Returns-based style

Why the following statement is incorrect?

Returns-based style analysis is better because it can more easily detect style drift over time.

Holding analysis u can see the holdings. Nothing is easier to detect then to see it with your own eyes.

But for return based, you can just run the regression. For holding based, you have to assess each holdings contributions to portfolio style.

Obviously you need to access to the holdingsā€¦ If you dont have access to the holdings then you arent doing holdings analysis.

What is easier, to look at how fat a wallet is and take a guess how much money is in it (returns based analysis)? Or to open out a wallet and count the individual bills (holdings based analysis)?