Machine learning in trading: theory, models, practice and algo-trading - page 1593
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
Can't you apply any flat strategy to this chart?
Well if it's (log)returns then I can't, it's obvious that it's not about the price)))
What makes you think that white noise is unpredictable?
By definition, it is designed that way. Of course theoretically it is pseudorandom, but the dependence is so complex and non-smooth that its presence can be neglected.
If the series is stationary, there is no need to use increments.
This is clear, but why bring it up if in reality it is not so? In fact, the requirement of statistical stationarity is very strict, quite minimal piecewise constant autocorrelations are enough to build a grail.
The sticking point is that financial series are DONE by INTENTIONALLY SUSPECTED PEOPLE, it's a game, a hunt. Even if the market were not influenced at all by fundamentals, it would still be constantly changing, as participants PLAY, trying to predict the behavior of others on average, thereby compensating for the time patterns that occur in the system. It makes no sense to explain such systems in terms of old statistics, it's like trying to predict the trajectory of a soccer ball using pure statistics.
The catch is that financial series are DONE by INTELLIGENT PARTIES, it's a game, a hunt. Even if the market were not influenced at all by fundamentals, it would still be constantly changing, because participants ARE playing, trying to predict the behavior of others on average, thereby compensating for temporal patterns that arise in the system. It is meaningless to explain such systems in terms of old statistics, it is like trying to predict the trajectory of a soccer ball using pure statistics.
The price series has"minimal piecewise constant autocorrelation.
All social statistics are based on an analysis of series generated by interested intelligent beings."
Well if it (log)returns then I can not, it is obvious that it is not about the price)))
By definition, it's designed that way. Of course theoretically it is pseudo-random, but the dependence is so complex and not smooth that its presence can be neglected.
This is understandable, but why bring it up if in reality it is not so? In fact, the requirement of statistical stationarity is very strict, quite minimal piecewise constant autocorrelations are enough to build a grail.
The sticking point is that financial series are DONE by INTENTIONALLY SUSPECTED PEOPLE, it's a game, a hunt. Even if the market were not influenced at all by fundamentals, it would still be constantly changing, as participants PLAY, trying to predict the behavior of others on average, thereby compensating for the time patterns that occur in the system. It makes no sense to explain such systems in terms of old statistics, it's like trying to predict the trajectory of a soccer ball using pure statistics.
Ok, it makes no sense using statistical terms. What terms should be used to describe market quotes?
The price series has"minimal piecewise constant autocorrelation.
All social statistics are based on an analysis of series generated by interested intelligent beings.
Eh if only...)) More precisely eh if they could be easily traded:))
But they are so ephemeral and sometimes deceptive, again for the reason that they are created by people who compete for one tidbit, alas most people lose, and the profits are taken by a select few, just like in sports, many people do exercises and go to the gym, but very few people earn half a billion dollars in sports.
Okay, in terms of statistics it makes no sense - what terms would make sense to describe market quotes?
I didn't say "describe" I said explain.
Eh if only...)) More precisely eh if they could be easily traded:)
Of course there is autocorrelation and other patterns, but they are so ephemeral and sometimes deceptive, again for the reason that they create people who compete for one tidbit, that alas most people lose, and profits take a selected few, just like in sports, doing exercises and going to the gym a lot, but half a billion dollars earn on the sort of units.
What to use instead of statistics?
What to use instead of statistics?
Statistics are our everything. Machine learning is an extension of statistics. Describing and analyzing data differently is the best way to do it. But you have to try to understand the reasons that drive the price, try to extract "reasons" from economic data, political data, social data, BIG DATA, as they are now called, everything that can affect the price. Yes, there is some alpha in the price itself too, but it is all eaten up by market costs, you need a lot more data and custom attributes based on an understanding of the market.
Statistics is our everything. Machine learning is an extension of statistics. No one has figured out a better way to describe data. But you have to try to understand the reasons that drive the price, try to extract the "reasons" from economic data, political data, social data, BIG DATA, as they are now called, everything that can affect the price. Yes, there is some alpha in the price itself too, but it's all eaten up by market costs, you need a lot more data and custom attributes based on market understanding.
Where is the source of "big data"?
Is there a base?
Reindeer, get out of here
Natasha, you are being bad again!
Start your own thread.