The essence of optimisation - page 9
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
There is a suspicion that a robust variant choice (without such a choice the optimization is not an optimization, it has already been dealt with) for an unknown process (and presumably not random) which is the market, can be made by approximation of a statistical set of parameters working at different parts of the optimization, following one another.
What would it look like in practice?
And how will it be different from the usual optimization with a search of variables one by one? In any optimisation, something is approximated. And even if you find that on one part of history one set works better than another - what next?
joo, 2014.03.31 22:05
There is a suspicion that selecting a robust option (and without such a choice optimization is not optimization, this has already been dealt with) for an unknown process (and presumably not random), which is the market, can be done by approximating a statistical set of parameters working on different parts of the OOS, following each other.Paragormon:
One way or another, everyone is starting from the postulate of market inertia. My point is that no one argues about other properties of the market and about optimisation according to some other principles.
Yes!!!, but but...
Inertia is definitely the main principle, but it should not be confused with its physical counterparts, which are quite simple and have long been inapplicable to the market.
"Inertia" is the so-called "signs" that summarize in the most concise way (in the algorithmic sense) the superposition of herd motivations of trading participants (humans and bots), their prevailing anticipations for the near future, which are obtained by statistical generalizations of the past, they are the ones that shift MO. Such statistical patterns of majority reactions to typical market states do change relatively smoothly, but faster and faster. The reason is that the whole crowd cannot change response patterns at once and the effect is sort of Gaussian blurred in time, due to the failure of some postulates about an efficient market, in this case about instantaneous equal informability and the same information processing algorithms.
But these are not Dow Jones-style patterns, autocorrelations of the first lag. Things are now more complex and changing much faster.
What would it look like in practice?
And how will it be different from the usual optimization with a search of variables one by one? In any optimisation, something is approximated. And even if you find that at one point in history one set works better than another -- what next?
don't know yet... working on it...
And why would you kill an Expert Advisor if it is still trading on the plus side?
and uses in its logic the most "inertial sign". which time blur by Gauss is more or less long?
Let him use it as much as possible.
...The ever-living population of EAs comes to mind, a few of those in the top population are trading...
Have you decided on the number of EAs in the top?
If, for example, there are several thousand EAs in the population.
I am thinking of limiting the top to 25 instances. Which will trade simultaneously and each will risk 2%.
And the margin will be divided in order of priority of their location in the top, but in total not more than 50% of the entire depo margin.
And why kill an EA if it's still trading on the plus side
and uses in its logic the most "inertial sign" which blur in time according to Gauss is more or less long?
Let him use it as much as possible.
Better tell me where will the robust cream come from if the genes have wizards and the like.
IMHO the genes should not be built on wrenches.
It seems to me promising to use statistical dependencies of movement of one tool relative to another.