From theory to practice - page 197
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
And one more, for general fun, so to speak:
If you strain a bit, by your sophisticated WMA is it as I understand it?, the trail goes to a regular MA.
If you strain a bit, by your sophisticated WMA is it as I understand it?, the trail goes to the regular MA.
This is me comparing the built-in average LWMA and MA with mine from Wissim
Now I spend a lot of my time dealing with non-stationary flow of tick quotes (tick arrival rate is absolutely non-stationary process). I am literally bending this flow with an exponent in order to convert it into a stationary Poisson flow, but it is hard to resist. It can take a lot of time and I feel sorry for it.
Does this conversion make any sense? Should we abandon these endeavours? After all, non-stationarity is already taken into account in my formulas (if you don't know how to do it - I can tell you in private) ?
Does this conversion make any sense? Maybe we should leave it alone? After all, non-stationarity is already taken into account in my formulas (if you don't know how to do it, I can tell you in private)?
Imho, I suppose that it does not. From my vantage point, it doesn't interfere with anything at all.
I have been dealing with non-stationary processes (signal processing) all my conscious life, and I do not feel any discomfort from it.
Imho, I suppose it doesn't. From my vantage point, it doesn't interfere with anything at all.
I've been dealing with non-stationary processes (signal processing) all my conscious life, and I don't feel any discomfort from it.
Uh-huh. Thank you. It's kind of hard and tedious.
Uh-huh. Thank you. It's kind of hard and tedious.
In general, unsteadiness is a blessing, not a misfortune. There is no need to fight it). But the question is very philosophical... I think you will get to the bottom of why.
In general, unsteadiness is a blessing, not a misfortune. There's no need to fight it). But the question is very philosophical... I think you will get to the bottom of why.
In general, stationarity is easier to predict. That's why in the next thread nothing works and never will work with respect to forecasting.
Non-stationarity does not bother me at all at the moment. But! I wanted to kill two birds with one stone - to make some money on Wiener's process with demolition, and to start making neural networks already.
But, apparently, this is not the case... The transition to stationarity is extremely difficult - the first shoots are observed only in the sliding window = 16 hours! This significantly reduces the number of trades, and the result of the neural network is still only in the fog...
I think I have to stop my theoretical research in time.
Generally speaking, it is easier to make predictions on stationarity. That's why in the next thread nothing will work and never will work in terms of forecasting.
I don't want to start this conversation about non-stationarity, because it will turn out to be too long, which I'm frankly not ready for).
Non-stationarity (and long tails as well) may indicate (not indicate, but may indicate)) that the process is pseudo-random, i.e. the process carries some kind of information load. By removing non-stationarity we release this load at the same time.
Another question is that we can use information only a posteriori, after it is received and interpreted. That is, after everything has already happened).
I don't want to start this conversation about non-stationarity, because it would be too long, which I am frankly not prepared to do).
Non-stationarity (and long tails as well) may indicate (not indicate, but may indicate)) that the process is pseudo-random, i.e. the process carries some kind of information load. By removing non-stationarity we release this load at the same time.
Another question is that we can use information only a posteriori, after it is received and interpreted. That is, after everything has already happened).
Whatever. I don't want to either - I've read enough here about the possibility/impossibility of making money from Wiener processes. Gentlemen, in the heat of the argument, though, forgot that you can only approximate a process to make a Wiener process, I suspect that only starting from window = 24 hours, and "memory" still can't be completely eradicated. So it will still be possible to make money "on the tails". But! The possibility of precisely forecasting appears!
All right. I got carried away. I repent.