DeFi & Web3Guide

Impermanent Loss Explained: The Math, Fees and Break-Even

A $4,841 ETH/USDC position lost $40.55 to impermanent loss in the 30 days to 15 November 2026 and earned about $28 in fees — the math, the break-even volume, and how to check any pool before depositing.

Impermanent Loss Explained: The Math, Fees and Break-Even
Illustration: CoinPaprika

Introduction

Ether (ETH) fell 59.0% between 1 September 2025 and 16 November 2026, yet a liquidity provider (LP) who held a full-range ETH/USDC position through the drop trailed plain holding by 9.20%. The loss peaked at 11.97% in June 2026, shrank to 2.89% as the price recovered in October and widened again in November. Impermanent loss is real, but it is far smaller than the price move behind it and it moves in both directions. This guide computes the formula and a price-ratio table, then tests both against a live Uniswap pool with dated prices and volume. It also shows how much trading volume a pool needs to break even, why concentrated ranges raise the stakes, and which fixes trade one risk for another.

Key Takeaways

  • A doubling or halving of one token against the other costs an LP 5.72% against holding, and a fourfold move costs 20.0%.
  • In the 30 days to 15 November 2026, the Uniswap V2 ETH/USDC pool paid its LPs only about 0.7 times its 0.95% impermanent loss in fees.
  • A pool needs daily volume of 21.2% of its value to cover a doubling held for 90 days at a 0.30% fee.
  • A range of 0.8 to 1.25 times the deposit price earns about 9.5 times the fees per dollar but loses 5.9% on a 25% move.
  • Bancor paused its impermanent loss insurance on 19 June 2022, so protection from an issuer is only as strong as the issuer.

What Is Impermanent Loss and Why Does Providing Liquidity Create It?

Impermanent loss is the amount a liquidity provider (LP) trails the same tokens held in a wallet after prices move. A pool always ends up holding more of the token that fell and less of the one that rose. The shortfall exists on paper only, and it vanishes if the price returns to where it started.

A Pool Sells What Rises and Buys What Falls

A liquidity pool holds two tokens, and every trade changes the mix. An automated market maker (AMM) such as Uniswap prices the pair from its reserves. Its constant-product rule, x × y = k, raises a token's price inside the pool with each purchase (Uniswap v2 docs, 2020). As ether (ETH) rises against the dollar stablecoin USDC, buyers take ETH out and pay USDC in. The LP's share shifts toward USDC, the token that lagged, while a wallet that kept the original tokens still holds all of its ETH. That gap is impermanent loss, also called divergence loss. Every one of the token swaps on Uniswap ↗ is a trade against an LP's position.

Arbitrageurs Drive the Rebalancing

The pool cannot see the outside market price. Arbitrageurs, traders who exploit price gaps between venues, trade against the pool until its price matches other markets. Each trade moves the pool further from the LP's original holdings. Researchers at Columbia University and the University of Chicago named this cost loss-versus-rebalancing (LVR). They identified it as the main adverse selection cost LPs pay (Milionis et al., 2022). Over 99.991% of the return variance of the Uniswap v2 ETH-USDC pool came from exposure to the market price itself (Milionis et al., 2022). The ETH price matters more to an LP than any other number.

One ratio fixes the size of the gap exactly: the price at withdrawal divided by the price at deposit.

How Do You Calculate Impermanent Loss for a Constant-Product Pool?

Impermanent loss depends on one number, the price ratio r: the token's price at withdrawal divided by its price at deposit. The formula is IL = 2√r ÷ (1 + r) − 1. The negative result is the share of the held value that the pool position lacks (Uniswap v2 docs, 2020).

The Formula From the Price Ratio

With k as the product of the two reserves ↗, a pool at price P holds √(k ÷ P) of the first token and √(k × P) of the second. Its value is 2√(k × P). Dividing that by the value of the original tokens at the new price leaves 2√r ÷ (1 + r). At r = 1 the loss is zero. Every other ratio gives a negative number: at r = 2 the formula reads 2 × 1.4142 ÷ 3 − 1 = −5.72%. Trading fees are excluded, so the formula isolates the price effect; any fees earned in the period are added on top and judged separately. Some authors measure the shortfall against the initial value, which changes the percentages. This article uses the Uniswap convention (Aigner and Dhaliwal, 2021).

