In this example we picked a stock and selected a MESA-measured cycle-based indicator that we believed would be predictive, and then fed it into a neural net.  We did not need to build a trading strategy beyond the one that the Prediction Wizard creates. This is a fairly typical example of how to take advantage of the neural nets in the NeuroShell Trader.  This and the other examples assume that you are already pretty familiar with the mechanics of using the NeuroShell Trader.  If not, please review the Tutorial and videos first.

The stock we picked is OfficeMax.  We like this stock because its price varies cyclically without major trends over the period of the data.

First, we had to choose the indicators we believed would be predictive of the cyclic variations.  From our experience, we know that a BandPass Filter that is tuned to the measured dominant cycle of the data extracts the coherent component of the market activity.  If the market has a short term coherency, then we can expect the dominant cycle cyclic component to continue into the future for a short while.

Next, we inserted our prediction with the Prediction Wizard.  We decided to predict the change in the close 10 days ahead of time.  For inputs, we chose only the MESA8 BandPass indicator.    Note that the neural network in the prediction automatically assumed a reversal system: exiting a long into a short, and vice versa.

The buy signals are blue upward triangles, and the sell signals are the red downward triangles. If you zoom in on one of these signals, you will see an X on the open of the next trading day. This is the date and price of the simulated fill. The NeuroShell Trader assumes you are looking at your chart at the end of the day after downloading another day of data. Any orders you place will be filled near tomorrow's open price.

To enter the Prediction Wizard and examine the results, right-click on the label at the top named "Trades: Predicted 10 Bar %Change in Open". When the pop up menu appears, select "Modify Selected Prediction." You will be placed at the end of Prediction Wizard.  Click on the Prediction Analysis button, and then select the Trading Statistics tab.

While the buy/sell points were not exactly at the places that we would have wished they occurred, they were nevertheless good enough to make an annual return on trades of 45.3% during the five year period when the stock declined in price. This is exclusive of commissions and slippage, of course.