|
PureBytes Links
Trading Reference Links
|
Pardon for interrupting, but when intraday version is offered,
systems can be tested out on tomorrow's data; a lot of time units can
be counted in a week or two. Does this counter the
inefficacy of testing on past data?
Good economists do not believe in empirical testing, either; on the
grounds that decisions leading to price variations, in free markets,
are the result of human action (but not human design), and can not be
analysed via the infinitessimal calculus; Why? because, being human
actions, as opposed to natural material, the data advances in
discrete units, not infinite ones.
As a consequence, good theories can only be derived from
incontrovertible, a priori, axioms; directions can be determined,
exact outcomes cannot.
P.S.
One system developer, and money manager, told me, about ten years
ago, that applying a whole handful of technical indicators in a
system was an unrealistic test; on the grounds that you will be
asking the market to do too much for your system, and the market just
won't do that; it is, rather, a better approach, he said; "..to use
an MA, to build some predictibility into your analysis".
Hope something useful can result from this observation and anectdotre.
|