September 2015
Beginner to intermediate
336 pages
7h 44m
English
—Mary Malliaris, Loyola University Chicago
We developed clusters on a training set and then used the cluster assignment and derived variables as inputs into a support vector machine model. The results show that both oil and the S&P play an important part in forecasting gold’s direction tomorrow. The support vector machine does a very good job predicting Up movements but is not reliable at predicting Down days. We used a 24-year training horizon to forecast daily for the next 25 months. The results were stable over the two-year-plus period, with Up-direction forecasts correct 84% of the time and Down-direction forecasts correct 21% of the time.
Predicting price movement in gold ...
Read now
Unlock full access