[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Testing systems with futures



PureBytes Links

Trading Reference Links

Yep, it's the old problem, figuring out what, if anything,
the reported results mean when you're not working with
stocks.  I don't use systems, just play with them occasionally
when I want to test an idea for possible discretionary use,
and so haven't paid much attention to this.  Thus, I've just
realized something that everyone else probably knows, but
I don't understand at all:

I have always assumed that there must be some sort of
proportionality in the system tester's results.  That is, even
if the numbers don't work out right in absolute terms, a
change of a given magnitude should be reported the same way
each time.  However, I just tested a "system" on T-bonds, with
a tick value of 1/32.  Looking at three consecutive trades on
7/28 and 7/29, the chart shows profits of 8/32, 13/32, and 12/32.
(I could live with it!)  The system tester reports these as being
3.57, 4.47, and 3.60, respectively.  Somehow, that didn't
look quite right, so I did the arithmetic.  The first figure translates
to a reported profit of .44525 for each 1/32 of price change that
appears on the chart.  The second comes out as .3438 (rounded.)
The third is exactly 0.3 per 1/32.  Looking at two other trades
gave values of .269 and .1329!

Does this make sense to anyone?  If not, can anyone figure out
where I've gone wrong?

EQUIS:  Please enter this as one more plea to include a little
database that lets MetaSTOCK(!) report accurate win/loss
figures when dealing with futures contracts.

Many thanks.

Owen Davies