Equity Curve R squared

autotradingalgos

autotradingalgos

For any traders out there with more than 500+ trades have you looked at the R Squared values for your equity curves? I combine various systems in the portfolio with Rsquare of between 0.95 and 0.982 and now have an equity curve R2 of 0.9872 since 2004 with 3922 trades. Anyone else done that analysis?

Regards to all and good trading.
 
My architecture can combine several strategies at the same time and output performance metrics across all strategies. I then can assess variability of returns for the whole "book".



autotradingalgos said:
For any traders out there with more than 500+ trades have you looked at the R Squared values for your equity curves? I combine various systems in the portfolio with Rsquare of between 0.95 and 0.982 and now have an equity curve R2 of 0.9872 since 2004 with 3922 trades. Anyone else done that analysis?

Regards to all and good trading.
More...
 
conduit said:
I find such measure very inferior to better measures to assess variability of returns. Risk adjusted returns to start with
More...

All the standard performance metrics are highly correlated. Heck, most of the time you're just trying to get the sign right!
 
autotradingalgos said:
For any traders out there with more than 500+ trades have you looked at the R Squared values for your equity curves? I combine various systems in the portfolio with Rsquare of between 0.95 and 0.982 and now have an equity curve R2 of 0.9872 since 2004 with 3922 trades. Anyone else done that analysis?

Regards to all and good trading.
More...

You shouldn't use R2 without being aware of its assumptions, eg linearity of (log) returns. If your system has a lot of skew then R2 won't be great; although in fairness the same applies to Sharpe Ratio and any other measure of risk that assumes symettry [same comments apply to higher moments].

GAT
 
Wait, now that is confusing. Leverage itself is a source of risk. When you increase leverage you certainly increase risk. So how can R^2 be a measure of risk when as you say it stays constant with higher risk?

Its more like a rhetorical question because I find R^2 a terrible measure of risk and do not even follow why anyone would use it for anything in regards to risk metrics. Sharpe ratio has its disadvantages (penalizes for upside return volatility) but is leagues better than r^2. Seriously what do you try to express with that?

globalarbtrader said:
R^2 will account for risk - it is invariant to leverage (double the leverage and R^2 is constant)

GAT
More...
 
conduit said:
Wait, now that is confusing. Leverage itself is a source of risk. When you increase leverage you certainly increase risk. So how can R^2 be a measure of risk when as you say it stays constant with higher risk?

Its more like a rhetorical question because I find R^2 a terrible measure of risk and do not even follow why anyone would use it for anything in regards to risk metrics. Sharpe ratio has its disadvantages (penalizes for upside return volatility) but is leagues better than r^2. Seriously what do you try to express with that?
More...

Sorry I didn't mean to confuse you. My point was that R^2 is a measure of risk adjusted returns. Sharpe Ratio is also a measure of risk adjusted returns. Neithier measure risk itself.

Let's suppose I have a strategy that returns 5% a year, 10% standard deviation of returns, R squared 0.9 (I have no idea if this is realistic as I don't use the measure myself).

Now if I double the leverage, I double the risk. However I will also double the returns. The Sharpe Ratio will remain the same. The R squared will also remain the same. Therefore both measures must be looking at risk adjusted returns.

Is that clearer?

By all means say that there are issues with using R squared, but it does look at risk adjusted returns.

GAT
 
You must log in or register to reply here.
Back
Top