|
PureBytes Links
Trading Reference Links
|
Look at your test data. How much are you looking at? I've seen examples
where 10 years of daily data produced 50 trades and a profit factor of
6. But 50 trades is statistically problematic in this case. The equity
curve it produced may be a random walk masquerading as a sexy profit
vector. Such a system will fail as soon as you start it in real time.
I've also seen examples where 3 months of intraday data produced 50 trades
and a profit factor of 5. But 3 months is a problem because it's too short
of a time span to test how your system weathers as the market
mutates. This system will fail in the future because the system was
curve-fitted to only 3 months of market behavior.
Also, don't forget the recent tech stock bubble, which is a special
case. You may think that a test period between 1999 and early 2001 is long
enough, but it represents a market behavior that was kind to the most
mediocre trend/momentum based systems. Consequently, there are a lot of
systems floating around following the late 90's bubble and promise amazing
returns. Find out their test periods. Do they include recent market data
since March or April of 2001? And if they do, do you see any "rolling off"
in their equity curves?
50%+ annual returns is in itself asking too much, because most people don't
achieve it. But that doesn't mean you can't develop such a system. You
won't get it from people who yap about their systems, either. Those who
truly have it don't talk about it. Having said that, you need to accept
the possibility that it may take you another 2+ years. Don't give up.
At 06:59 AM 12/6/2001, Noel Falzon wrote:
>Is having a system that consistanly makes 1% a week
>asking too much ?
>
>Having worked hard with TS2k for over 2years, trying
>various systems (several that I and friends have put together as well),
>So many fall apart in realtime.
>
>Is anyone consistanly making say 2% a week (100% a year), over the
>last 3 or 4 years.
>
>Is this asking too much ?
>
>Noel
|