I previously wrote a defence of CAPM in reply to this post. Even given that the empirical empirical evidence shows that high beta portfolios do not generate higher returns, the beta is still useful for investors.

It is possible to argue that CAPM is correct but that the definitions of market portfolio used to test it are too narrow. The definitions used to test it, are those used to apply it. Given that, as used in the real world, CAPM does not correctly predict returns, I will accept the empirical evidence for the sake of argument. This does not mean I have changed my view that the CAPM, or a very similar model, is fundamentally correct.

So, accepting the empirical evidence, is CAPM completely useless to investors? I would say that beta is still useful.

Beta is still a useful measure of risk. It is true to say that an investor with a high beta portfolio is taking a higher risk than one with a low beta portfolio.

Therefore, if the empirical studies show that there is no reward for taking on that higher risk, then investors should avoid the risk and hold low beta portfolios.

Combining this with the evidence for the value and size effects, and the poor performance of index trackers, the best strategy seem to be low beta small cap value.