What Are Automated Market Makers in DeFi?

Automated market makers are the algorithmic systems that let decentralized exchanges function without a human or firm actively quoting buy and sell prices — instead, a fixed mathematical formula, applied to a pool of deposited tokens, sets the price automatically for every trade. That much is fairly well known. What’s less understood is a structural weakness baked directly into how AMMs work: because they can only react to trades after the fact rather than continuously updating like a human market maker would, they systematically lose money to arbitrageurs, a cost that academic researchers have recently given a precise name — loss-versus-rebalancing.

Here’s the finding that should reshape how anyone thinks about “just providing liquidity” as passive income: researchers studying the Uniswap V2 WETH-USDC pool found that 99.991% of an unhedged liquidity provider’s return variance simply reflects holding exposure to ETH’s price, not anything specific to market making at all. The remaining sliver — fees earned minus losses to arbitrageurs — is where the real strategic risk and reward actually lives, and that sliver has a name: loss-versus-rebalancing, a systematic, predictable cost that AMMs pay to arbitrageurs every single time the market price moves and the pool hasn’t caught up yet.

What Makes a Market Maker “Automated”

Before getting into the structural weakness, it’s worth being precise about what an AMM actually replaces.

• A traditional market maker is a human trader or firm that continuously quotes both a buy and sell price, updating those quotes based on new information as it arrives
• An automated market maker replaces that human decision-making with a fixed algorithm — most commonly the constant product formula — applied to a pool of deposited token reserves
• The AMM’s price is purely mechanical, determined by the ratio between the two token reserves at any given moment
• This removes the need for a human counterparty or matched order for every trade, enabling continuous, permissionless, 24/7 trading

The Structural Weakness: AMMs Can’t React, Only Get Reacted To

This is the core insight separating AMMs from genuinely comparable human market making.

• A human market maker watches news, order flow, and related markets, adjusting quotes proactively the instant new information suggests the “true” price has shifted
• An AMM has no awareness of anything happening outside its own pool — it simply holds whatever price its reserve ratio implies until someone actually trades against it
• This means an AMM’s displayed price can lag the real, broader market price for a meaningful window of time, especially right after a significant price move elsewhere
• During that lag, the AMM is effectively offering a stale, mispriced quote to anyone paying attention

How Arbitrageurs Keep AMM Prices in Line With the Real Market

This lag isn’t a rare edge case — it’s a structural, recurring feature that a specific class of traders exists specifically to exploit.

• When the broader market price moves and an AMM’s internal price hasn’t caught up, an arbitrageur can trade against the pool at the stale price and immediately profit
• That arbitrage trade is precisely what pulls the AMM’s price back into alignment with the wider market — it’s a genuinely useful function for the ecosystem’s overall price accuracy
• But the profit captured by the arbitrageur comes directly out of the liquidity providers’ pockets, since the pool systematically executed that trade at a worse price than the “true” market rate
• This transfer of value from LPs to arbitrageurs happens continuously, every time prices move meaningfully, not just occasionally during extreme volatility

LVR vs Impermanent Loss: Two Ways of Measuring the Same Underlying Cost

These two concepts get discussed somewhat interchangeably, but they frame the same underlying phenomenon differently.

• Impermanent loss compares a liquidity position’s value at one point in time against simply holding the two tokens separately since deposit — a comparison that depends heavily on an arbitrary starting point
• Loss-versus-rebalancing instead compares the AMM’s actual trades against a hypothetical strategy that continuously rebalances at true market prices, isolating specifically the cost caused by stale-price arbitrage
• Recent research has found that for smaller price movements, both measures produce nearly identical expected values, despite framing the cost quite differently
• LVR is described by researchers as predictable and non-decreasing — meaning it accumulates steadily as a genuine, structural cost of the AMM design, rather than simply reflecting bad luck from volatility

What Reduces LVR in Practice

Since this cost is structural rather than incidental, protocol designers have specifically worked on ways to reduce it.

• Faster blockchain block times shrink the window during which an AMM’s price can remain stale before an arbitrageur corrects it, directly reducing the size of each individual arbitrage opportunity
• Trading fees partially offset LVR, since fee income accrues to LPs regardless of whether a given trade came from a genuine trader or an arbitrageur
• Concentrated liquidity designs, like Uniswap V3, can reduce the effective capital exposed to LVR within a chosen price range, though this comes with its own trade-offs around range management
• Newer experimental AMM designs, such as “partially active” market makers that hold back a portion of reserves from active trading, aim to structurally reduce adverse selection costs while accepting other trade-offs in exchange

Why AMMs Still Won Despite This Structural Cost

Given this genuine, quantifiable drawback, it’s worth understanding why AMMs became the dominant model for decentralized trading anyway.

• AMMs are computationally simple and cheap to run on-chain, compared to the complexity of replicating a full order book system in smart contract code
• They enable continuous, permissionless liquidity provision without requiring a licensed market-making firm or specialized infrastructure
• Trading fee income, especially on high-volume pools, can meaningfully offset LVR costs for liquidity providers, particularly during periods of lower volatility
• For many LPs, the strategy functions less as pure “market making” and more as a leveraged, fee-generating exposure to holding the underlying assets — which is precisely what that earlier statistic about return variance illustrates

This is general educational information about AMM mechanics and current academic research on liquidity provision costs, not investment advice. Providing liquidity to AMMs carries genuine, quantifiable structural costs beyond simple price volatility, and independent research is worth doing before depositing funds into any pool.

FAQs

Q1. Is loss-versus-rebalancing the same thing as impermanent loss?

They’re closely related but framed differently — LVR specifically isolates the cost from arbitrageurs trading against stale AMM prices, while impermanent loss compares a position against simply holding the tokens since an arbitrary starting point.

Q2. Why can’t an automated market maker just update its prices as fast as a human trader would?

An AMM has no awareness of external market events — it only reprices in response to an actual trade against its reserves, meaning it structurally lags the broader market until an arbitrageur corrects it.