Discussion of article "Forecasting market movements using the Bayesian classification and indicators based on Singular Spectrum Analysis"

 

New article Forecasting market movements using the Bayesian classification and indicators based on Singular Spectrum Analysis has been published:

The article considers the ideology and methodology of building a recommendatory system for time-efficient trading by combining the capabilities of forecasting with the singular spectrum analysis (SSA) and important machine learning method on the basis of Bayes' Theorem.

Since the price predictions presented in the Figure 1 contain a set of errors in direction, let us find out how a combination of the indicators allows reducing the number of such errors. The difference in the values of the actual and predicted price in Figure 1 is irrelevant, because it does not cause losses when in a concordant position. 

There are values: (MACD-Signal), difference derivatives for series of the actual and predicted price and stochastic, smoothed by three points. The consistency of the conditions defined earlier, which indicate the direction of the price change, is further used to analyze the efficiency of their combination. 


Figure 5. Concurrency of the actual and predicted directions of the changes in price when using three indicators

Author: Roman Korotchenko

 

I don't believe that SSA will give a correct forecast in the presence of external news, which knock quotes out of the found model. So I would put up a red warning that you should look at the calendar in parallel.

And the idea of the article on probabilities is stolen from me ;-). But I wanted to do everything with open sources and without linking to specific indicators (there is even a universal Expert Advisor, where you can add different indices and calculate probabilities of events). But here - more about SSA.

 
Stanislav Korotky:

The idea of the article on probabilities was stolen from me ;-). But I wanted to do everything with open sources and without linking to specific indicators (there is even a universal Expert Advisor, where you can add different indices and calculate probabilities of events).

Can you give me a link to read it?
 
Stanislav Korotky:

I don't believe that SSA will give a correct forecast in the presence of external news, which knock quotes out of the found model. So I would put up a red warning that you should look at the calendar in parallel.

And the idea of the article on probabilities is stolen from me ;-). But I wanted to do everything with open sources and without linking to specific indicators (there is even a universal Expert Advisor, where you can add different indices and calculate probabilities of events). But here it is more about SSA.


Then we are waiting for your article :) In general, the caterpillar has not been able to achieve anything from it, no matter how much we chased it with the dude.
 
Ilya Flegentov:
Can you give me a link to read....
There's no link yet. I have an idea and a not yet finalised expert. I wanted to get the probability formula in analytical form and compare the results with head-on calculation on historical data.
 
Stanislav Korotky:
There's no link yet. I have an idea and a not yet finalised expert. I wanted to get the probability formula in an analytical form and compare the results with a head-on calculation on historical data.
Is there enough material for an article? It would be good
 
Rashid Umarov:
Is there enough material for an article? That'd be good.
It is not ready yet. There are problems with the formula, and the expert under MT4 (so historically), but this, as far as I understand, is not welcomed now, so it will be necessary to translate it under MT5.
 

There is nothing to discuss yet, unfortunately.

Maybe we should have started publishing from the third part? Side question, couldn't you have put everything in one?

Many people have experimented with this topic (SSA). It would be interesting to compare the result.

Good luck

 
Vladimir Perervenko:

There is nothing to discuss yet, unfortunately.

Maybe we should have started publishing from the third part? Side question, couldn't you have put everything in one?

Many people have experimented with this topic (SSA). It would be interesting to compare the result.

Good luck


Given the rather high level of the article, I want to raise doubts with the author, namely the need to write an expert at this stage. There is no evidence in the article that the results of testing an expert will be trusted. Well, if we get 2 or 3, even 10 values of the profit factor, or any other values from the tester - is this a statistic? What is the guarantee that the expert will behave in the same way in the FUTURE?

At the heart of these doubts is the author's assertion that SSA is capable of working in NOT stationary markets. Where is the proof? I don't recall any such proof.

Suppose I have missed something in this matter. But the article does not specify at all what varieties of non-stationarity SSA solves and what the result is. Is it possible to isolate the trend? But then the residual from subtracting the trend is not necessarily stationary. This issue is addressed in great detail in the framework of ARCH models. Because of the variety of residuals, a very large variety of ARCH models has arisen.

This piece is not in the article, and therefore there is no evidence that trading decisions are made on a stationary time series. Hence it follows that the future behaviour of the TS on these ideas is NOT predictable.


PS.

About 10 years ago I used "caterpillar" (FATL-SATL). Expert Advisors lived from 3 to 6 months, then started to drain. The main problem is not only NOT stationarity in the classical sense (MO and dispersion change), but also in changing periodicity, which can be clearly seen in ZZ.

 

СанСаныч Фоменко:

PS.

About 10 years ago I used "caterpillar" (FATL-SATL). Expert Advisors lived from 3 to 6 months, then started to drain. The main problem is not only the non-stationarity in the classical sense (MO and variance change), but also the changing periodicity, which can be clearly seen in the ZZ.

Re-optimisation or spreading the EA over different instruments did not help?
 
Rashid Umarov:
Re-optimisation or spreading the EA over different instruments didn't help?

Started to drain - re-optimisation - started to drain.....

The worst thing is that the"lifetime" is not known. I tried to optimise every weekend - doesn't help. On an intuitive level I stick to the point of view that after about 3-6 months the quote changes in some way, which leads to the drain of all previously successfully traded Expert Advisors. You can see it very well from the signals. It is pointless to take a 3-month signal.

It is necessary to do datamining. It is the basis of stability of future trading.