Regression equation - page 15
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
What does it mean to say that the cointegration problem has been solved? The purpose of using cointegration is to get a stationary BP. If successful, you can start mowing the lawn.
Do I understand correctly that mowing the lawn at stationary allows constant autocovariance (or rather its dependence on lag only) and unchanged MO? To be honest, I don't quite understand how constant autocovariance affects the amount of grass.
I've got it that autocovariance staggers, but it staggers quite predictably...
In general, you don't need stationarity in the strict sense to make money. Some persistent dependence in the autocovariance (or more strictly - autocorrelation) of BP is enough. For example, the periodicity of local extrema. Perhaps so dependent VR can be easily transformed to stationary one and that is why all in the end speak about the necessity of solving the cointegration problem for earning.
But still I don't see how constant autocovariance can help.
In my research I understood why the FOREX market is much more complicated than others. The fact that in FOREX there are very few financial instruments: a dozen or two. And the correlations between them are extremely unstable. In other markets, there are many more financial instruments and portfolios whose equity is almost stationary can easily be found.
Do I understand correctly that mowing the lawn at stationary allows constant autocovariance (or rather its dependence on lag only) and unchanging MO? To be honest, I don't quite understand how constant autocovariance affects the amount of grass.
I've got it that autocovariance staggers, but it staggers quite predictably...
In general, you don't need stationarity in the strict sense to make money. Some persistent dependence in the autocovariance (or more strictly - autocorrelation) of BP is enough. For example, the periodicity of local extrema. Perhaps so dependent VR can be easily transformed to stationary one and that is why all in the end speak about the necessity of solving the cointegration problem for earning.
But still I don't see how constant autocovariance can help.
In my research I understood why the FOREX market is much more complicated than others. The fact that in FOREX there are very few financial instruments: a dozen or two. And the correlations between them are extremely unstable. In other markets, there are many more financial instruments and portfolios whose equity is almost stationary can easily be found.
I'm missing the point again. how do you get to the stationarity of the non-stationary... does regression allow this or is it just a pen sketch?
Mowing the lawn allows you to be sure that the parameters are constant. At its simplest, a stationary series is mean-reverting, i.e. if it has moved away from the mean, it is bound to return to it. Don't care about autocorrelation, don't care about extremes. The arithmetic mean is the road to riches, trade from the channel walls to the centre.
Constant autocorrelation is simply part of the definition of stationarity. To be happy, it is enough for the average to be constant. Let the autocorrelation walk. If you want, you can try the GARCH model that will help you to detect clusters of higher/lower volatility and determine the exact borders to trade from, because they can be different at different times. Alternatively, you may choose not to bother and just trade from two SCOs. Very soon there will be nothing to dispose of. If the process is really stationary.
Forex is much easier than other markets. They already have ready-made spreads, natural cointegration analysis. There is no need to try to cointegrate them again. They are ready long-short portfolios - buy euro - sell dollar. All currency pairs are already stationary and mean-reverting. You just need to look not at minutes or hours, but weeks or even months. Even on the hour any pair is random wandering of pure water. The 1:100 leverage blinds people, it's like looking at an elephant through a microscope.
Forex is a very slow and very low volatility market. The right way to trade there would be one or two trades a year. Looking at stocks, a price change of 3-5-7 percent or more in one day? Easy! Sometimes even several times a day. That's where the real action is. Forex, as most people try to trade it, is digging through the noise, looking for a needle in a haystack. One should go to those markets where there is a real opportunity to make money and not play roulette on forex.
Mowing the lawn allows you to be sure that the parameters are constant. At its simplest, a stationary series is mean-reverting, i.e. if it has moved away from the mean, it is bound to return to it. Don't care about autocorrelation, don't care about extremes. The arithmetic mean is the road to riches, trade from the walls of the channel to the centre.
Constant autocorrelation is simply part of the definition of stationarity. To be happy, it is enough for the average to be constant. Let the autocorrelation walk. If you want, you can try the GARCH model that will help you to detect clusters of higher/lower volatility and determine the exact borders to trade from, because they can be different at different times. Alternatively, you may choose not to bother and just trade from two SCOs. Very soon there will be nothing to dispose of. If the process is really stationary.
Might be an off-topic, as I haven't reread the thread.
I'm honestly not much surprised that this topic has taken so long to discuss. I think it's simple enough.
I've attached a parameter selection module to my indicator and I've drafted the code in about fifteen-twenty minutes.
It was necessary to confirm the words that multivariate linear regression is unequal to the weighting coefficients.
For example:
The weight coefficients obtained in both cases are not proportional: {k1; k2; k3} !~ {n1; n2; n3}.
This sometimes not immediately obvious fact is best shown by an example:
timbo:
You just need to look not at one-minute or one-hour, but at one-week or even at one-month. Even on the clock any pair is a random wandering of pure water.
Forex is a very slow and very low volatility market. The right way to trade there would be one or two trades a year. Looking at stocks, a price change of 3-5-7 percent or more in one day? Easy! Sometimes even several times a day. That's where the real action is. Forex, as most people try to trade it, is digging through the noise, looking for a needle in a haystack. One should go to those markets where there is a real opportunity to make money and not to play roulette on Forex.
I wonder how you can explain such a picture on a time frame from about 20:00 to 23:00 -
I think you can trade on tick charts, or on one-minute charts, or on 5-minute charts, or on any timeframe - everywhere and everywhere
There will be channels, resistance and support lines.
It took you to confirm the words that multivariate linear regression is unequal to the weighting coefficients.
How long did it take you to make this discovery? :)
Look up Pearson's 1901 work on orthogonal regression.
How long did it take you to make this discovery? :)