| dc.description.abstract | The purpose of the study is to investigate whether price movements in the U.S. stock market can be used predict price movements in the U.K. stock market. The stock markets are represented by the S&P 500 and the FTSE 100, respectively.
We investigate three combinations of explanatory and dependent variables for the predictions: (i) using explanatory variables from the S&P 500 Stocks to predict the intraday returns of the FTSE 100 Stocks, (ii) using explanatory variables from the S&P 500 Stocks to predict the intraday return of the FTSE 100 Index, and (iii) using explanatory variables from the S&P 500 Index to predict the intraday return of the FTSE 100 Index.
Artificial Neural Networks and Ordinary Least Squares Regression are employed as the prediction methods, and they are separately tested in the study. Moreover, three virtual traders with different levels of risk aversion are employed to trade based on the predictions of the model.
Initial data analyses demonstrate that there are significant correlations between the intraday returns of the U.S and the U.K. stock markets. Furthermore, the predictions indicate that it is possible to predict the intraday returns of the FTSE 100 Stocks and the FTSE 100 Index, based on information from the S&P 500 Stocks and the S&P 500 Index. Specifically, by simulating trades based on the predicted returns of the FTSE 100 Index, we obtain and accuracy of 71.9% and an annual return of 44.2%. This is considerably higher than the return of the FTSE 100 Index of the same time period, which amounts to 3.3% on an annual basis.
Predicting the intraday returns of the FTSE 100 Stocks yields inferior results compared to predicting the FTSE 100 Index, with an accuracy of 56.1% and an annual return of 22.9%.
Both the ANNs, which represent a sophisticated forecasting method, and the OLS, which is a simpler forecasting technique, yield similar results when predicting the intraday return of the FTSE 100 Index. Hence, the predictive power of the model largely originates from the data employed. | en |
