Machine learning in trading: theory, models, practice and algo-trading - page 3220

[Maxim Dmitrievsky]

fxsaber #:

CSV: time bid ask.

How long do I need a series?

I took so many ticks D'2023.03.01', from MQ demo EURUSD, it turned out to be about 10mln

And the output is a series without timestamps, what can I do about it? I can link to the old ones, I can do without it.

[Maxim Dmitrievsky]

Even when generating Bid and Ask (their increments), the cumulative sums are very different, due to random sampling

It does not work together, i.e. you have to generate either one or the other.

[Maxim Dmitrievsky]

and I can also look for patterns in the ticks to try and explain what is being traded

[fxsaber]

Maxim Dmitrievsky #:

How long do you need the row to be?

The source is less than 7 million ticks. As a source.

I took so many ticks D'2023.03.01', with MQ demo EURUSD, I got about 10mn ticks

Why MQ-demo?

And the output is a series without timestamps, what can I do about it? I can link to the old ones, I can do without it.

You can't do it without timestamps. The patterns depend on the time of day. The same rollover, for example.

[fxsaber]

Maxim Dmitrievsky #:

Even when generating Bid and Ask (their increments), the cumulative sums are very different, due to random sampling

It does not work together, i.e. you have to generate either one or the other.

You can generate the average.

Forum on trading, automated trading systems and testing trading strategies.

Machine Learning in Trading: Theory, Models, Practice and Algorithm Trading

fxsaber, 2023.08.19 08:36 AM

The randomisation algorithm is like this:

1. A real tick history is taken.
2. A sequence of increments of the average ((bid+ask)/2) price is made from it.
3. In this sequence, each term is randomly multiplied either by +1 or -1.
4. A new tick history is collected from the obtained sequence of increments, where time and spread coincide with point 1.
5. The new tick history is written in a custom symbol.

Then timestamps and spreads will coincide.

[fxsaber]

Maxim Dmitrievsky look for patterns in the ticks to try to explain what's being traded.

That would be educational.

[fxsaber]

Maxim Dmitrievsky #:
I'll try tonight.

The only thing is, we've got to do something to avoid that.

Forum on trading, automated trading systems and testing trading strategies

Machine learning in trading: theory, models, practice and algo-trading

fxsaber, 2023.09.05 09:00

if you take a large ZigZag, the same flat scalper does not merge (even earns something there). But this is for the reason that the flat plots are bypassed by the randomiser.

That is, the generator can be so sumney that it will leave some chunks almost unchanged. At the same time, you can't see it at all with an unarmed (and even an armed one should try hard) eye.

[Maxim Dmitrievsky]

fxsaber #:

You can generate an average.

p.2 and p.4. Then timestamps and spreads will coincide.

https://disk.yandex.ru/d/_p7ZGNw70AiABQ

[Maxim Dmitrievsky]

fxsaber #:

The only thing is, we've got to do something to avoid it.

I.e. the generator can be so crazy that it will leave some pieces almost unchanged. At the same time, the naked (and even the armed should try hard) eye can't see it at all.

I think that here it will depend on the number of gaussians, this file is generated in 15 pieces.

plus sampling from the resulting distribution is random, i.e. it pulls each new sample randomly from the whole distribution.

[fxsaber]

Maxim Dmitrievsky #:

https://disk.yandex.ru/d/_p7ZGNw70AiABQ

Timestamps are not the same as the original. MQ-demo is not interesting.