|
PureBytes Links
Trading Reference Links
|
>Furthermore, Thomas Stridsman's book as become my bible and he
>strongly suggests using RAD to perform system testing.
>
>Would love to hear from those on the list that have *insightful*
>experience using this type of data, good or bad.
Bad.
1. System tests performed on ratio-adjusted data don't equate to
results seen in real trading, in my experience. I see several
instances (especially for long-term trading systems) where the
relationships between critical turning points are reversed in ratio
adjusted data; e.g. a higher high in real life might be a lower high
when ratio-adjusted. For this reason, ratio-adjustment can fail to
identify long-term trends (say over a 10 year window).
2. Signals generated from ratio-adjusted data may or may not occur
when applied to true contract data. If you have a system that works
well on ratio-adjusted data, that tells you nothing about how it
will perform on a real futures contract, and vice versa.
My conclusion: Use spot market data for generating signals, which
get traded on the actual historical market data. That way, you're
counting on real, un-adjusted data to make decisions.
Pinnacle told me they now offer historical spot market data along
with their daily futures data feed, for an extra $3/month.
--
,|___ Alex Matulich -- alex@xxxxxxxxxxxxxx
// +__> Director of Research and Development
// \ Unicorn Research Corporation -- http://unicorn.us.com
// __) HTML FORMATTED MAIL SENT HERE WILL BE REJECTED AS SPAM.
|