Grow the player base, increase the number of predictions, you'll end up with some real super-stars who just go on winning. They'll probably write books and lecture about their techniques. Right up until the moment they lose. Kind of like hedge fund managers.
I read this example in a book or article, but I can't remember what that was called.
With 1,000,000 people predicting 100 coinflips each, the top ten had an accuracy of:
75% (1)
74% (1)
73% (1)
72% (7)
He sent different predictions to different people, some saw his predictions come true, and he sent more letters to all those people. And then repeated that, eventually narrowing it down to one person, who saw week after week correct gambling predictions, and really believed he had a system for winning.
She borrowed family money to bet big - then he revealed that she was the result of survivor bias and he had no system.
The Romans considered luck to be an important trait. When regarding military leaders, they considered luck to be even more important than skill. It meant that the gods favored the lucky person.
Julius Caesar was a skillful general, but he was even more admired for his incredible luck. There were many times where he was almost beaten, yet he got some lucky break that allowed him to be victorious.
It's harder to model with VC's because they are measured in rewards that get exponentially better with success (money) and are self-reinforcing (better deals from recognition). My guess is that they figured out how to eliminate the absolutely horrible ones and got lucky enough times to have the best deals come to them.
That does not describe Berkshire Hathaway at all.