Sign in
Article On-chain

Impermanent loss explained: liquidity pools and LP returns

Impermanent loss explained with a 50/50-pool formula, ETH examples, fees and concentrated-liquidity risks. How LP returns differ from holding tokens.

Impermanent loss explained: liquidity pools and LP returns
Иллюстрация CoinLab / AI
Contents
  1. How a pool changes your asset mix
  2. The formula for a full-range 50/50 pool
  3. A positive dollar return can still trail holding
  4. Can trading fees offset the gap?
  5. Why a narrow range changes the calculation
  6. How to read a specific position’s results
  7. Limitations and common mistakes
  8. Frequently asked questions

Impermanent loss is the gap between the value of a liquidity-provider position and the value of the same original tokens if they had simply remained in a wallet. It arises because an automated market maker changes the position’s asset mix as relative prices move. A position can earn a positive dollar return while still underperforming that unchanged portfolio. Profit since deposit and performance against holding are different measurements.

A liquidity provider, or LP, supplies assets for swaps and receives the associated share of trading fees. In exchange, the provider accepts changing inventory and other protocol risks. Neither a pool’s advertised fee nor an attractive annualized return describes the complete outcome. This guide explains the mechanism, the arithmetic and the limits of the model, without selecting a particular position for the reader.

How a pool changes your asset mix

An order-book seller specifies a price and quantity. An automated market maker executes trades against reserves under rules encoded in its contracts. The classic constant-product model uses x × y = k, where x and y are the quantities of two tokens. Ignoring fees for now, their product stays constant. Uniswap’s v2 returns documentation explains this mechanism and its implications for providers.

When one token appreciates relative to the other, arbitrage trades change the reserves: traders remove the more expensive token and add the other. The LP’s claim consequently contains less of the appreciating asset and more of its counterpart. A move in the opposite direction changes the composition the other way. This is an outcome of the pool’s rule, rather than a discretionary manager deciding to sell a rally.

Comparing the ending balance only with the original dollar deposit therefore leaves something out. There are at least two useful benchmarks: the assets’ initial value and what the unchanged original token quantities would be worth at the end. Impermanent loss concerns the second comparison. Both tokens can also fall in dollar terms without a large change in their relative price.

The formula for a full-range 50/50 pool

For a constant-product, full-range position initially funded with equal values of two assets, the standard expression is:

IL = 2 × √r / (1 + r) − 1

Here r is the ending relative price divided by the starting relative price. If the price of the first token in units of the second doubles, r is 2. The formula measures underperformance against holding before fees and additional costs. It is not a universal calculator for every DeFi position.

Relative price movement r Underperformance against holding
Down 75% 0.25 −20.00%
Down 50% 0.50 −5.72%
Down 20% 0.80 −0.62%
Unchanged 1.00 0.00%
Up 25% 1.25 −0.62%
Doubles 2.00 −5.72%
Quadruples 4.00 −20.00%

CoinLab calculated this table from the documented formula. Its symmetry applies to reciprocal ratios: doubling and halving produce the same relative underperformance. A 50% rise and a 50% fall are not reciprocal changes.

The assumptions matter: constant-product pricing, an initial 50/50 value split, the full price range, no fees, no strategy changes and no additional deposits. Weighted pools, stable-swap curves and bounded concentrated-liquidity positions require their own appropriate calculations.

A positive dollar return can still trail holding

Consider entirely hypothetical starting conditions: one ETH priced at $2,000 plus 2,000 stablecoin units, with the stablecoin assumed to remain worth $1 throughout this model. The initial value is $4,000. These are teaching inputs, not current ETH prices, actual pool reserves or a promise that a stablecoin will maintain its peg.

Suppose ETH rises to $2,500. Without fees, the model produces approximately 0.894427 ETH and 2,236.07 stablecoin units, worth $4,472.14 in total. Holding the original one ETH and 2,000 stablecoins would instead be worth $4,500. The gap is $27.86, or approximately 0.62% of the holding benchmark.

The provider nevertheless has a positive dollar result of about $472.14 before costs. Impermanent loss does not necessarily mean losing money against the initial $4,000. It describes the different result of using the same starting assets in a liquidity position rather than leaving their quantities unchanged.

If ETH halves instead, the model position is worth approximately $2,828.43 against $3,000 for holding. Underperformance is about 5.72%, while the decline against the original $4,000 is about 29.29%. Those percentages answer different questions and should not be substituted for one another.

The real-market context is separate from this teaching model: CoinLab’s November 18, 2026 server snapshot at 1:00 PM ET put aggregate stablecoin supply at approximately $288.07B (DefiLlama). This is a segment-wide total, not a particular pool’s reserves or a guarantee that every token maintains its peg. Our stablecoin-supply explainer discusses why aggregate market size does not protect an individual position. Updated conditions are available in the CoinLab daily brief.

