Machine learning in trading: theory, models, practice and algo-trading - page 2579
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
Any such row can be approximated to a stationary one
Like this one?
I've been looking for four years. It's faster than the Kalman.
Like this one?
I've been looking for four years. Faster than a Kalman.
The main essence of arbitrage, as of any post factum, is the observation that there is a balance in the market between the mutual movement of currency pairs
that is I would say an interrelated movement of them
and this is nothing other than the spread
in fact, the fight against the spread, which widens not in favor of the trader and can infinitely pump the dough out of his pocket while he has it ;)
this is a game of chess with a grandmaster, who rules the movement of kotir
it's a hell of a game.....
I advise not to crash into portfolios, or see my post about two triangles, which wobble relative to each other like pendulums.
An interesting and profitable strategy because they are both separately in balance and almost neutral to the trader.
The pairs are superimposed as follows:
1.0 - Bid/Bid0.
Equity on the example of EUR/USD/GBP
EURGBP + GBPUSD - EURUSD = 0.000 throughout history !!!
Good luck !
No, I meant non-stationary spread (which will be a frequent phenomenon). Stationarity is needed only when training the model while trading will be non-stationary in any implementation. The main thing is to have enough examples.
I thought you would understand. This is just that, for non-stationary, autocorrelated series, with transitions.
Here's a comparison of the error rate, of the two filters.
In green, Kalman.
I thought you would understand. This is just that, for non-stationary, autocorrelated series, with transitions.
Here's a comparison of the error rate, of the two filters.
The one in green is Kalman.
I thought you would understand. This is just that, for non-stationary, autocorrelated series, with transitions.
Here's a comparison of the error rate, of the two filters.
The green one is kalman.
So what is this?
We need to somehow divide the prices of the two pairs into
1) spread which earns
2) noise
Do PCA or other decomposition, maybe with AMO, autoencoders or networks (but it is more probable that everything will "die out" on the new data), so PCA is better
And we need a "number" to describe "a good spread" after the spread decomposition. Co-integration alone is not enough here, we need the spread to earn because it is not a linear product of prices but a non-linear part of prices after decomposition
the funny thing is that the spread of two pairs is the chart of the third one
What's the point of trading?
The funny thing is that the spread of two pairs is the chart of the third
What is the sense of trading?
The funny thing is that the spread of two pairs is the chart of the third
What's the point of trading that?
it makes sense to trade triangles. And the neuronet, if it is trusted, then train on three.
simply because there is no nishish in the history of 1 arbitrary symbol. It's effectively traded and in any theory is bound to be noise.
And in three you can catch the current situation with a small window.
Who has quotes for the pound and the euro more than 5 years 5min. throw me please!!!
SanSanych's hint was DucasKopi