|
PureBytes Links
Trading Reference Links
|
> My daytrading system for the S&P 500 must be in a draw-down... is
> there any data, studies, or information in general about how long a
> draw-down period should last?
Until it stops. :-)
Seriously, I very much doubt there's any general data about this.
It's going to be very dependent on the trading methodology. Some
systems are very sensitive to market volatility, and may improve or
degrade with increased volatility. Tight MM stops can make you more
susceptable to losses unless your system picks excellent entries.
Have you gone back and carefully studied the system's past history?
Is this drawdown really out-of-character for the system? The dollar
amount may be higher, but that might be because of the increased
volatility. Is the actual pattern of losses unprecedented? Look
closely at the trade-by-trade performance. Looking just at the
system report can give you a "view from 50,000 feet" that misses out
on all the pain and drawdowns you would have suffered during actual
trading.
> Although the last 5 trading days in March didn't work out very
> well for me (that's when my draw-downs started) I still made ...
Sounds like you've been having an excellent run. I think you're on
the right track to stop trading until your system gets back on its
feet again. Spend the time testing and trying to find what's gone
wrong with your system. When the system seems to be recovering, you
might want to start out with reduced leverage. Given your monthly
ROI figures, I suspect you're leveraging it pretty aggressively.
> (My research shows an optimum stop placement to be
> 4.8 points in the S&P -- Risking $1,200 per trade).
Again, this is going to be very dependent on individual systems, and
on how those systems respond to market conditions. Volatility has
been going through the roof lately, so maybe that stop is too tight
now. Have you been getting stopped out a lot, only to have the
market turn and go the direction you expected? Try backtesting a
larger MM stop to see how it works for the last few weeks and the
last few months. Maybe you should scale your MM stop (and/or other
system parameters) based on recent volatility so it can adapt to
changing market conditions.
If you develop a system over a sufficiently long time period (I
prefer at least 100-150 trades in a test), you ought to capture
enough different market conditions that you can be reasonably
confident of the system's performance in future conditions. Try to
develop robust systems, i.e. ones that have few optimized parameters
and that perform well even on less-than-optimal parameter choices.
That reduces the chances of curve-fitting the system to past market
behavior. It's a very good idea to test the system's out-of-sample
performance to see how it holds up on unseen data; e.g. optimize the
system on 1998 data and see how it performs in 1999.
And hold your breath, and hope your system recovers...
Gary
|