Performance fees are based on an illusion. Avoid funds that pay them.

How do they usually work?
Typical performance fees will take a slug of a fund's performance above a certain benchmark. For example, a very active fund benchmarked against the FTSE350 index might take 20% of any excess return.

This seems fair doesn't it? 20% seems a lot, but after all we are talking about excess performance here. That still leaves 80% for us. What can possibly go wrong?

Here's the catch
The catch is what happens in the down years. The fund manager is not offering to repay us 20% of any underperformance. So he's taking a slice of the gains, but he's leaving us with all the losses

Well, you might say, he has to take some sort of payment for his services. You'd rather it was in this form, related to the performance of the fund, than as a flat percentage.

Wrong. The 'performance' fee is related to volatility, not total return. Look at these examples.

If your fund mixes +10% years with -10% years (measured against the benchmark - not a bad peformance, actually) the manager will take 20% of 10% every other year - average 1% annually. But if your fund mixes +40% years with - 50% years the manager will take 20% of 40% every other year - average 4% annually. Even though the fund itself will be going down the pan at 10/2 = 5% a year gross + 4% fee = 9% total. And your money with it.

It's actually worse than this. We shortcut the maths above. See Percentage Games if you are interested.

....and the consequences are....
What investment style would you expect the manager to adopt under this particular 'performance' incentive? If he has a choice between a strategy giving a safe 8% annual return and another giving + or - 50%, which one do you think he's going to chose?

Why do you think hedge funds are so keen on this type of fee? How much money do you think the technology funds made in performance fees on the way up before the bubble burst? Actually, you don't really want to know.

...to be fair to some funds
Some funds have performance fees where, after a payout, no further fees are paid until the fund returns to the previous trigger level (or 'high water mark'). This is much, much better. But the first year is still a one way bet. And each subsequent year still has a shot at doing better (think of a child on a swing). And if a fund falls too far below its high water mark after a few years it can always change its fee formula. Or merge into another fund. People have short memories.