Candidate A and B will face off. There is a guaranteed winner. The market maker takes a five percent cut. For initial simplicity, 1 dollar will be bet by market participants. The market maker's equations are really simple.

P(A) + P(B) = 1 (the probability of A plus the probability of B is the total market funding)

The market maker knows he needs a 5% cut, so the payout is .95 to the correct predictors. Easy! 5% is nothing, right? We can ignore it and do. Who cares if markets are slightly off? They always are.

But what about taxes? The predictor's profits are taxed at 25% (let's pretend taxes are lower).

The market does not add up now. The market exists, but it seems irrational. The people betting on A and the people betting on B have to have radically different views on the election (information sets).

These markets exist now and are filled with relatively smart traders. These markets continuously fund agreements where expected value is not agreed upon, where the traders of each outcome possibility have to have radically different information sets.

Now consider a different type of prediction market: https://bit.ly/2DnOATR

This problem, let's call it Outcome Payout Disequilibrium, is easier to study in these types of prediction markets.

Any market with persistent and severe Outcome Payout Disequilibrium does not fit any version of the efficient market hypothesis.

Because...

Consider a Yes/No PIP market on a stock's price--will the stock's price be higher tomorrow or not?--and the stock's price itself. Arbitrage is quickly eaten up. But if there is severe and persistent Outcome Payout Disequilibrium in the Yes/No PIP market, then we know the stock's price is being determine by discordant information sets.

-roosterred