Impermanent Loss Explained: When It Actually Matters and When It Doesn't

Most explainers either drown the topic in formulas or skip the mechanism entirely. This one walks through what actually happens to your tokens, when the loss matters, and when it stays marginal.

Impermanent loss is the difference between what your tokens are worth after providing liquidity to a decentralized exchange and what they would have been worth if you had simply held them. If you have ever looked at a liquidity pool and quietly assumed that providing liquidity is just a way to earn passive yield, this article fills in what most explainers leave out. At Blockready, we teach this concept inside Module 11, and it is one of the most consistently misunderstood ideas in DeFi, partly because the name is misleading and partly because the math gets in the way of the mechanism.

By the end of this piece, you should be able to look at any liquidity pool and form a defensible opinion on whether it is worth your capital.

What is impermanent loss?

The word "impermanent" is the source of much of the confusion. It suggests something that goes away on its own. The reality is conditional: the loss disappears only if the price ratio between your two tokens returns to where it was when you deposited. If you withdraw before that happens, the loss becomes permanent. A name like "divergence loss" would describe the phenomenon more accurately.

You also do not lose tokens. The pool always pays you out. What you lose is the difference between the value of your withdrawal and the value you would have had by holding. That sounds reassuring until you realize that an opportunity cost large enough to dwarf your fees is, for practical purposes, indistinguishable from a loss.

How impermanent loss actually happens

The constant product formula in plain English

Most automated market makers, including the kind that power decentralized exchanges, use a pricing rule called the constant product formula: x times y equals k. The x is the quantity of one token in the pool, the y is the quantity of the other, and k is a fixed constant. Whenever someone trades against the pool, the underlying smart contract adjusts the quantities so the product still equals k. The pool does not have an opinion about prices. It just preserves the product.

The arbitrageur's role

Here is what most articles skip. The pool's internal price does not respond to external markets on its own. It responds to traders. When the external price of one token rises and the pool's internal price has not caught up, an arbitrageur sees free money. They buy the cheaper token from the pool and sell it elsewhere at the higher external price, repeating until the prices match. That is not a flaw in the design. It is the design.

For you, the liquidity provider, this rebalancing has a specific consequence. Every arbitrage trade pulls the appreciating token out of the pool and adds more of the depreciating one. By the time the prices line up again, your share of the pool contains less of the asset that went up and more of the asset that went down. The arbitrageur kept the price difference. You absorbed it.

A worked example using round numbers

You deposit 1 ETH and 1,000 USDC into a 50/50 pool. ETH costs $1,000 at deposit, so your contribution is worth $2,000. The pool already holds 9 ETH and 9,000 USDC from other LPs, so after your deposit it holds 10 ETH and 10,000 USDC. You own 10% of the pool. The constant product is 10 multiplied by 10,000, which equals 100,000.

Now ETH's price on every other exchange doubles to $2,000. The pool itself has not changed; inside it, ETH still trades at $1,000. That gap creates an arbitrage opportunity. Arbitrageurs buy ETH from the pool at $1,000 and sell it elsewhere for $2,000. Each trade reduces the pool's ETH and increases its USDC, until the internal price catches up.

When the internal price reaches $2,000, the pool contains roughly 7.071 ETH and 14,142 USDC. (Multiply those: still 100,000.) Your 10% share is now 0.7071 ETH and 1,414.2 USDC.

Time to compare. If you withdraw, you get tokens worth $2,828.40 (0.7071 ETH at $2,000, plus 1,414.2 USDC at $1). Not bad. But if you had simply held your original 1 ETH and 1,000 USDC, you would have $3,000. The gap is $171.60, or 5.7% of what you would have had by holding. That gap is your impermanent loss. The pool was not broken and the smart contract did exactly what it was supposed to do. Yet you ended up worse off than holding, because every arbitrage trade rebalanced your position toward the asset that was losing relative value.

How impermanent loss scales with price moves

The relationship between price divergence and impermanent loss is not linear. A 2x move and a 4x move do not produce a 2x and 4x loss. They produce something steeper.

Two things matter. Direction does not. ETH doubling against USDC produces the same loss as ETH halving against USDC; the pool sees magnitude, not sign. And the curve is gentle near the deposit price and steep at the extremes.

When does impermanent loss actually matter?

The math is easy to find. The judgment about magnitude is harder, and understanding the mechanism without it leaves you nervous, uncertain, and unable to decide.

