Views from Clifford S. Asness
One of the big problems with quantitative work is data mining. (I will
argue here that it’s not just quants who suffer from this.) That is, finding
things that have “worked” in the past just because you looked at a lot of
things and found some random coincidental effect.
The best cure for data mining is an out-of-sample test (i.e., trying
it somewhere else or over a different time period).
Well, there’s a paradox to this industry. To attract traditional assets, you need a
five-year track record, some gray hair on your team, and it does not hurt
to have the solidity of a Goldman Sachs behind you. To attract hedge
fund assets you only need to be 29 years old and say that you are closing. Of course I’m being too flip. For starting either one a great track record,
pedigree, and investment process are crucial. However, the strange difference
in difficulty between starting a hedge fund and a traditional firm is
quite real. So we launched