How to pick a manager

Perhaps it is actually harder to pick a manager than to pick investments. The general statistics against active management are well-known. In 2012, for example, 63% of large cap mutual funds failed to match the market (even worse less for hedge funds).

As a result, the full choice set for selecting investments includes both asset class as well as manager, and the probability of choosing a good asset class And a good manager is less than a good asset class itself. Given that most investors usually time managers incorrectly, are there some nontraditional metrics which would work?

Tudor's infamous divorce heuristic notwithstanding , what about some more specific methods to pick a manager? How about through the lens of incentives. A manager should be incentivized for long term outperformance.

1) The manager should still be "hungry." If you are already a household name and have "made it" as a big time manager, you cannot have the same drive for performance as before. A key barrier could be $1B AUM, but the idea should be clear.

2) The manager should be focused primarily on investing. If he has a such a large staff that he spends most of the time either buying baseball teams or opening a huge new office which makes headlines.

3) The manager should not be on CNBC etc. enough to be a talking head. If the manager has enough time and/or interest to continually market himself, he/she is too focused on raising money. Raising money/awareness of his/her ideas does not really make performance better.

What other ones could work?