|
PureBytes Links
Trading Reference Links
|
Simon Roberts usefully pointed out:
> Owen,
> What is legitimate is to define what % swing is significant for any
> security.
> Having decided that a swing of x % is significant, for whatever time frame
> you are studying, you can then draw x percent swings on your chart, and as
> they develop clock the recent turning point.
> You can be certain that this is purely mechanical, at the current time,
and
> does not rely on any discretion of interpretation.
> Swing traders do this all the time. Read Robert Krauz, or Robert Miner in
> Dynamic Trading.
Thanks, Simon. This is pretty much what I was attempting to do. The
instructions, since lost with a dead hard drive, said, basically, "enter
when the price has dipped X% below the high of the peak or risen X%
above the low of the trough." My hope was that this would compensate
for the crystal-ball effect by delaying entries until after the peak or
trough would have been recognized in real time. And, of course, this
technique seemed lovely when applied to historical data.
In real time, one would have lost money for a while, taking signals that
appeared in real time, but not in hindsight Periodically, the indicators
changed their minds and decided that instead of the series of small up
and down moves the "system" had identified over the last week, an up
(or down) trend had been in effect the entire time. When run repeatedly
as a system, it would signal buys and sales according to the short-term
moves, and then suddenly eliminate them all and retrospectively report
signals that fit a pattern that had not appeared until the latest bar.
This occurred during a generally rising market. My interpretation was
that as the price grew, the absolute size of an X% move expanded with it,
and the peak and trough functions recalculated the past, based on X% of
the latest price. No idea whether the effect would have been mirrored
exactly on the way down, as I tossed the technique immediately. I'd have
given a lot for peak and trough functions that viewed the past as it would
have been seen in real time--assuming that I've diagnosed the problem
correctly.
Thanks for reminding me about Krauz and Minor. I know of them, but
they've sat on my "get around to them someday" list. I'll move them up.
Owen Davies
|