Can trading fees offset the gap?

They can, but there is no guarantee. Fees depend on actual swap volume, the fee setting, the provider’s share of active liquidity and the time spent participating. A pool’s total trading volume is not your fee income, and a transaction fee percentage is not an annual return.

Uniswap’s fee documentation also distinguishes protocol versions. In v2, fees increase reserves; in v3 and v4 they accrue separately to positions. Any portion allocated to the protocol cannot simultaneously be counted as the LP’s income. Current parameters must be checked for the particular pool.

In the teaching example, suppose $4,472.14 excludes separately accrued fees, another $30 of fees is earned and costs are $10. The result becomes $4,492.14, still below the $4,500 holding benchmark. These are hypothetical figures. In actual accounting, do not add fees a second time if they are already included in reserves or a position valuation.

APR and APY also need a measurement period and an explanation of what counts as income. A short successful interval does not guarantee the same rate over a year. Incentives paid in another token need separate treatment: token units may increase while their dollar value declines. LP returns also differ from crypto staking, which has another reward mechanism and risk profile.

Why a narrow range changes the calculation

With concentrated liquidity, an LP chooses a price interval rather than distributing capital across the full curve. This can increase active liquidity per dollar of committed capital, but changes the position’s exposure. The details are described in the concentrated-liquidity documentation.

Once price leaves the range, the principal becomes single-sided and stops earning swap fees until price returns inside. Going out of range is not itself a liquidation or proof that ownership has been lost. However, the token mix no longer matches the original allocation.

Moving the range is another decision. It may involve accepting the current inventory, swapping some assets and paying transaction costs. Repeated repositioning does not magically erase economic underperformance. An actively managed strategy needs an explicit benchmark and accounting for all of its cash flows.

A single-sided position is not automatically equivalent to an ordinary stop order either. The range-order explanation shows that conversion can reverse if price crosses back through the interval. The simple full-range formula above cannot, by itself, evaluate a narrow-range strategy.

How to read a specific position’s results

  1. Record the original quantities of both tokens, the entry time and the accounting currency.
  2. Identify the protocol version, pool type, range and rules for changing the position.
  3. Value the unchanged starting tokens at ending prices to establish the holding benchmark.
  4. Separately account for assets in the position, accrued but unclaimed fees, incentives received and amounts already withdrawn.
  5. Include deposits, withdrawals, repositioning swaps, network costs and any management charges.
  6. Compare equivalent cash flows over the same period and check that fees have not been counted twice.

If you added funds, the holding benchmark must receive the same token quantities at the same times. Otherwise, the comparison mixes different capital contributions with investment performance.

An interface’s green percentage and current value are therefore insufficient. Transaction history, inventory and fee-accrual rules matter too. The official v3 position guide shows that the range and selected pool are distinct inputs. Matching token names alone does not make two positions equivalent.

Limitations and common mistakes

“Impermanent” does not mean “certain to disappear.” In the simple model, a return to the original relative price reduces the gap, but that price recovery is not promised. The economic underperformance already exists before withdrawal; after a strategy change, the original model’s assumptions may no longer apply.

A stablecoin pair is not risk-free either. If one token loses its peg, the pool can accumulate the relatively cheaper asset. Contract flaws, administrative permissions, token-specific behavior and network risks remain separate from the price calculation. A bridge introduces another layer, discussed in our crypto-bridge explainer.

Finally, IL compares the position with unchanged holdings, not with a continuously rebalanced trading strategy. These are different benchmarks. One successful historical position does not establish future returns, and a small IL figure does not rule out a large dollar loss. The terminology is useful only when the underlying comparison is clearly specified.

Frequently asked questions

Can I earn a profit and still have impermanent loss? Yes. Both portfolios can appreciate, while the LP position grows less than the original tokens would have if held unchanged.

Is waiting for price to return enough? Neither the return nor unchanged position conditions are guaranteed. Fees, costs and strategy changes require separate accounting.

Does an out-of-range position become empty? No. It generally holds one asset and earns no swap fees until price re-enters. The asset depends on the direction of movement and the position’s bounds.

Is there a universally safe LP yield? No. The Uniswap Labs explanation of IL describes price-change risk rather than promising compensation. A useful assessment needs the actual position’s conditions, not a single promotional percentage.

This is not investment advice.

Morning brief and news on TelegramEvery morning at 9:00 AM ET, before the US open: price, 24h scenario, squeeze risk. Key news around the clock.
Coins
Category
Tags
#DeFi#liquidity pools#impermanent loss#fees
Ryan Mitchell
On-chain and mining

Breaks down blockchain and mining data: hashrate, difficulty, holder and whale behavior.

Comments

Read next