Blockchains are a new settlement and ownership layer, one that’s programmable, open, and global by default, unlocking new forms of entrepreneurship, creativity, and infrastructure. Growth in monthly active crypto addresses is generally tracking the
Prediction markets are a hot topic again — even cartoon characters are talking about them (cf South Park). But beyond the buzz, what is a prediction market, exactly? How do they work, how are they designed, and what makes them work?
We answer all
In this explainer, Tim Roughgarden, head of research at a16z crypto, presents a new way to think about the cost side of providing liquidity in an AMM, which centers around a quantity called LVR (“loss versus rebalancing,” pronounced “lever”).
There are two types of participants in an automated market maker (AMM): traders, who exchange one of the AMM’s tokens for another (say, ETH and USDC), and liquidity providers (LPs), who provide tokens to the AMM in the first place, generally in exchange for a share of the trading fees.
But to date it’s been unclear when it makes economic sense to participate as an LP. When does the benefit exceed the cost? The benefit side of this comparison is easy to understand: revenue from shared trading fees, plus in some cases additional token rewards.
LVR is a mathematical formula that isolates an LP’s adverse selection costs, allowing them to compare their potential LVR with the AMM’s fees, and to reason about whether and when to provide liquidity. And it provides AMMs with a way to keep their LPs happy by keeping LVR small.
This is joint work with Jason Milionis, Ciamac Moallemi, and Anthony Lee Zhang
About the presenter Tim Roughgarden is a Professor of Computer Science and a member of the Data Science Institute at Columbia University, and Head of Research at a16z crypto.