>I don't really see how that cashes out in an inaccurate overall system
Imagine a strategy that has a 50% chance of blowing you up completely over 10 years, but nets you 50% more profit if it works. The average return of this strategy is 25% less than a safe strategy (zero risk, zero gain), but it has a 50% chance of winning in this marketplace. If it were real money, nobody would go near it.
Ok, now imagine there's 100 people running strategies with that profile and 100 running the safe strategy. All of the top spots will be held by the suicidal strategy. This is an artificial example but you get the point. Real money forces you to avoid that risk profile as soon as you identify it.