|
PureBytes Links
Trading Reference Links
|
Michael,
On Mon, 14 Aug 2000 13:17:06 +0200, you wrote:
>about nine years ago I worked with a German trading company on exactly such
>a project. We did our own C++ programming (quite a novel language at that
>time) with a programming staff of about a dozen highly qualified people,
>plus two Cray supercomputers and a number of UNIX workstations.
Maybe, you were too early, and you had too much computational
resources available ...B-)...
>Now with the advent of new electronic exchanges, and you are right to
>mention the DAX contracts here, the situation might slowly change. Still,
>even with the DAX, I don't know if there would be a bid/ask history on
>options that goes back a sufficient number of years.
Is it really necessary to have bid/ask histories? Imo, most TA people
do not use them in their systems, even if they may be very helpful.
>I can only repeat that I have yet to see a real mechanical option trading
>system. If you know of one, or have written one yourself, I'd be interested
>to know.
I have done some work on an option trading system, which not
concentrates on "non-directional strategies" (like most option traders
seemingly do), but looks at option markets as _dynamic systems_ (in
contrast to most stationary option pricing models).
The system employs two components with different "frequency horizons",
where the "low frequency" part (5 ticks or more) deals with the
current "dynamic market trend" (based on a _dynamic_ option pricing
model), and the "high frequency" part (1 tick or more) handles
external disturbances coming along from economics, politics, etc.
The "high frequency" part as a pure TA model (, which I also use for
practical trading at this time,) is based on a sliding ROC indicator
working with "continuous" prices "at-the-money" & 20 days to expiry.
Results are satisfying, even if the out-of-sample profit is only about
2% of the value that could be achieved in a-posteriori calculations.
Presently, I work on some refinements of the "high frequency" model
and on basic out-of sample tests of the "low-frequency" model. For
both I have a long list of potential refinements and ToDos, which wait
for implementation because of my limited time and resources available
for this project.
Let me know, if you are interested in any detail.
mfg rudolf stricker
| Disclaimer: The views of this user are strictly his own.
|