Is There +EV on Prediction Markets?
What to actually buy, based on 72 million trades
Short answer: yes — but the devil is in the details, as always.
I have two accounts on prediction markets. The first one is purely degenerate. I occasionally throw money at sports, esports, and politics just for fun. There’s probably no real edge there, which is why I do it rarely.
The second account is different. I'm running a hypothesis there: small bets first, scale up if it works. A bot handles all the trades, exclusively on sports events. The logic is dead simple — so simple I suspect a lot of sports bettors are doing the same thing. I'll write about it properly once the experiment reaches a conclusion. For now, here are the interim numbers: nearly 3,000 predictions, ~1% ROI.
I also had another hunch: bets with 90%+ implied probability should carry a small structural edge, because people chasing big multipliers are on the other side. Turns out this hunch has evidence behind it.
Spoiler: the gap between the best and worst position is 64 percentage points. More on that below.
