Discussion of article "R-squared as an estimation of quality of the strategy balance curve" - page 6
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
New article R-squared as an estimation of quality of the strategy balance curve has been published:
Author: Vasiliy Sokolov
The correlation coefficient of the equity curve is an awesome metric, I've used in another platform but I did not have the know how to develop in mql5.
I was looking for it for years!Great writing Vasiliy Sokolov, Thank you soo much!
Vinicius from Brazil
An article on how to choose an even eqity.
Vasily,
Thank you for the great article.
Is it possible to add accounting for the slope of the balance curve and make a combined optimisation parameter ?
Thank you for your reply.
Basil,
Thank you for a great article.
Is there any possibility to add the balance curve slope and make a combined optimisation parameter ?
Thank you for your reply.
Good afternoon, the angle is a tricky one. The point is that the Y axis is balance or equity and the X axis is time or number of transactions. There can be no angle between time and balance axes, because they are different characteristics. It is like trying to find the angle of dependence between the volume of a thing and its weight. The relationship is obvious, but there are no angles, just as there are no angles between power and speed, age and coefficient of survival, etc.
However, in terms of meaning, the most appropriate parameter here is the yield expressed in per cent per annum calculated exactly to the regression line. Obviously, the higher the per cent per annum, the greater the visual angle.
Good day, it is difficult with the angle. The point is that the Y-axis is balance or equity, and the X-axis is time or number of transactions. There can be no angle between the axes of time and balance, because they are different characteristics. It is like trying to find the angle of dependence between the volume of a thing and its weight. The relationship is obvious, but there are no angles, just as there are no angles between power and speed, age and coefficient of survival, etc.
However, in terms of meaning, the most appropriate parameter here is the yield expressed in per cent per annum calculated exactly to the regression line. Obviously the higher the annual percentage, the greater the visual angle.
Vasily, thank you for your reply.
In fact, here we don't need to recalculate the slope angle with each update of R-squared and keep its change in memory. We are only interested in the final angle and comparing these final angles between several runs on the tester.
As an option :
1. When EA starts, read from the Tester End Date log and on that date geometrically calculate the angle just before stopping;
2. The incomparability of time and balance (or equity and number of trades) can be solved by simplifying: (1) specify that the initial balance in the tester should always be 10 000 USD, (2) the duration of the run should be balanced by a correction factor in the input, e.g. for day-month run time the coefficient is 0.1, for 1-3 months the coefficient is 0.3, for 3-6 months the coefficient is 0.5, for 6-12 months the coefficient is 1, for 1-3 years the coefficient is 3, etc. In the end, we are still comparing all runs over the same time period. The difficulty is how much weight to give to the individual elements of the Rsquare * AngleCoeff equation.
Of course simplistically I can do it in Excel :))))
Maybe from the point of view of automatic use in code, the annual percentage is really simpler and more reliable.
Thanks again for the article!
Vasily, thank you for your reply.
Actually, here too we don't need to recalculate the slope angle with each R-square update and keep its change in memory. We are only interested in the final angle and comparing these final angles between several runs on the tester.
As an option :
1. When starting EA read from the Tester End Date log and on this date geometrically calculate the angle just before stopping;
2. The incomparability of time and balance (or equity and number of trades) can be solved by simplifying: (1) specify that the initial balance in the tester should always be 10 000 USD, (2) the duration of the run should be balanced by a correction factor in the input, e.g. for day-month run time the coefficient is 0.1, for 1-3 months the coefficient is 0.3, for 3-6 months the coefficient is 0.5, for 6-12 months the coefficient is 1, for 1-3 years the coefficient is 3, etc. In the end, we are still comparing all runs over the same time period. The difficulty is how much weight to give to the individual elements of the Rsquare * AngleCoeff equation.
Of course simplistically I can do it in Excel :))))
Maybe from the point of view of automatic use in code, the annual percentage is really simpler and more reliable.
Thanks again for the article!
You just need to set the chart size in pixels, then all charts will have the same size and you can just work with them as a picture, measuring the slope of the angle. But wouldn't it be better to just look at profitability in that case?
You just need to set the chart size in pixels, then all charts will have the same size and you can just work with them as a picture, measuring the slope of the angle. But wouldn't it be better to just look at profitability in that case?
I like the pixel idea :)))
Meaning use profitability as a factor and get a combined optimisation parameter - maybe even better than angle slope. Thanks for the idea.
You just need to set the chart size in pixels, then all charts will have the same size and you can just work with them as a picture, measuring the angle slope.
Bullshit.
I like the idea of pixels :)))
Meaning use profitability as a coefficient and get a combined optimisation parameter - maybe even better than slope angle. Thanks for the idea.
You can also measure profitability on different intervals, both on the chart and on time intervals, then average the value and compare it with normal profitability. You can compare absolute values, or you can calculate the average deviation, or the percentage of time above/below the average.
Bullshit.
Justify it.
Justify it.
Okay. Just first calculate the angle of inclination formed by the following two points: