Suppose the fund gets 2% of all profits in a profitable year and gets nothing in an unprofitable year.
Scenario 1: 100 million dollar profits for two years in a row. Fund gets 4 million, customer gets 196 million.
Scenario 2: 300 million dollar profit in the first year, 300 million dollar loss in the second year. Fund gets 6 million, customer ends up with -6 million.
On a large scale, what this means is that the fund can look for large ups and downs because they gain every time there's an up, but they don't lose when there's a down.
Now if you parallelize this (so it's happening at the same time with many different customers), it means they can focus on risky bets. Maybe half their customers will win and half will lose, but the fund makes money on the winners and isn't hurt by the losers.
Most of the arguments in this comments sections are outdated. The same goes to the 'Buffet argument'. Most of the funds (after 2008) run with High-Watermark rule. Thus, in the case of your Scenario 2, fund will get nothing from the next 300 million it makes (apart from fixed fees). Thus, the investor will end up with 294 million at period 3. Just like if the fund made 100 million for 3 periods consequently.
Scenario 1: 100 million dollar profits for two years in a row. Fund gets 4 million, customer gets 196 million.
Scenario 2: 300 million dollar profit in the first year, 300 million dollar loss in the second year. Fund gets 6 million, customer ends up with -6 million.