Either really good math or really good insider trading.
He's a MIT/Berkley Math professor.
The Simons Center for Geometry and Physics has an art gallery and an associated lecture series, and James often shows up to participate in these events. Luckily, these events are free to enter for all graduate students. At one of these events, I'd commented that a particular piece reminded me of a crystalline structure I was interested in studying (as a li-ion cathode). James overheard me and came over to quiz me on some of my group's work. About 8 months later, I ran into him at another lecture. While he didn't remember my name, he asked how my hollandite simulations had turned out.
This anecdote doesn't really prove anything, other than that Dr Simons is a pretty cool dude.
In real life you can't run a fund by finding insiders willing to commit crimes routinely. But you often can get the same information by other means by collecting intelligence. Observing traffic, the flow of materials, following people and finding who meets who.
IANAL, but I’d imagine that then it’s no longer non-public information (at least in the US).