Not the Grail, just a regular one - Bablokos!!! - page 86
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
I have one question, why "IQUITY"???
I guess that's what it sounds like in Albanian :)
I had another idea. Not really on the subject, but for some reason I stumbled upon it several times when I was thinking about equity and TA. We take a primitive NOT system (Close[OPT1]>Close[OPT2] and close in OPT3 days), optimize it on the interval and do the same but Close[OPT1]<Close[OPT2]. We find an option where we have two profitable equity with the same options. It can happen, for example, if there is a trend. We monitor equity on the less profitable option and trades on the more profitable one. The idea is that a good trend is good in any combination as "while the fat one dries, the thin one dies"
.
Look for the equity variant (select the options) where there is a pattern on the equity and there is a signal on it - trade the received options.
Then we repeat the process.
Collect from ticks candles by their number, not by time. Therefore, trends are stretched and flat patterns are compressed, which improves further analysis.
And I collect them not from left to right, but from right to left, so that at the moment of analysis the last tick is completely formed, where the last tick is the clause of the candle.
The system is always in the market - only flips.
It is difficult to say what is the timeframe. One candle is 400 ticks.
Market analysis every 800 ticks.
I.e. the machine drew different models and analysed their statistical advantage...
It turned out, trivial and simple...
The model gave more statistical advantage, the less frequent it was...
In each analysis there were 10,000 models found to estimate the statistical advantage...
I.e. if a model was less frequent, more history was needed for the set of at least 10.000 deals...
The dependence turned out to be exponential...
That is, with a linear increase of the statistical advantage of the model, the number of trades fell exponentially on the same amount of price sampling...
This fitted very well with common sense...
As they say once a year and the stick shoots...
I.e. theoretically a 100% model will work once in an infinite sample...
I.e. is essentially an impossible event...
And a frequent here-and-now model will work out 50/50...
But I will say that there is one idea to get around this problem...
That is, to have a statistical advantage for the rare models of the here and now...
Where do you get your estimate of the coefficients, maybe they have an unmeasured variance, or they don't converge to their mo?
RLS is an adaptive filter. The mathematical expectation of the coefficients is not constant, as the interaction forces between the price series are constantly changing.
NEO
this is partly true, provided that they traditionally regard SB as a kind of integral continuous process, for which the creation of a practical model at all is highly unlikely. However, if we speak about some discrete SB and not about the process as a whole, but only about its some but specific fragments or, if you like, sections, then for some, comparatively short in time discrete sections, the construction of a model seems possible and the practical value of such models is proved by real trading.
A selection of interesting thoughts.
The problems seem to be 2 or more, not just one.
1- Gain a stat advantage, which of course will be negligible in small cases, compared to a very long sample, but that's half the problem.
2- Make it so as to increase the number of such situations, i.e. bypass the rarity problem.
The second problem was hinted at more than a dozen pages ago.
Good afternoon gentlemen. Please look at the above, I think there is no need to comment. These are the results of my audit. I am not the author of this thread and I have nothing to do with its authorship.
To the moderators or whoever chopped up the author's accounts - please stop.
Alexander ( or whatever his name is...) - I have no words, respect!
Good afternoon gentlemen. Please look at the above, I think there is no need to comment. These are the results of my audit. I am not the author of this thread and I have nothing to do with its authorship.
To the moderators or whoever chopped up the author's accounts - please stop.
Alexander ( or whatever his name is...) - I have no words, respect!
The filter according to the author's system was applied to my automatics, which I rarely use. As a result about 80% of signals went to the trash and I got a sampling of 18 deals, only 1 of which turned out to be unprofitable. I did not use any optimizations or anything else.