Develop Multiple Regression Model to Predict Index Fund Performance

1. Choose your favorite Index fund (e.g., DJIA, S&P 500, Wilshire 5000) and get at least 25 years of monthly averaged historical data of index performance

2. Choose at least three market indicators (e.g., unemployment, GDP, 10-Yr. Treas. Rate, CPI, Initial Jobless Claims, inflation of US dollar) and get at least 25 years of historical data.

3. Create a multiple regression model that predicts the value of the monthly CHANGE of the index fund based on the three market indicators that you have chosen.

4. Perform a residual analysis to determine the validity of the regression model

5. Determine the usefulness of each of the independent variables using by performing a hypothesis test on each of the slopes for the independent variables.

6. Use a Durbin Watson statistic to determine if autocorrelation is present in your data.

7. Check for interaction between your variables. 8. Predict the average index value based on a 95% confidence level for three different combinations of values of your market indicators.  

Report should include the following:

1. 10 page limit, double spaced

2. Sections on Introduction, Methods, Results and Conclusions

3. Appropriate figures and tables to describe your data, analysis and results

4. List all assumptions made in your analysis

Report will be graded as follows:

•  Introduction. Does the report contain a clear introduction describing the fund and market indicators chosen? /10
• Methods. Does the report contain a methods section that describes how the regression model was created? Are the methods for the hypothesis tests described? /10
• Data. Are the data included in the report or appendices that were used to create the model? /10
• Results. Are the results of the regression analysis clearly described including the model and appropriate measures of the goodness of the model (e.g., coefficient of determination)? /20
• Residual Analysis. Was a residual analyses performed and appropriate plots included in the report to support the residual analysis? Are any of the assumptions of the regression analysis violated as demonstrated by the residual analysis? /10
• Hypothesis test for the slopes. Was the hypothesis test for the slopes performed, the null and alternative hypotheses listed, and correct conclusions made? /5
• Durbin-Watson. Is the Durbin-Watson test performed and described? /5
• Interaction. Did you describe how you checked for interactions between the variables and correctly conclude if there were any? /5
•  Predictions. Did you make a prediction and determine the 95% confidence interval for three different combinations of values? /5
• Conclusions. Is there a conclusion section? Are the conclusions reasonable based on the analysis? /10

You may work with a partner to collect the data and create the model, but you each must turn in your own report with your own analysis and conclusions.