Understanding CPMM: How Automated Market Makers Price Prediction Contracts

A constant product market maker, often abbreviated CPMM, is a mechanism for automatically pricing trades using a mathematical formula instead of matching individual buy and sell orders in a traditional order book. The core idea, borrowed from decentralized finance and adapted for prediction markets, is that a liquidity pool holds reserves of two related assets, in this case shares representing the yes and no sides of an outcome, and the product of those two reserve quantities is held constant according to the formula x times y equals k. Every trade shifts the ratio of reserves in the pool, and that shifted ratio is what determines the new price.

To make this concrete with an illustrative example, imagine a pool starts with equal reserves of yes-shares and no-shares, implying a fifty percent probability. A trader who wants to buy yes-shares deposits money into the pool and withdraws yes-shares, which reduces the pool's yes-share reserve relative to its no-share reserve. Because the formula requires the product of the two reserves to stay constant, removing yes-shares causes the implied price of a yes-share to rise. The larger the trade relative to the size of the pool, the more the price moves, which is the mechanism's built-in way of representing slippage.

This design has a useful property: a CPMM can always quote a price and execute a trade, even for a contract with very little natural interest, because the formula generates a price mechanically from the pool's reserves rather than requiring a matching counterparty to show up at the same moment. That is valuable for markets on niche or early-stage questions where an order book might otherwise sit empty. The tradeoff is that prices in a CPMM can move more than they would on a deep order book for the same trade size, particularly when the pool itself is small.

Liquidity providers are the other side of this system. Anyone can typically deposit funds into a pool to become a liquidity provider, earning a share of trading fees generated by others' trades in exchange for taking on the risk that the pool's reserves end up skewed toward whichever outcome resolves against them. This is conceptually similar to providing liquidity in a decentralized finance exchange: the provider is compensated for enabling trading, but is exposed to a version of the risk known in other AMM contexts as impermanent loss, here manifesting as the pool's payout composition shifting unfavorably as the true probability becomes clearer.

It is worth noting that the CPMM approach represents one design choice among several for pricing prediction market contracts, and it is not the only, or even necessarily the dominant, model in use today. Many prediction markets, including some of the largest and most liquid ones, instead rely primarily on a central limit order book, where buyers and sellers post prices and get matched directly, much like a conventional exchange. Some platforms have also used variations such as logarithmic market scoring rules, which behave somewhat differently from a strict constant product formula while serving a similar purpose of always providing a price.

Understanding the CPMM formula is useful less because any one platform's exact current mechanism depends on it, and more because it illustrates a general principle that applies across automated market making: price is a function of relative scarcity within a pool, trade size affects price nonlinearly, and someone has to bear the risk of holding inventory in exchange for a fee. Whether a specific prediction market runs on a pure CPMM, an order book, or a hybrid of the two, that underlying tension between liquidity provision, price impact, and fair compensation for risk is the same one every market design has to solve.