|
PureBytes Links
Trading Reference Links
|
--- Ernie Bonugli <ebonugli@xxxxxxxx> wrote:
> The main point I was trying to answer was whether there is more to a
> successful system than just the Sharpe Ratio. I posed the question by
> presenting the result of a system which I developed which shows a
> relatively "good" Sharpe number and yet we can see that the system
> goes flat at certain times in its history. I do not use this system.
One interesting aspect of sharp ratio on flat period.
When a trading system is flat on its equity curve,
but its sharpe ratio does not drop, most of the time
it implies a strong money management scheme is built
into the system which let it "workout" its bad times with
very protective measures as most trading signals do have
their times that it works less than desirable comparing
to its normal behaviour.
>
> My personal conclusion is that while Sharpe gives us a very good
> indicator for the "goodness" of the system, it is not all telling. Or
> if it is all telling then a system with a ratio of 4 or less is not
> all that good. If the Sharpe ratio is the best measure of "goodness" then
> the system only becomes worthy of being put into production is it exhibits
> well over a value of 4. My current best effort is ~4.5. So a Sharpe of 5
> or better would allow me to sleep at nite knowing my funds are well invested.
not really, a sharpe of 5 is simply than every 5 years,
you can get a real bad one. It is all probability and
I can assure you that it is no fun seeing your system
with 5+ sharpe ratio trading your capital down 15% to 20%
in only a few weeks when the system has been in use for
just a year or so :)
The main usage of sharpe is that you know the probability
of the risk/reward that a particular system is giving you
and than you can track your system by an objective means
whether it is dropping out of its normal performance range.
For a system with 5+ sharpe ratio, its equity curve would
bounces back very fast once the "bad" period is over relative
to say a sharpe of 3 or 4.
> 5-10 - good, you can rest now!
>
> This scale probably shows my bias and state of affairs.
>
try 6 to 6.5 as the cut off for comfort level.
remember 5 implies about 20% of the time in terms of probability
you will suffer some losses on a annualized basis.
so, a 6 to 6.5 scale reduces that to less than 15% or so.
and finally, do not trust the TS sharpe result.
Export the equity list out and compute that in Excel.
=====
Lawrence Chan http://www.tickquest.com
Transform market data into opportunities
|