Modify Field Dialog: Desired Output

The Modify Field Dialog allows you to analyze and modify the properties of a field. The Desired Outputs page allows you to add or change the desired outputs being used in a prediction.

Ä    Note: Fields that are currently being used by the Solution Service cannot be modified.

&  For help with predictions, see Predicting and Modeling Financial Data.

Selecting a Value to Predict

To specify the value to predict, use the following options.

¨    Value to Predict

This value specifies the name of the data series field to predict.

Ä    Note: Only fields from the current data series or group are available for the desired outputs. If you would like to predict the value of a field in another data series, create a calculated field that is equal to the value of the field you want to predict, then use that calculated field for the desired output.

¨    Samples in Advance

This value specifies the number of values to look ahead in the desired output field.

:    Example: If you are trying to predict a value two trading days in advance from a data series with daily data, this value should be set to 2.

Ä    Note: If this value is zero, the prediction will be for the current value. This can be useful when predicting values like the optimal signal, which suggest actions based on future values.

¨    Preprocessing

This value specifies the type of preprocessing that should be used on the desired output. Preprocessing is typically used to give the desired output a more uniform range, often it making it more predictable. TradingSolutions automatically postprocesses the predicted values when preprocessing is used.

:    Example: When Percent Change preprocessing is used, the percent change in value is the value that is sent to the neural network. Any values that are received from the neural network are treated as a percent change in value and converted automatically to what the value would be if that percent change occurred.

Ä    Note: This value is only available when the prediction is for at least one sample in advance.

·      Percent Change

This option is typically used when you are trying to predict a continuous, positive, non-zero value that does not have a constant range, such as price information.

·      Change

This option is typically used when you are trying to predict a continuous value that may not have a constant range, such as some momentum indices.

·      None

This option is typically used when you are trying to predict discrete values that have a constant range, such as entry/exit signals.

Ä    Note: Preprocessing on outputs is automatically reversed when values are produced by the neural network. For example, prices predicted using percent change preprocessing will result in prices which are determined by a predicted percent change from the appropriate previous value.

þ    Output Data is Pre-Differenced

This setting indicates that the values of the desired output data represent change or percent change data. It affects how Optimize for Directional Accuracy operates in that the criterion component will need to calculate the change from the previous value is the data is not pre-differenced. It also affects how the Directional Accuracy analysis is calculated on the Directional Accuracy Sub-page.

Ä    Note: This setting is repeated on the Advanced Topology Settings Dialog: Criterion page.

p    Signal Postprocessing Settings

This button displays the Signal Postprocessing Settings Dialog, which allows you to view or modify the postprocessing settings used for a predicted signal.

Ä    Note: This button is only available when the predicted value is an entry/exit signal.

Specifying Multiple Outputs for a Prediction

Most predictions will have only one desired output and, therefore, only output one field. To predict multiple values at one time, select Predict multiple values with this prediction.

This option displays the list of outputs for this prediction. When this option is first selected, only the current desired output will be listed. To add an additional output, press the Add New button and specify the information for the selected output. To remove extra outputs from a prediction, press the Remove button to remove the selected output.

There are two ways to approach predicting multiple values in one data series:

·      Create individual predictions for each value being predicted.

This option will typically produce more specialized models of the values being predicted. This is typically preferable, especially with values that are hard to model. However, highly specialized models may have trouble generalizing results for new values. By default, a cross validation set is used to help prevent models from specializing on the training set.

·      Predict multiple values with one prediction.

This option will typically produce less specialized models of the values being predicted. This is useful for reducing the amount of specialization on the training data, which in turn may allow the models to generalize better on new values.

Ä    Note: Additional outputs can be added to a prediction at any time. However, once a prediction is created, individual outputs cannot be removed.

Ä    Note: Postprocessing of predicted signals is performed separately for each output.

What Do I Do Next?

When you are finished updating the properties of the field, press the OK button to save your changes or select another property page. If you would like to exit from this dialog without performing any changes, press the Cancel button.

Additional Property Pages

The Modify Field Dialog has many different property pages that will be displayed as is appropriate for the type of field being modified. For information on the other property pages available, see the help for the Modify Field Dialog.

How Did I Get Here?

The Modify Field Dialog appears when you select to modify or analyze the selected field from the Modify Data Series Dialog or the Modify Portfolio/Group Dialog. It is also displayed when you select Modify Field… from the context menu for a field in a chart or spreadsheet.