Non-fitting system - main features - page 3
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
OK. Then please (everyone) answer me a dumb question: what is the purpose of optimization?
I agree. I was talking about this at the beginning.
Optimisation is not necessary. and the first tests can be done on the tester. (tautological but true)
And the manifest is about that.
>> Who reads it?
OK. Then please (everyone) answer me the retarded question: what is the purpose of optimisation?
If we are looking for optimal parameters, then what is it, if not a fitting? Or are we using the optimizer to investigate the TS itself? A test drive like this.
Then maybe we should focus on the very criteria of researching the TS by means of optimisation?
I agree that optimization is more of a statistical gathering - a robustness study of a trading idea.
I'll throw in a few bits and pieces as well:
1 Assertion. Any system is in itself a market optimisation. We choose favourable entry and exit points and protect ourselves with the rules of mm. Right now I'm writing a system that has not a single parameter to optimise at all in the tester. But even it has at least four parameters: entry, exit, position maintenance, mm management. All these parameters are market optimisation.
2 Assertion. It follows from the first one. There is no non-optimized trading system (manual or mechanical). The only exception is a completely random system, in which you randomly select the volume of a position, the time of opening and closing.
3 Assertion. Since optimization is invisibly present in any trading system, then talks that any optimization is bad make no sense.
4 Assertion. I thought many times that the market works this way but the optimization shows that the market works in a different way. Therefore, assertions that the system parameters must be selected solely on the basis of logic and reason are wrong.
Assertion 5. It follows from the previous one. The main task of optimization (for me, at least) is to find a stable state of system development, regardless of the market (with some reservations) and timeframe (also with some reservations).
The presented "statements" are purely my IMHO, not claiming anything, so please don't peck too hard.
Ok. Then please (everyone) answer me a stupid question: what's the purpose of optimization?
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 this.
Of course it is a test drive. At least with me. And without genetics switched on, full enumeration of parameters, field of results, selection of a stable, smooth zone, separate miscalculation on logn\shorts\all together, window shifting, all over again. And what the optimizer finds is the first run, just to see how it is in general.
There is no single criterion. More precisely, it's dangerous to limit yourself to one criterion. Especially when it comes to betting on the real.
I'll throw in a few bits and pieces as well:
1 Assertion. Any system is in itself a market optimisation. We choose favourable entry and exit points and protect ourselves with the rules of mm. Right now I'm writing a system that has not a single parameter to optimise at all in the tester. But even it has at least four parameters: entry, exit, position maintenance, mm management. All these parameters are market optimisation.
2 Assertion. It follows from the first one. There is no non-optimized trading system (manual or mechanical). The only exception is a completely random system, in which you randomly select the volume of a position, the time of opening and closing.
3 Assertion. Since optimization is invisibly present in any trading system, then talks that any optimization is bad make no sense.
4 Assertion. I thought many times that the market works this way but the optimization shows that the market works in a different way. Therefore, assertions that the system parameters must be selected solely on the basis of logic and reason are wrong.
Assertion 5. It follows from the previous one. The main task of optimizing (for me at least) is to find the stable development of the system, regardless of the market (with some reservations) and timeframe (also with some reservations).
The above "claims" are purely my own opinion and do not pretend to be anything, so please do not make too much of it.
I support the separation of the term optimisation of the system state at a particular point in time (this includes the current currency position itself, the market condition its trend etc.)
- the system itself has to do it!
and the so called optimization of input parameters in MT.
These are different optimisations (added)
Heaven and earth! (or rather a lame swamp near the chemical plant)
I'll throw in a few bits and pieces as well:
1 Assertion. Any system is in itself a market optimisation. We choose favourable entry and exit points and protect ourselves with the rules of mm. Right now I'm writing a system that has not a single parameter to optimise at all in the tester. But even it has at least four parameters: entry, exit, position maintenance, mm management. All these parameters are market optimisation.
The market cannot be optimised - it is a fact given to us in our senses. Our relationship with the market - yes - can be optimised.
2 Assertion. It follows from the first one. There is no non-optimized trading system (manual or mechanical). The only exception is a completely random system, in which you randomly select the volume of a position, the time of opening and closing.
It's hard to argue with that. Unless, of course, we concentrate on "optimising the market".
3 Assertion. Since optimization is invisibly present in any trading system, then talk that any optimization is bad makes no sense.
Do you know what a syllogism is? It is a statement where all logical conclusions follow from an incorrect premise.
4 Assertion. Many times I thought that the market works this way, but the optimization shows that the market works in a different way. Consequently, assertions that system parameters must be selected solely from outside, based on logic, and reason are wrong.
Yeah... And I gave the example of the monkey as a joke. But there it is...
5 The assertion. It follows from the previous one. The main task of optimization (for me at least) - is to find a stable state of system development, regardless of the market (with some reservations) and timeframe (also with some reservations).
It's an interesting bouquet of words - a stable state of development of the system. Only the "information-energy field" is missing for completeness of feeling and style. (Ask the healer Epiphany.)
The presented "statements" are purely my IMHO, not claiming anything, so please don't peck too much.
I haven't started yet. Although ... what is there to start. IMHO is IMHO.
Since the market cannot be optimised in principle there cannot be any trading system that shows better results than a random number generator.
Swinosaurs, please don't comment on my posts anymore. I just can't have argumentative arguments with fools (no one can).
There's a book on creating good systems and testing them properly.
There is a book on creating and properly testing good systems. There is not a single sufficient criterion for robust system, but a big pile of necessary ones. I'm talking about Pardo's book (there's a branch created over 2 years ago, but never brought to its logical conclusion).
Forward analysis (probably what HideYourRichess called "sliding window optimization") - is just a model of changes in external parameters in real time, leaving the system in some stable area. It provides no guarantees either, but a system that has passed the forward analysis (along with dozens of other tests and criteria) has a significant chance to be robust. And in this approach to system design, external parameters do not have to be logically valid (we are not discussing a perfect system now).
Optimisation is useful, but it is a whole set of actions in which finding a sufficiently wide stability zone is not the last step of system design.
I haven't used an optimizer for a long time, as I try to look for ways to avoid arbitrary external parameters that would have to be optimized. Well, for example: if the first version of the system uses a scale with a period of 20, then I try to extend the system so that it uses other scales with periods in the range of, say, 5 to 40, on the same rights. This is also parametrization, but still far from as rigid as simply fixing the number 20 as extern.
Since the market cannot be optimised there cannot in principle be any trading system that performs better than a random number generator.
Swinosaurus, please don't comment on my posts anymore. I just can't have argumentative arguments with fools (no one can).
I am not going to comment. Can I call you names?))
===
What kind of comments were you expecting on your pseudo-scientific opus claiming to be marbleized? Like, slow down, we're taking notes?
You know, if you have already said a stupid thing - it does not happen to anyone, then it is the last thing to behave according to the principle "You are a fool". You seem like a sane person.