Sharpe and MAR

[AyePee]

Hi guys,

Does anyone have a feeling for what sort of Sharpe Ratio and MAR are attainable for a mechanical trading system?

Any thoughts on what may be better (more relevant) measures of performance for a mechanical trading system?

Cheers,

AP

 

[amnonco]

hi AP,
check out curtis faiths "way of the turtle" for a briefing about such performance ratios.
or you can google the ones I'll mention later...

comparing sharpe ratios is not a good way to evaluate all types of systems. it's good for evaluating systems that trade same/similar markets or that trade in the same/similar fashion.

also MAR ratio is very problematic because of 2 reasons:
1. CAGR% - can vary greatly depending on the test period, start backtesting half a year before or after, and the CAGR% can change significantly.
2 MaxDD - the same goes for the max drawdown. there's a difference between a syste with 23%, 21%, 18% and 19% drawdowns to a system with 25%, 6%, 4%, 5% drawdowns...

the solution is called R-cubed. which uses:
RAR% - which is the regression of annual return - a more robust measure of return.
avgDD - -which uses the average of 5 (or other number) max drawdowns..

google these...
Good Luck

 

[Adamus]

You need to keep an eye on several measures simulatneously.

I measure total equity, total # trades, $ per trade, correlation of equity curve to straight line from zero to final profit, geometric mean trade to name some of them.