| Forecast
Accuracy We receive numerous requests to post the performance of our forecasts versus the actual outcome. We currently do not do this because of the false sense of security it may offer. While the use of artificial intelligence to produce these forecasts has never ceased to amaze us trying to model the behavior of very complex systems such as global financial markets is a very daunting task and is always full of surprises. The best measure of our performance is you the viewer. By monitoring the forecasts over time you can judge for yourself their strengths and weaknesses and their appropriate use. As a surrogate we post the standard deviations for each forecast value which is a measure of how much the forecast value can possibly deviate from the actual value. By using the standard deviations as a measure of uncertainty, there is always some chance the actual value can be quite different than the forecast value. We also provide an example of a typical graph of the predicted versus the actual values from a neural network model (in this case, the Dow Jones industrial average). As you can see in the graph below, over the long term the predicted values match very well to the actual values, although there may be some short term deviations. In developing our forecast models, the neural networks may be set-up to increase the accuracy of the most current values, but a great deal of general market behavior may be lost in the process. Likewise we can produce models that encompass much more general market behavior, but possibly at the expense of present-day accuracy. Thus in developing our forecasts there is a constant tension between forecasting current values and forecasting general market behavior. The Financial Forecast Center's policy is to favor the greatest range of general market behavior over more accurate present value but less generalized forecasts. Usually this means looking at global markets going back at least into the 1970's and possibly into the 1940's to get a broad range of behaviors to develop the most generalized and robust forecast models. Our primary goal is to forecast the storms on the horizon over the breezes of today. Dow Jones Index of 30 Industrials Stock Forecast Accuracy Dow Jones Index of 30 Industrials Stock Index
at Close of Month
Predicted Vs. Actual |
