quantia

Prediction markets as usually presented seem quite problematic because the mechanism that gives you “good” prices seems to rely entirely on adverse selection.

The only people who have a good reason to trade either have informational edge (in which case, why would anyone be a counterparty?) or are satisfying a hedging requirement (in which case, they’re shifting the price away from the true probability).

With most “normal” financial markets, hedging is just expected behaviour, and liquidity providers + unsophisticated speculators are able to make money without always getting ripped off by more knowledgeable investors. There’s no expectation that prices are solely a reflection of the probabilities of various future outcomes, because the market isn’t initially created for speculation, but rather for raising capital, swapping risks, etc.

The theoretical picture of prediction markets is essentially presenting participants with a negative-sum game (due to risk-averse utility functions, transaction costs, and the costs of staying as informed as everybody else), and expecting this to induce them to reveal information. This sounds a little like expecting a free lunch.

Subsidising the market might fix things, but it’s tricky to find a payoff structure that doesn’t make trading more attractive than it ought to be and just add noise to market prices. It also doesn’t address the problem of people using the prediction market as a hedging tool, since a market priced according to true probabilities is nicer to trade in than a market where most participants are trading in the same direction as you because they have identical risks to hedge.

A one-sided prediction market is a great way to demonstrate confidence in a set of forecasts, but if people trust the forecasts to be right it doesn’t seem like very many of them would actually bet without having some kind of insider information (unless they just want to see whether you’re telling the truth about your willingness to back up your forecasts with money).

In general, the perception that financial markets exist to serve some sort of informational need is somewhat naive. Markets exist to buy and sell things; price signals serve only to encourage investment in producing more of a thing, and not primarily as a substitute for news, though they may be cautiously used this way.