A Worked Example With One ETH and USDC

The numbers below are illustrative. An LP deposits 1 ETH and 2,000 USDC when ETH costs $2,000, a $4,000 position with k = 2,000. ETH then doubles to $4,000, so r = 2. The pool now holds √(2,000 ÷ 4,000) = 0.7071 ETH and √(2,000 × 4,000) = 2,828.43 USDC, worth $5,656.85. Holding the original tokens would be worth $4,000 + $2,000 = $6,000. The gap is $343.15, or 5.72%, matching 2√2 ÷ 3 − 1. The LP sold 0.2929 ETH on the way up, at prices below $4,000, and still finished behind a holder. Fees earned along the way would reduce that gap, and the next sections measure them.

How Much Impermanent Loss Does Each Price Move Cause?

A 50% rise costs 2.0%, a doubling costs 5.7% and a fourfold move costs 20.0% of the value a holder would have kept. The table below was computed from the formula and reproduces the figures Uniswap publishes (Uniswap v2 docs, 2020).

Reading the Price-Ratio Table

The loss grows faster than the price move. A 25% move costs 0.62%, while a 100% move, four times larger, costs 5.72%, nine times the loss. A move of 200% costs 13.4%, and a 400% move costs 25.5%. "Price move" means the change in one token against the other: if both tokens rise 50% together, the ratio stays at 1 and no loss occurs. That is why pairs of tokens with prices that move in step carry little impermanent loss. In dollar terms, an LP whose tokens would have been worth $10,000 after a fourfold move holds $8,000 in the pool.

Why the Loss Is Symmetric and Reversible

A fall costs exactly what the matching rise costs. Halving the price (r = 0.5) loses 5.72%, the same as doubling it. A 75% fall (r = 0.25) loses 20.0%, the same as a fourfold rise (Uniswap v2 docs, 2020). Impermanent loss also depends only on the current ratio, not on the route taken to reach it. If the price returns to its level at deposit, the loss returns to zero. The loss becomes permanent only when the LP withdraws while the ratio is away from 1. The curve is also flat near 1 and steepens with distance, so exposure rises quickly once a price runs far from its starting point.

Price movePrice ratioImpermanent lossPool worth per $100 held
−75%0.25−20.0%$80.00
−50%0.50−5.72%$94.28
−25%0.75−1.03%$98.97
+25%1.25−0.62%$99.38
+50%1.50−2.02%$97.98
+100%2.00−5.72%$94.28
+200%3.00−13.40%$86.60
+300%4.00−20.00%$80.00
+400%5.00−25.46%$74.54

Data current as of November 2026.

Line chart of impermanent loss: 0.6% at a 25% rise, 5.7% at a doubling, 20.0% at a fourfold move, 25.5% at fivefold

The curve is flat near a ratio of 1, which is why short holding periods in calm markets cost so little.

Do Trading Fees Outrun Impermanent Loss in a Real ETH/USDC Pool?

In the 30 days to 15 November 2026, the Uniswap V2 ETH/USDC pool paid its LPs about 0.7 times its impermanent loss, even though trading tripled. Over 14 months, fees at a normal pace covered about a third of the loss. Both results use measured prices and volume, not invented figures.

The Measured Window

ETH cost $2,420.56 at the start of 17 October 2026 and $1,834.53 at the close of 15 November 2026, a 24.2% fall. The ratio r = 0.7579 gives a loss of 0.95% (DefiLlama, 2026). An illustrative position of 1 ETH plus $2,420.56 in USDC, $4,841.12 in total, was worth $4,214.54 in the pool against $4,255.09 held, a shortfall of $40.55. The pool is the original full-range Uniswap V2 ETH/USDC pair, so liquidity sits at every price and the formula applies without adjustment, as it does to any full-range position, including one wrapped as a pair token ↗. It charges a flat 0.30% fee on every swap. Since Uniswap governance switched on the protocol fee on 29 December 2025, LPs receive 0.25% in proportion to their share and the protocol keeps 0.05% (Uniswap docs, 2026).

