Lets say I want to estimate a replicating portfolio by doing a linear regression between the returns of a long-only portfolio and several long-short factors like Fama-French 5-factor or Betting Against Beta (CAPM framework).
I discover that I can replicate and explain about 60% of the variance in my long-only portfolio ($R^2$ of ~0.6) with 3 significant coefficients to some of the long-short factors mentioned above. Does this comparison even make sense when replicating a long-only strategy with multiple long-short strategies?
