Machine learning in trading: theory, models, practice and algo-trading - page 1244
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
As for the topic of MO itself, its very long-term existence shows the complete futility of the approach: MO as a trading system. Studying the elephant's excrement, there is not much to say about the elephant itself, much less predict it. And quotes, nothing else.
At the moment indicator-logic systems, with less expenses, show quite real results. So maybe we should use MO in such systems. Let's say, forest-trees - already prepared teachable logic for such systems, than to bother to write it all. In general, MO as applied solutions for existing systems is quite a working topic.
In contrast to TS on indices, MO allows to see the truth that there is no grails, that prices are noise and algotrading is a casino, I'm not even speaking about manual trading.
As for gurus, for the foreseeable past we had only one - SanSanych, and that for lack of a better one.
+100500
The MO, in contrast to the TS on the turkeys, allows you to face the truth that there are no grails, that prices are noise, and algotrading is a casino, not to mention the manual one.
Your MO doesn't show a damn thing. Everything, this was known without MO, and long before MO.
So I do, through GA, overkill is still primitive (in the article example, one "of")
the problem is not even how to "fix" it properly, but how to prolong the fun now... because usually it doesn't work out more than 100% of trail
I can't help it if I've got a gimmick, it's not a stationary one.And what are Gaussian mixes, I was too shy to ask, what can I do with them? I want to develop the probabilistic approach more, but I hesitate to start
Because the way of the samurai was suggested by Alexander, but to make it clear to no one at all, it is necessary to do it through probabilistic models of the MO.
Max, there is no other way than Doc's way. It is necessary to look at the results under different distributions of returns - normal, triangular, lognormal, etc. As soon as you see good results, you should select the same ones on the real BP. This is the only way and no other way, until the Wizard helps. It's a long, long way - but, it's worth it.
And the different Asaulenok - don't listen. I'm the only real physicist here. The rest are just... Mathematicians or worse. Good luck!
I only listen to people who really offer something interesting, Asaulenko only criticizes everything and tramples on the root :)) You have a lot of sensible ideas, I think ... in terms of probabilistic approach, of course.
Well, I may be exaggerating my merits, of course. :))
But, that's not the point. Please input to the NS different distributions of returnees. Put it in a table, as Doc did. Let's see.
Then put the best results wherever you want - in the forests, NS. Wherever you want. This will only give you +1-2% accuracy, not more.
I will do exactly that, but in a slightly different way
I'll show you the results soon, next year maybe))
OK. And the Hindu - give him tasks, do not be shy. Greetings!
OK. And give the Hindu a hard time - give him tasks, don't be shy. Happy holidays!!!
Happy New Year ))
Especially for you, one of Doc's last posts on my PM:
You already have an almost perfect time series (real ticks). By making a thinning you more or less keep all the hidden states, dependencies, and other things that are in that time series, and for the model in training this thinning is not a problem. But in doing so, you increase the average increase in price, and accordingly the spread becomes less and less of a problem.
In general, you can't spoil your oatmeal. Your tics are great. The erlang thinning helps to overcome the spread.
But it all rests on your tick source, the ticks have already been processed in a way that we don't know about, and that is the most important thing. If you take the ticks from the terminal, your whole strategy will probably not work. If I take terminal ticks, most probably, the entire strategy will not work. I've been trying to bring the prices available in MT5 to the stationarity of your real ticks for several weeks, but I've come very close.
Happy New Year ))
Happy New Year to you too).
You've done a lot of work this year.
But often chatter alone.
Without a pattern, you can't teach the woods, the needles, the trees, the taiga.
Look for the root first.