Fees Against Loss on the Pool

The pool's on-chain swap records show $46.68M of trading volume in the 30 days to 15 November 2026, against $19.88M of liquidity at the start of the window (Ethereum RPC, 2026). At the LPs' 0.25%, those trades paid about $116,700, or 0.59% of the pool's starting value. That is roughly $28 on the example position, 0.7 times the $40.55 loss, and equals 7.1% annualized. The pool traded 8.3% of its value each day, more than three times its pace in the 30 days to 7 October, as the falling market drew heavy trading. It would still have needed about 11.9% a day to cover the loss. A 24% fall in one month sits close to the −25% row of the table above, which costs 1.03%.

Fourteen Months of Price Moves

A longer window shows the loss moving with the price. ETH cost $4,389.71 on 1 September 2025 and $1,797.71 on 16 November 2026, down 59.0%, a loss of 9.20% against holding (DefiLlama, 2026; CoinPaprika, 2026). At the 25 June 2026 low of $1,564.84, the same deposit trailed holding by 11.97%. The loss shrank to 2.89% as ETH recovered to $2,697.17 on 7 October 2026, then widened again as it fell, which is what "impermanent" means in practice: the gap follows the price both ways. At October's turnover of 2.6% a day, a 0.25% LP fee pays about 2.4% a year; at November's 8.3% it pays about 7.6%. A 9.20% loss over 441 days costs about 7.6% a year. Only a month as busy as November matched the loss, and a typical month covered about a third of it. The 14-month fee total is not measured here, so the comparison uses 30-day turnover as a stand-in.

MeasureValue
ETH price, 17 October 2026$2,420.56
ETH price, 15 November 2026 close$1,834.53 (−24.2%)
Impermanent loss, full range−0.95% ($40.55 on $4,841.12)
Pool volume, 30 days$46.68M
Fees to LPs at 0.25%about $116,700
Fees as share of pool value ($19.88M)0.59%
Fees on the example positionabout $28
Fees against lossabout 0.7 times

Data current as of November 2026.

Stat cards: ETH −24.2%, impermanent loss 0.95%, $116.7K LP fees on a $19.9M pool, 8.3% of pool value traded daily

Even a busy month can fall short, so the next question is how much volume a pool needs when prices move far.

How Much Trading Volume Does a Pool Need to Cover Impermanent Loss?

Break-even daily volume, as a share of pool value, equals the impermanent loss divided by the fee rate times the days held. A pool that turns over more than that figure pays its LPs for the risk; a pool that turns over less does not.

The Break-Even Turnover Formula

A pool earns the fee rate multiplied by its daily turnover, so the loss divides by fee × days. A price move of +100% held for 90 days costs 5.72%. At a 0.30% fee that needs 21.2% of pool value traded every day. The table lists the daily turnover required across holding periods. The calculation ignores gas costs, which fall on the LP for each deposit and withdrawal. The figures are illustrative: they assume the price ratio sits at the stated level for the whole period and the pool's size does not change.

What Real Pools Turn Over

The measured V2 pool traded 8.35% of its value a day over the 30 days to 15 November 2026, against 2.64% in the 30 days to 7 October (Ethereum RPC, 2026; DexPaprika API, 2026). Uniswap V2 LPs now keep 0.25% rather than 0.30%, so each figure in the table rises by a fifth for them. At November's pace the pool covers a doubling held for a year (6.2%) and about a 25% move held for 30 days (8.3%), but not a 50% move held for 90 days (9.0%). At October's pace it covered only moderate moves held for a year. The ETH/USDC pool on Uniswap V3 with a 0.05% fee traded $3.01B in the same window, on $76.2M of liquidity on 16 November. That is a turnover of 131% a day and fees of 1.97% of pool value, of which LPs keep three-quarters (Ethereum RPC, 2026; DexPaprika API, 2026). Early evidence from 17 Uniswap v3 pools was less kind. Fees totalled $199.3M against impermanent loss of $260.1M, and 49.5% of LPs earned negative returns (Loesch et al., 2021). Volume and liquidity for any pool are public on DEX pool data pages ↗.

Price move heldImpermanent loss30 days90 days365 days
+25%−0.62%6.9%2.3%0.6%
+50%−2.02%22.4%7.5%1.8%
+100%−5.72%63.5%21.2%5.2%
+300%−20.00%222.2%74.1%18.3%

Data current as of November 2026.

The 365-day column is the one a patient LP can hit, and only for moderate moves.

How Do Concentrated Liquidity Ranges Amplify Impermanent Loss?

