Spread trading in Meta Trader - page 150

 

К. В. Vorontsov, E. V. Egorova "Dynamically Adaptable Compositions of Forecasting Algorithms"

При прогнозировании зашумленных нестационарных временных рядов возникает проблема выбора адекватной модели временного ряда. Модели нестационарных процессов, такие как GARCH, несомненно, расширяют область применимости классических статистических моделей. Однако они опираются на априорные предположения о природе нестационарности (например, гипотезу о непостоянстве дисперсии) и потому являются в той же степени эвристическими, что и классические стационарные модели.

The idea of applying several heuristic prediction algorithms together[1] seems more universal. In this case the following hypothesis is accepted: a series can go from one state to another, and in each state its behaviour is well described by one of the standard models...

Interesting article in the context of application of different algorithms - Can We Learn to Beat the Best Stock.
 
goldtrader:
The implementation of state arbitration is described in books of 300-500 pages.
Unfortunately, I have not come across any literature describing the implementation of statistical arbitrage in the general case of market-neutral portfolio construction or optimal portfolio for a market-neutral strategy. The most that is described is a special case of statistical arbitrage: pair trading (spread trading).
 

Portfolio trading is not it. There is such a book, but it weighs 6MB, it cannot be uploaded here.

 
goldtrader:

Portfolio trading is not it.

What does it mean? I think I have written above what statistical arbitrage is in general terms.

There is such a book, but it weighs 6MB, so it cannot be uploaded here.

You can get bits and pieces of it.
 
hrenfx:
You can do it in chunks.
This is the first piece. Add ".001" to the file name. Packed with 7-zip.
 
hrenfx:
You can do it in chunks.
This is the second one. Add ".002" to the file name and unzip both.
 
goldtrader:
MarketModels by CarolAlexander.djvu

Yeah, thanks. Immediately read a piece like this:

Unfortunately, once again I encounter the pernicious influence of classical formulations of the portfolio formation problem.

The mathematical solution (on the second page) is absolutely correct. But what's the sense in the fact that the sum of weights equals one! This is nonsense!

If the weights are the shares occupied by each asset in the portfolio, then the sum of absolute values must equal unity!

One has the feeling that the authors of these theories do not know algebra beyond the second-grade level. And they adjust the condition (up to the point of absurdity) of the problem to a beautiful analytical solution.

I too have sinned almost the same:

Now about why the sum of squares of coefficients equals one. One, because it is the normalization of the vector. And squares because if JPYUSD is used instead of USDJPY, for example, it should not influence the estimate of interrelations. In this case only the sign of the appropriate coefficient will change. Ideally, the sum of absolute coefficient values, not the sum of squares, should be equal to one. I failed to find an acceptable solution for such a condition, so I settled on the sum-of-squares condition (and the solution is still simple). If you want to trade synthetic Recycle, you will need to normalize the optimal vector found with the sum-of-squares condition to the condition equal to one of the sum of absolute values. This would not be an ideal solution with the original sum-of-absolutes condition, but it would give a chance to evaluate the ideal solution.

But at least the sense is left not delusional. In general, books should be read very carefully.

Thanks again for the book!

 
goldtrader:
If possible, post more books on similar subjects.
 
goldtrader:
..... Your approach is very simplistic.

Maybe. But the simpler, the better ("strike while the iron is hot") is said for a reason.

Of course, one can delve into multi-volume abstruse talmuds and "sculpt" megabyte constructions with clever names, using sophisticated (and trendy) mathematical apparatuses.

But will they help in practice? I do not think so.

Meanwhile, the stated theme is: - "trading in metatrader". And not at all in the Russian government computer centre.

What I am proposing is a very concrete way to implement the stated theme "without leaving the box office"! Exactly in the metatrader.

I'm not arguing. Perhaps a "multi-smart approach" would be appropriate for employees of the largest specialist hedge funds and other "barclays". But to trade in metatrader (in my opinion) one should start with the simplest, - exaggerated ways and techniques of arbitrage trading.

All the more so, that most of the visitors of this branch are unlikely to be "docents with PhDs"(c).

 
hrenfx:
If possible, post more books on similar subjects.