|
PureBytes Links
Trading Reference Links
|
>Before calculating expectancy, I first toss out the largest win as
>an outlier, reducing the total number of wins by one. If it's an
>outlier, then discarding it makes the expectancy more representative
>of the system's expected performance. If it isn't an outlier, then
>discarding it won't matter anyway.
>Granted, if you have *two* outlier winning trades in your data, then
>this crude technique won't work.
In the past few months I started to learn the importance of money management
(the hard way... fortunately I slowed down before hitting "ruin") and when I
read this post of calculating expectancy, I was thinking another method to
get rid of outliers is to use a "three sigma" filter. Basically, its a
simple filter that removes all points that lie outside of 3 standard
deviations and then recalculates the SD and checks for outliers again until
all points lie within 3 SDs. This would solve the problem if there happened
to be more than one big winner. If you want to be more conservative, I
guess you could use 2.5 SDs (or for those who get off on Fibonacci's, 2.618
SDs :P ).
-- Wyatt
|