Over the past week or so, a sizable gap has opened up between prediction markets and statistical models referencing the outcome of the November election.
The argument for multiple markets with different rules as a way to help triangulate information inputs seems amenable to formalization (e.g. improving model identifiability in some sort of hierarchical Bayesian model with market-specific likelihood functions; perhaps even using data at the transaction level and per-trader utility function type as a latent variable, that sort of thing). Is there prior and/or ongoing work in this direction?
The argument for multiple markets with different rules as a way to help triangulate information inputs seems amenable to formalization (e.g. improving model identifiability in some sort of hierarchical Bayesian model with market-specific likelihood functions; perhaps even using data at the transaction level and per-trader utility function type as a latent variable, that sort of thing). Is there prior and/or ongoing work in this direction?
No work that I know of, but the idea is interesting
Love the police work.
The sleuthing is mostly Domer (at the link), I added some financial analysis, but thanks!