|
Home Page
Yahoo Charts
| | The Standard GPF is no longer available, and upgrading to the GPF Pro
is recommended. Only a few of you have been using the Standard GPF
anyway. Here are the specific features of the GPF Pro:
- Possibility to have a custom objective. This can be used to gear the genetic
algorithm to use your parameters. You can for instance create your own Excel
formulas to calculate the profit & loss potential using stop losses, profit caps, ...
A more interesting use which has been requested by users consists in applying
a profit & loss objective from another (dependent) stock so that a causality be
detected in patterns, for instance, using pattern data from Intel to maximize P&L in
Compaq.
For simplicity, all data must be present in the same workbook.
- Adding your own Boolean indicators to filter patterns. This will force the
G.P.F.
to skip bars if your indicator says so. It is a simple way to
jack-knife your data set.
- Pattern Topology: Although this should be used with much care, it can be
interesting to better extract consistencies in winning patterns in specific market
configurations. It is possible to ask the G.P.F. to focus on reversal or
follow-through patterns. A pattern strength has even been added to make sure mild
signals are ignored.
This must used with caution though. For each bar, the Rate of Change
over the n last bars, with n being your maximum search space or bars back
(default: 5) is first calculated. Then the pattern strength is
calculated as the profit potential (according to your optimization
objective) multiplied by that RoC. As this uses the profit potential, it
obviously peeks into the future. This would make it a very unsuitable
neural net entry. But for genetic algorithms, as long as you know what you
are doing, you may do so. You must not be fooled by the beautiful
equity curve, which will obviously look great, but this is the best way to
determine whether a pattern has consistently revealed a turn-around or a
run-away rally.
- Full Out-Of-Sample Analysis
Data at the end of your sample set can be held out and reserved for out of
sample analysis. While this is not a requirement like in neural nets,
we believe this gives more credit to your patterns.
- For intraday traders, it is now possible to focus on a particular time frame.
- Intraday features have been moved to the G.P.F. Pro, and are no longer part of the
standard G.P.F.
|