Re: Criteria for determing a "good" trading system Part 2

[Bob Fulks <bfulks@xxxxxxxxxxxx>]

At 1:16 PM -0800 3/27/02, carrslem wrote:

>I'm amazed at the number of performance numbers which mathematicians
>can devise! As for me, inspection of a chart of the equity curve is
>the determining factor. I want that "curve" to approximate a straight
>line as closely as possible and to do so consistently and
>independently of the changing trends in the market. This property is
>much more important than the magnitude of the end result (provided,
>of course, that the line DOES have an upward slope). If such a system
>shows a mediocre rise, I can boost it simply by increasing the trade
>size, something I would NOT do if the curve were erratic with
>occasional protracted loss periods.

I would agree.

But a related single number is the Sharpe Ratio (properly calculated
   - unlike that in TS2000i):

   Linear equity curve vs. time = constant return vs. time.

   Constant return vs. time = very low standard deviation of returns vs. time

   Sharpe ratio = average return / standard deviation of returns

so linear equity curve = consistent performance vs. time
                       = low drawdown
                       = high Sharpe Ratio.

Objective measurements:

    Sharpe Ratio            Comment
        0.3          Buy/hold of the Nasdaq
        1.0          Only the best mutual funds
        1.5          Minimum for a "decent" trading system
        2.0          Pretty good trading system
        3.0          Really good trading system
        5.0 up       Exceptional trading system
             

Bob Fulks