|
PureBytes Links
Trading Reference Links
|
RT’s,
I wanted to tell you about something that a friend and I tried. We
decided to have a contest to see if we could forecast the markets. We
selected a basket 10 commodities as we were interested in trading futures.
Each of us tried our best to simply pick whether each future would be up of
down the next day. I started out pretty good even picking 9 of 10 one day
but soon my friend started out doing me as my average dropped down below 5
of 10. After a few weeks soon it became apparent that we were both
averaging about 5 of 10 or about 50/50. Now I Figure that many forecasting
systems will turn out like we did. If a neural net or and other software
cannot do better then 50/50 then it is no better then flipping a coin.
Sometimes its easy to get fooled because like myself a system or program
can select 9 of 10 at least once. I have tried a number of programs now,
and after blind back testing they ended up about the same, either coming
out 50/50 or worse. I figure that a lot of vendors got rich last year on
systems that said to buy some kind of stock almost anything would do. Now
that the stock market is going sideways things aren’t so rosy. Some systems
are so clouded in variables and rules that you can’t really tell how things
are working. After asking many pros success seems to come down to
experience. Even great systems are probably a result of experience.
Good Trading,
Brent
|