“Prediction markets can test a candidate's true impact if you design them like a natural experiment”
Proposes using quasi-experimental methods from econometrics — regression discontinuity design and difference-in-differences — to turn prediction markets into causal tools for evaluating candidates and policies. Instead of futarchy's wholesale replacement of governance with markets, this approach uses conditional markets on vote margins and election-day weather to isolate a candidate's true impact from confounding factors.
Some technical background helpful