I'm sure I've seen Peter write the following somewhere (along similar lines, although I might have made it up ):
"People have found a pattern that yields consistent returns, but they cannot do the background work to prove it."
I'd put myself in that category. My experience means mostly I've "stumbled" on certain opportunities, but it's pretty much all "instinct" (whereby my journal is based on videos, not data). This is acceptable to a point (i.e. fillrates), but it lacks finesse & ultimately "true" scalability (i.e. does my pattern work equally well on this race? on this runner? etc etc). Maybe I'm a rare breed who wants to prove their edges in cold hard data. But I think all top top traders, have a fundamental base in statistics & data handling - therefore can prove why what they do works - which can be taught.