Liquidity and loss for AMMs

Generally, automated market makers (AMMs) have to compensate for the “impermanent loss” that liquidity providers (LPs) suffer from when prices and asset compositions of their pool shares change. While this is not an issue for e.g. stableswap markets, in many other cases projects distribute inflationary token incentives to the LPs on top of trading fees, which is not sustainable.

Proper (unbiased) prediction markets are particularly tricky to run because they must guarantee defined payouts for all correct bets. Only under such promise can the market obtain the information we are after: the probabilities for either outcome to happen.

For illustration, betting on “head” of a fair coin flip is only profitable if the payout is larger than 2x the investment, and, conversely, from traders taking such bets for at least 2x returns on both sides, “head” and “tail”, we can conclude that the chance of either outcome is 50% (according to the market’s belief).

In contrast to guaranteed payouts at the end, common DEX markets for trading assets put the responsibility of “selling the top” on the traders for maximal profit, which makes them easier to run.

In the case of AMM prediction markets, payouts for winners must come from the liquidity providers, who tend to lose against traders actively taking the winning position (which can actually be easy when important information comes in or shortly before the maturity of the market). In other words, the impermanent loss caused by the extreme price changes that are enforced at maturity is very large. Thus, despite the fact that AMMs have been invented in times before blockchains to enable prediction markets with low trading activity, they turn out to be ill-suited for this purpose: either incentives are unsustainable or liquidity provision is irrational.