Thoughts on some of the absurdity of multi-currency analysis. - page 15

New comment
 
Yes.
 

Archiving:

As you know, there are two types of data compression: lossy(JPEG, MPEG-4) and lossless(RLE, LZW).

Hypothesis:

The market produces information at the same rate.

Every hour the market produces N bytes of pure information. This means that if all the information produced by the market in an hour is perfectly compressed, it will take N-bytes.

There is a compression algorithm that compresses market information at an almost constant bitrate.

Usage:

Apply compression algorithms to market data in order to compress the information with a constant data flow (bitrate).

Example:

The information produced by the market in the last 50 minutes is processed. It already occupies (in compressed form) M bytes(M < N). Given that M should get closer to N in an hour and knowing the compression algorithm we can look through all possible versions of such information for the remaining 10 minutes which would give us these final N bytes in a compressed form. And accurately discard those variants that wouldn't give us that much. I.e. a rough prediction, and the closer to the end of the hour, the more accurate the prediction. At the 59th minute the prediction a minute ahead would be highly probable.

Algorithm choice:

Most likely we must stop at a lossy compression method, close to streaming video (audio) compression algorithms.

Baseline conversion:

But still if we stop at algorithms without loss of information. Before compressing, most often there are transformations(BWT, PPM) of data for more effective compression. This transformation is basis transformation. The most convenient basis is the one which is the most compressible.

P.S.

MT5 developers have announced that they have managed to create a highly specialized algorithm of compression of quote history that is superior to all universal archivers. Its secret is an effective data conversion before compression. This is not a conversion of basis majors, but transformation of single trading instruments. If this transformation is disclosed by the developers, it can be used (without reinventing your own wheel) as a starting point to the goal of baseline transformation.

 

Beautiful, getch.

 

getch писал(а) >>

The most convenient baseline is the one that lends itself best to compression.

Or predictable. It's buttery. :)

 
MetaDriver >>:

Или прогнозированию. Масло маслянное. :)

Predictability is a problem to check. Compressibility is not.

 

getch писал(а) >>

The developers of MT5 said that they managed to create a narrowly focused algorithm for compressing quote history, which is superior to all universal archivers in terms of compression. Its secret is the efficient transformation of data before compression. This is not a conversion of basis majors, but transformation of single trading instruments. If this transformation is uncovered, it can be used (without reinventing your own wheel) as a starting point to the goal of baseline transformation.

One could easily increase the infocapacity of the quote stream many times over - switch to delta coding. Which is equivalent to returns-conversion. :) :)
 
MetaDriver >>:
Можно легко в разы увеличить инфо-ёмкость потока котировок - перейти на дельта-кодирование. Что эквивалентно ретурнс-преобразованию. :) :)

Delta encoding is one of the most primitive and ancient conversions. Universal archivers use excellent data conversion efficiency. But the developers of MT5 claim to have managed to outperform them considerably. Perhaps this is just marketing.

 
getch >>:

Прогнозируемость проверить - проблема. Сжимаемость - нет.

No problem either - a tester.

 
MetaDriver >>:

Тоже не проблема - тестер.

How can a tester help in determining the level of predictability?

 
getch >>:

Дельта-кодирование - одно из самых примитивных и древних преобразований. Универсальные архиваторы используют отличные по эффективности преобразования данных. Но разработчики MT5 утверждают, что им удалось их существенно переплюнуть. Возможно, это только маркетинг.

Universal archivers make little use of data specifics. If you take it into account (know in advance) you can put a lot of things out of brackets.

1...8910111213141516171819202122...40
New comment
Reason: