August 21, 2018

IN THIS ISSUE

WEBINAR OFFER REPEATED

______________________________________________________________________________________________________

by Denham Ward

You’ve probably noticed that some of your models work better on trending markets while others work better in cyclic markets.  You can manually switch models based on market conditions, but what if you could automatically switch your trading system according to the maximum value of a selection criteria like Equity, Recent Equity Curve Correlation, Recent Change in Equity, Winners Minus Losers, etc.

The method we are going to demonstrate works for both rule based systems and neural network predictions.

The first step is to figure out the maximum value of your selection criteria across all of your trading systems.  If you had four systems, you would use the Max4 indicator found in the arithmetic indicator category with the appropriate criteria calculation/indicator for each system as shown below.
MaxCriteria = Max4( System1Criteria, System2Criteria , System3Criteria , System4Criteria )

where System1Criteria might be Momentum(SystemEquity(Trading Strategy #1), 10 ) if you wanted the system with the maximum change in equity over the last 10 bars.

Note: If more than four systems, then you will need to use a combination of Max# indicators.  So for instance if you had seven systems you would use Max2( Max4 ( C1, C2, C3, C4 ), Max3( C5, C6, C7 ) )

Once you have figured out the MaxCriteria value, then you would need to create a controlling trading strategy where each entry/exit condition is nested in a series of IfThenElse rules that select which trading system to use.

For instance the Buy Long condition would look like the following:
If System1Criteria = MaxCriteria Then
Else
If  System2 Criteria = MaxCriteria Then
Else
If System 3 Criteria = MaxCriteria Then
Else

Note that in the case of a tie when two or more systems have the same maximum value, then the one that occurs first in the nested if/then/else list will be the one used.
 The Long Entry condition above is the first of four nested rules that compare the 10 period change in equity for Prediction 1 against the max value for all of the predictions in our example chart.
You would create similar nested if/then/else indicators for long exit, short entry and short exits as needed or desired.

The method described above works for ranking different trading systems for a SINGLE security on one chart page.

If you want to balance all of the stocks in your portfolio, check out the model in the July 2018 Newsletter that uses the upper rank indicator.  The newsletter is available from our technical support website:  NSTSupport.wardsystemsgroup.com.

______________________________________________________________________________________________________

BUILDING BETTER MODELS WITH NEUROSHELL TRADER WEBINAR
A few years ago we presented a webinar called Building Better Models with NeuroShell Trader.  Last month we offered the materials from this webinar to our users for \$179.  We had a lot of interest and have decided to repeat the offer this month.
The webinar covers the following topics:
• Preventing Overfitting
• Optimization Tips
• Pattern Recognition is Key
• Automatically Adapting Models to Changing Markets
• How to Use Adaptive TurboProp2, Neural Indicators, and Adaptive Net Indicators
• Different Uses for Averages: Variable Length Moving Avg and Average Trading Rules
• Practical Suggestions: How Much Data, What to Do When a Model Isn’t Working, and Ensemble Models