How and where to set up a hedge fund? - page 11
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
Is there something to hang on to? For example, a moving average based on a linear regression is better than a simple average.
Yeah, there's a lot to go on. And there's a lot of maths to go around.
But there is always just one "BUT!"
The market has once again proved to be stronger
It's better to count money and trade - maths is essential.
Yes, there's a lot to get behind. And there's a lot of maths to come up with on it all.
But there is always only one "BUT!"
The market has once again turned out to be stronger.
But it is better to count money and trade - mathematics is of great importance here.
This is the third day I am writing the system of correlation, I started to trade manually some time ago and it was pretty successful.
In general terms: we calculate the Pearson correlation for all currency pairs, I selected 28 sane pairs in total. If I have found pairs with +-0.95 level, then I calculate Spearman correlation for these two pairs and decide what to sell and what to buy, sell a pair with a higher correlation and buy a pair with a lower one.
The result is still good, I checked it on history manually, in a couple of days I think I will convert the algorithm to a bot and run it for tests.
Here is the maths.
Screenshot at the moment:
In general, the market is a mathematical model, but it is difficult to pick a formula for profit.
This is the third day of writing a correlation system, I once started trading manually and quite successfully.
In general terms: we calculate Pearson correlation for all currency pairs, I selected 28 sane pairs in total. If I have found pairs with +-0.95 level, then I calculate Spearman correlation for these two pairs and decide what to sell and what to buy, sell a pair with a higher correlation and buy a pair with a lower one.
The result is still good, I checked it on history manually, in a couple of days I think I will convert the algorithm to a bot and run it for tests.
Here is the maths.
Screenshot at the moment:
In general, the market is a mathematical model, but it is difficult to find a formula for profit.
If it were easy - we would be eating rocks. Everyone would quit their jobs (and bakers would stop baking bread) and rush to the Forex market.
This will never happen, people are used to living in their "comfort zone" from which not everyone is able to get out.
The same hedge-fund, or PAMM - those who want to make money pour in there, and if they burn out, they habitually pour it in again, but elsewhere. If a man has never made money, he will not invest even $100. After all, many have been taught that money can only be earned with the hands, and have not been told that you can earn it with your head.
This is the third day of writing a correlation system, I once started trading manually and quite successfully.
In general terms: we calculate Pearson correlation for all currency pairs, I selected 28 sane pairs in total. If I have found pairs with +-0.95 level, then I calculate Spearman correlation for these two pairs and decide what to sell and what to buy, sell a pair with a higher correlation and buy a pair with a lower one.
The result is still good, I checked it on history manually, in a couple of days I think I will convert the algorithm to a bot and run it for tests.
Here is the maths.
Screenshot at the moment:
In general, the market is a mathematical model, but it is difficult to find a formula for profit.
Yeah, we used to do that.
If the correlation coefficient is large, the correlation is likely to decrease, i.e. pairs will diverge. Only it is not clear at the very beginning - which one goes where. I.e., the strategy in such interpretation will turn trades back and forth. This auto-impression of spread will be great for the quotes.
The solution was found at that time as follows:
- selection of pairs with high correlation coefficient, but it is traded from the moment when correlation of these pairs worsened. I.e. pair convergence is traded, not divergence.
Estimation of predicted profit is also possible.
We take a decent part of history, estimate the coefficient of correlation, and measure differences between pairs in pips. The statistical average is calculated. This figure will be the second confirmation of the signal along with the correlation coefficient.
You know, like linear regression, Fourier etc.
I tried Fourier - it doesn't work, it's all noise. Which is understandable, Fourier is mainly intended for analysis of periodic signals. Wavelets are better, I've been going back to them from time to time, but no practical results yet.
I made a couple of clips, like a movie, showing real time filter and wavelet in the second part of the video.
https://youtu.be/Jvk8eW6tNxY
https://youtu.be/75JmYzPQdyw
I tried Fourier - it doesn't work, just noise. Which is understandable, Fourier is mainly intended for analysis of periodic signals. Wavelets are better, I go back to them periodically, but there is no practical result yet.
I made a couple of clips, like a movie, showing real time filter and wavelet in the second part of the video.
https://youtu.be/Jvk8eW6tNxY
https://youtu.be/75JmYzPQdyw
Have you tried two timeframes, the small one is used to enter, the big one is used for filtering?
Have you tried two timeframes, the small one to enter and the big one to filter?
Of course, not the timeframes from the terminal, but the principle is the same
This is the third day of writing a correlation system, I once started trading manually and quite successfully.
In general terms: we calculate Pearson correlation for all currency pairs, I selected 28 sane pairs in total. If I have found pairs with +-0.95 level, then I calculate Spearman correlation for these two pairs and decide what to sell and what to buy, sell a pair with a higher correlation and buy a pair with a lower one.
The result is still good, I checked it on history manually, in a couple of days I think I will convert the algorithm to a bot and run it for tests.
Here is the maths.
Screenshot at the moment:
In general, the market is a mathematical model, but it is difficult to find a formula for profit.
Yes, the problem may occur when there is correlation and the pair decreases or vice versa. It means that there can be correlation in the falling market and the system will buy.
Yep, have been doing that sort of thing.
At high correlation coefficient there is a probability that the correlation will decrease, i.e. pairs will go in different directions. Only it is not clear at the very beginning - which one goes where. I.e., the strategy in such interpretation will turn trades back and forth. This auto-impression of spread will be great for the quotes.
The solution was found at the time as follows:
- Pairs with a high correlation coefficient are selected, but traded from the moment when the correlation of these pairs deteriorated. I.e. pair convergence is traded, not divergence.
Estimation of predicted profit is also possible.
We take a decent part of history, estimate the coefficient of correlation, and measure differences between pairs in pips. The statistical average is calculated. This figure will be the second confirmation of the signal along with the correlation coefficient.
The advantage of what is written now is that it works in the tester. As soon as I get a signal in the tester on the history, I open two charts and count the result. I came to a conclusion that at 20 points plus I should close both positions and wait for another signal. Sometimes it pulls at 200, but can overstay, and the transaction will hang.
The history has not been checked deeply but I think it is enough to judge about its efficiency.