I'm starting over.
I saw good results when I did a quick and dirty test on the model using S&P constituents over the various years. I'm going to start this time by putting all that down.
This one is the S&P500 that existed in 2013 edited for companies that don't work with this model (financials primarily). I run the model on that group as if it was early 2014, just after most of the companies have issued their 2013 results. The model predicts the best performing stocks. I'll then graph the actual results over various time frames.
This was the genesis that started this testing. I'm going back to the beginning and recording those results.
Note: the edited S&P500 portfolio has 315 stocks in it.
One Year Return
Two Year Return Three Year Return Four Year Return Five Year Return Six Year Return Year Two Return Year Three Return Year Four Return Year Five Return Year Six Return Observations:
These results are different for the S&P500 Y-6 ones that I published previously. The reason is I defined the portfolio differently. This testing had 315, the other had 300. In the previous test, for the portfolio I removed all finance companies without regard to whether they might still have had the information needed to perform the analysis. With this analysis, I looked at the data and removed those that didn't fit.
How the initial portfolio is defined is important, however these results, as well as the original results are both good, and indicative of the model's predictive ability.