|
PureBytes Links
Trading Reference Links
|
> ME> For a given strategy that a person wants to sell or share
> ME> what proof or documentation is necessary that the strategy
> ME> works?
>
> it is not the percentage profit that is makes a good system it's the
> profit factor. everything else is and exercise of if you can
> tolerate the other parameters or not.
PF is a pretty good indicator **IF** the system performs
consistently. But you can have a good PF with a lousy system.
Example: PF = total_wins / total_losses. Let's say we have a
system with 100 trades: 99 losses of $1000, and 1 win of $500k.
This system has a fantastic PF of 5.05. Would you want to trade
it?
Obviously not. And obviously that's a fairly bogus example,
since it's a really bizarrely-skewed result. But it shows how
PF, like most measures that look only at the overall final
numbers, can be misleading.
In my opinion there's no substitute for seeing the equity curve.
I want to see a system that performs well, month in and month
out, in all market conditions. I've seen plenty of systems with
great PF's that would have killed you in realtime trading -- huge
drawdowns followed by huge runups, etc. I don't want that much
stress. Give me a nice, steady equity growth any day. Provided
it's realistic, i.e. it includes realistic slippage and costs, I
think it's more important than PF, win%, almost anything else.
The best measure I know of to put a numeric value on "nice equity
curve" is the Sharpe ratio. Higher Sharpes generally (not
always) translate to smoother equity curves. If I had to ask for
one number to value a system, it would be Sharpe. HOWEVER be
aware that there are many ways to calculate Sharpe; it's not an
absolute figure like PF or win%. Even Dr. Sharpe himself says
there are a number of CORRECT ways to calculate a "Sharpe ratio."
So you'd have to know the Sharpe (as calculated by that person)
of a known system, so you could compare it to the Sharpe of the
system under question.
Gary
|