mean?