The Optimization tab allows you to specify how you want the Trader to search for the best model.

The maximum number of inputs defaults to 6, which we believe is a reasonable value to prevent overfitting the model to the training data.

You can specify an exact number of minutes to optimize the model, but in general it is not necessary to specify a value with today’s fast computers.

Optimizing a model across all chart pages causes the Trader to select the same inputs and corresponding parameter values for each issue in the chart.

If you have preferences for how long you want to be in a trade on average, you can enter values for the shortest or longest average trading span. The resulting model may not match these values precisely, because the first priority is the objective you select on the Training tab.

The Trader offers four different optimization methods:

GeneHunter is our original pure genetic algorithm optimization method.

Evolution strategy optimization generally works best for most models.

Swarm optimization works fine on problems with fewer variables or for fine tuning models that were initially trained with evolutionary strategy optimization.

Brute force optimization examines every possible solution but can take forever working on a problem with a lot of variables.