I ran a forecast awhile ago (September), but forgot to post it here. At the time, the S&P 500 was trading about 820. Here's the results of what I wrote to LPS in a private message.
"Remember that thread on forecasting stock prices? Well I ran a 6 month forecast for the S&P 500, using the bond yield data, Fed funds rate, and 3 mo. Treasury yield, and came up with a forecast (multiplicative decomposition with linear regression) of about 1130, with the 95% confidence range falling between 1101 and 1158. The Mean Absolute Percentage Error was 2.06%, and the R-Squared was 99.68%. The RMSE was 22.88, and the Durbin-Watson was 1.51 (which would indicate some serial correlation-just what we'd expect because of the normal business cycle) These are pretty friggen good results. The forecasted graph fits the data fairly well. Let's see what happens."
This forecast was using ONLY Treasury data, specifically excluding other data such as Aaa/Baa Corporates, which would probably reduce the forecast, but not the trend because of the spread on corporate bonds is still relatively high, indicating a higher uncertainty of corporate results. I'll run another forecast including some sort of Aaa/Baa data and see what we come up with.
Danno |