|
PureBytes Links
Trading Reference Links
|
Somewhat in line with what Rick says, over the past four years (since summer
1998(the index future markets basically changed in late summer 98)), I have
never backtested a new indicator nor a change in my methodology more than
over the past 100 days or so.
Granted that since the basic construct of my indictors does not and has not
changed, what I am backtesting are my entry filters, stops, targets, and
exits.
So, in a practical sense, I hold that testing 1994 data for trades to be put
on Monday morning is not relevant. What works now, works. Yes, I know the
arguments against such a small sample, however I am looking for results now.
If a part of the methodology begins to fail, it becomes rather evident
quickly.
I doubt I would have been part of the crowd that watched my portfolio lose
80 per cent of its value over the past 18-30 months listening to the
investment community hammering home the buy and hold line at "any" cost.
If my method stops working next week or next month by failing to return at
the same profit level as I enjoy now, I'll change it now based on today's
data. And I'll backrest it over the past 3 month's data..not 1997 or 1993 or
1987's data.
As a matter of interest, the only changes we've made in recent times have
been in the entry filters in the SP/ES and ND/NQ. We are using the same
exact methodology in the currencies, bonds, grains, beans, and the crush as
we did five years ago. So we don't change methods at the first losing
day...but if we have to, see above. We use the data that's coming across
now, not data from the great commodity bull markets of the 1970's.
FWIW.
Jim
|