How is the correlation of the meta strategies calculated? Probably based on the equity curve, but how exact and what interval (per trade, daily, weekly, etc.)?
Correlation coefficient: when strategies get close to -1, they cancel each other out, don't they? One System - hypothetically - increases by 1%, the other falls by 1%.
Is this really "excellent diversification"? Shouldn't one aim for a correlation that is close to 0?
Correlation coefficient: when strategies get close to -1, they cancel each other out, don't they? One System - hypothetically - increases by 1%, the other falls by 1%.
Is this really "excellent diversification"? Shouldn't one aim for a correlation that is close to 0?
Rename
QUOTE:Equity curve, yes. It's the result of the Corr (Correlation Coefficient, CC) indicator on the entire curve in the base scale. A CC is found for each combination of strategies, summed and averaged.
How is the correlation of the meta strategies calculated? Probably based on the equity curve, but how exact and what interval (per trade, daily, weekly, etc.)?
Example:
Equity curves of 4 Strategies: A, B, C, D. The correlation of A to the others is:
( Corr(A, B) + Corr(A, C) + Corr(A, D) ) / 3
Thank you, @Cone
And this please?
Correlation coefficient: when strategies get close to -1, they cancel each other out, don't they? One System - hypothetically - increases by 1%, the other falls by 1%.
Is this really "excellent diversification"? Shouldn't one aim for a correlation that is close to 0?
And this please?
Correlation coefficient: when strategies get close to -1, they cancel each other out, don't they? One System - hypothetically - increases by 1%, the other falls by 1%.
Is this really "excellent diversification"? Shouldn't one aim for a correlation that is close to 0?
QUOTE:
Corr ... indicator on the entire curve in the base scale
Is it a) the equity curve "as is" or b) the daily/weekly changes in the equity curve?
With a) two profitable strategies with low drawdown also have an r close to 1.0 even if the fluctuations are not correlated. (This is the ideal combination)....
a) "as is" (but synchronized) using Corr.ComputeCoeff()
A perfect negative correlation (-1) doesn't guarantee a neutral return since correlation is just directional. Example: If you buy equal dollar amounts of QQQ and QID (2x inverse of QQQ), the correlation is -1, but the return will positive if the market is trending higher.
A perfect negative correlation (-1) doesn't guarantee a neutral return since correlation is just directional. Example: If you buy equal dollar amounts of QQQ and QID (2x inverse of QQQ), the correlation is -1, but the return will positive if the market is trending higher.
Your Response
Post
Edit Post
Login is required