When it stays marginal

Impermanent loss is small in three conditions: highly correlated pairs (two stablecoins, or wrapped versions of the same asset), short holding periods in quiet markets, and high-fee pools with sustained trading volume where fees outpace the divergence drip. A USDC and DAI pool, for example, will see almost no impermanent loss in normal conditions because both tokens track the dollar.

When it dominates

The loss dominates in the opposite scenarios. Volatile pairs, where one asset can swing 50% in a week, push you into the steep part of the curve quickly. Concentrated liquidity on Uniswap v3 amplifies both upside fees and downside divergence. And single-token collapses, like a memecoin going to zero or a stablecoin breaking its peg, can wipe out the appreciating side of the pool entirely.

What the data shows: the 49.5% finding

The most important piece of evidence on this question comes from a 2021 study by Topaze Blue and Bancor, which analyzed 17,000 wallets across 17 Uniswap v3 pools accounting for 43% of the platform's total value locked at the time. The full academic paper is available on arXiv. During the five-month sample period, those pools generated $199 million in trading fees but incurred over $260 million in impermanent loss. The result: 49.5% of LPs, after fees, would have been better off simply holding. The IntoTheBlock follow-up, looking wallet-by-wallet, found 51.75% of wallets unprofitable against a hold strategy.

The study's authors concluded, in findings widely reported in the DeFi press, that "both inexperienced retail users and sophisticated professionals struggle to turn a profit under this model." Concentrated liquidity provision is closer to professional market-making than passive yield, and the data has been clear about that since 2021. Understanding impermanent loss is the difference between depositing capital into a pool that quietly extracts value and one where fees genuinely compensate for the risk.

A real impermanent loss case study: USDC's March 2023 depeg

Stablecoin pools are treated as the safest place to provide liquidity. The conventional wisdom holds that two assets pegged to the same dollar will not diverge. That is correct in normal conditions. It is also exactly the assumption that broke on March 11, 2023.

That weekend, Silicon Valley Bank failed. Circle, the issuer of USDC, disclosed that $3.3 billion of its $40 billion in reserves was held at SVB, roughly 8% of the cash backing the stablecoin. Within hours, USDC traded as low as $0.87. DAI, USDP, and FRAX, which all hold portions of their reserves in USDC, depegged in sympathy. A Federal Reserve research note later documented how the depeg cascaded through DeFi.

The peg recovered within 72 hours after federal regulators announced full backing for SVB depositors. But for LPs who panicked and withdrew during the depeg window, the loss was very much permanent. Like other costly mistakes in crypto, this one came down to acting on assumptions instead of understanding the mechanism.

How to reduce impermanent loss exposure

You cannot eliminate impermanent loss while still earning trading fees. You can reduce it through three levers, each with its own tradeoff.

The cleanest reduction comes from pair selection. Stable-stable pairs minimize divergence in normal conditions but expose you to depeg events. Correlated pairs like stETH and ETH stay tight as long as the correlation holds. Volatile pairs offer the highest fees, but the data is clear that fees rarely compensate for divergence at scale.

Two other levers help. Match the pool's fee tier to the volatility you expect: a 1% tier suits volatile pairs, a 0.05% tier suits stables. And avoid concentrated liquidity unless you are prepared to actively manage the position. The half of v3 LPs who lose money are mostly the ones who set a range and walked away.

When holding beats providing liquidity

This is the question most posts on this topic carefully avoid. There are conditions in which holding the two tokens is the rational choice, and naming them is the only way to make this conversation useful.

Holding beats LP'ing in three concrete situations: when you have a directional view on either token (holding ETH directly captures upside without rebalancing drag), when expected fee income is unclear or low (new pools with thin volume rarely generate enough fees to cover meaningful divergence), and when you cannot actively monitor the position. The data on inactive LPs is bleak, and most retail liquidity providers fall into this category by default.

The formula, for reference

For readers who want it: the standard impermanent loss formula for a 50/50 constant product pool is IL = (2 times the square root of r) divided by (1 plus r), minus 1, where r is the new price ratio divided by the old. Plug in r = 2 (a doubling) and you get 5.7%. Plug in r = 5 and you get 25.5%. Useful for spreadsheet modeling, but you do not need it to understand the mechanism, which is why we put it last.

Frequently Asked Questions