Building a trading system using digital low-pass filters - page 16
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
bstone, I'm not talking about the real thing, there's no hard evidence, because the statistics are simply unknown. You were talking about a Wiener.
Let's assume that one can't make money, just as one can't make money on a Wiener process. Then in the framework of your problem, you can limit yourself to generating a price series based on a Wiener process. As a result you will have synthetics for testing the characteristics of TS on price series, on which according to the theory there is no chance to make profit in the long run.
And until there is no strict proof that the price series, unlike a Wiener one, allows to make profit in the long run, there is no sense in solving your problem in its original formulation. What do you think?
P.S. As far as I understand, the forex statistics allow us to believe that the real price series will not make money in the long run :)
And why do I need such rows, bstone? I need realistic rows, not ones for which it has already been proven that no money can be made. I'll drive any system to its grave with a Wiener row like that...
Well forex will do the same to any system. So I do not see the difference. Do you really believe in an infinitely long working grail?
Well here we are, what about what I posted, proof that it's not a BGS, hence not a Wiener process, i.e. it's possible to make money. I have laboured, laboured or maybe I'm missing something :-(.
I guess you are hoping to find a stationary characteristic and use it as a key to build an algorithm?
The figure shows the variability of the returns (at five-minute intervals) throughout the day. There is no and should not be any stationarity here. It is clear that if the range of H-L changes, then returns should also change, due to statistical reasons.
The figure shows the variability of the returns (on the five minutes) over the course of the day. There is no and should not be any stationarity here. It's clear that if the range H-L changes, then the returns should change due to statistical reasons.
Sorry, maybe we do not understand something or mix up the terms. Let's look at an example. We have a row of numbers 1 2 3 4 5 we do two procedures. The first one adds a random number to the row to get a number -2 3 1 5 7. The other procedure (from each successive number we subtract the previous number) we get a number 1 1 1. So we get two series - one non-stationary, the other stationary.
So this phrase "It's clear that if the range H-L changes, the return should change due to statistical reasons" is incorrect. Yes the second series may be non-stationary, but not for that reason. Although I still doubt it is non-stationary.
Nothing gets lost anywhere. The H-L is essentially a spread of changes, a rough estimate not badly correlated with cattle. On retuns, if you take them as changes, you get the same thing. It is the same range of variation, but only close. On the chart we can see that H-L and returns differ from each other by a factor of 2, - it should be so in theory and so it turns out on the data. There are a lot of other small points that agree well with the theory.
Besides this, it cannot be so that if two series of data obtained from one, - one is stationary and another is not, why? Primitive operations were performed, what do they change in the data? The fact that 2 turned into 5 means nothing, the scale of change is the same.
Also, the figure, actual data - what can be incomprehensible, when you can see that the changes in the characteristics of the data reach 2-3 times. It's not 5% but 200-300%.