An asset's Beta is the product of its market correlation and volatility ratio. Highly volatile assets can have beta near 1 even when they have low market correl

An asset's Beta is the product of its market correlation and volatility ratio. Highly volatile assets can have beta near 1 even when they have low market correlation. That's why CAPM statistics usually include R-squared along with Beta. More generally, in constructing optimized portfolios, correlations are important, not Betas. To your original point, 30 large firms are generally plenty to capture the market factor, regardless of weighting. For one thing, market-weighted indices are dominated by their top names (the top 30 names in the S&P 500 constitute over 40% of the index presently). More importantly, summing quickly suppresses uncorrelated variance, and there's a point after which adding additional names does very little. (I think there's a phase transition, actually – see Lello et al's recent report on polygenic risk scores, which use LASSO regression...[1]) [1] https://arxiv.org/abs/1709.06489 https://arxiv.org/abs/1709.06489