[zerosum]

[ QUOTE ]
Do you have more specifics on how you find the top 20 stocks?

[/ QUOTE ]

I can offer a general explanation.

The selection process is an expected return factor model, where the expected return is relative to the universe of stocks from which the factor measures and sensitivities are derived. Multivariate regression is used to disentangle the interrelationships among factors, allowing for consideration of the *pure* expected return to each particular factor. Each stock's sensitivity to each factor is estimated, normalized and expressed as *B* or Beta. A particular stock's expected return is the sum of its betas multiplied by their respective expected return factor values.

Here's an example, using k factors:

Expected Return = B1*F1 + B2*F2 + B3*F3 + . . . Bk*Fk

The model automatically adjusts to changes in market sentiment regarding the expected return that should be accorded each factor. The underlying assumption of the model is that the present importance of a particular factor is the best estimate of its future importance.