Concentrated liquidity multiplies fees and loss together. A range from 0.8 to 1.25 times the deposit price earns about 9.5 times the fees per dollar of full range while the price stays inside it, and loses 5.9% on a 25% move against 0.6%. A separate guide covers range selection ↗.

Narrow Ranges Multiply Fees and Loss

Uniswap v3 lets LPs bound liquidity within a price range. The same capital then acts like a much deeper pool while the price stays inside it (Uniswap v3 whitepaper, 2021). Deeper liquidity captures a larger share of each trade's fee, because the position backs more of every swap that crosses its range. The exposure scales the same way: researchers describe this leverage as raising the fees earned and the impermanent loss together (Loesch et al., 2021). When the price leaves the range, the position holds a single token and stops earning fees until the price returns (Uniswap docs, 2026).

What the Range Numbers Look Like

These figures are computed from the Uniswap v3 position formulas. A range of 0.5 to 2 times the deposit price delivers 3.4 times the capital efficiency and loses 2.1% on a 25% move. A range of 0.8 to 1.25 delivers 9.5 times and loses 5.9%. A range of 0.9 to 1.11 delivers 19.5 times and loses 8.7%. On a 50% move the same three ranges lose 6.9%, 15.3% and 17.8%, against 2.0% for full range. The fee side shows in the measured pools. The concentrated V3 pool at a 0.05% fee turned over about 16 times as much of its value per day as the full-range V2 pool in the 30 days to 15 November 2026 (Ethereum RPC, 2026).

Bar chart of loss on a 25% price move: full range 0.62%, range 0.5x-2x 2.11%, range 0.8x-1.25x 5.87%, range 0.9x-1.11x 8.71%

Narrowing the range raises both the reward and the bill, and the bill arrives faster.

Which Pools and Strategies Reduce Impermanent Loss?

Each way to reduce impermanent loss moves the risk elsewhere. Correlated pairs add depeg risk, lopsided pools add slippage, hedges leave a running cost, and insurance depends on the insurer surviving ↗.

Pick Pairs Whose Prices Move Together

The price ratio drives the loss, so a pair whose prices move together has little to lose. Uniswap's v2 whitepaper notes that LPs on correlated assets, such as two dollar stablecoins, suffer less impermanent loss than LPs on ETH pairs (Uniswap v2 whitepaper, 2020). The trade-off is that the link can break. A stablecoin that loses its peg turns a low-risk pool into a large loss, a risk covered in the guide to stablecoins in DeFi ↗. Fee income on quiet pairs is also small, since traders pay little to swap between assets that barely move apart.

Weighted Pools, Hedges and Insurance

Balancer pools can hold uneven weights such as 80/20, where the heavier token causes less loss when it moves (Balancer docs, 2026). Computed from the weighted formula, an 80/20 pool whose 80% token doubles loses 3.27%, against 5.72% in a 50/50 pool. Lopsided pools pay for it in slippage, because one side holds much less liquidity (Balancer docs, 2026). Market exposure can also be hedged, but LVR remains as a running cost that fees must offset (Milionis et al., 2022). Insurance carries its own risk. Bancor V3 launched in May 2022 with instant 100% impermanent loss protection. It paused the protection on 19 June 2022 and halted the BNT minting that paid compensation (Bancor DAO, 2022).

ApproachHow it cuts the lossWhat it costsExample
Correlated assetsPrice ratio stays near 1Link can break; fees on quiet pairs are smallTwo dollar stablecoins
Weighted poolHeavier token weight shrinks lossHigher slippage, so less volume80/20: −3.27% against −5.72% on a doubling
Wider price rangeLess exposure per dollarFewer fees per dollarRange 0.5x–2x: −2.1% against −5.9% for 0.8x–1.25x
Hedge the price exposureOffsets the market moveLVR and hedging costs remainDynamic hedge (Milionis et al., 2022)
Protocol insuranceIssuer reimburses the lossDepends on issuer solvencyBancor, paused 19 June 2022

Data current as of November 2026.

No option removes the risk; each one chooses which risk to hold instead.

What Should a Liquidity Provider Check Before Depositing?

A pool is worth joining when its measured daily turnover clears the break-even for the price moves the LP expects. Impermanent loss is one risk among several ↗, and the others appear as total losses rather than a gap against holding.

Run the Numbers First

Four checks take minutes. First, divide the pool's 30-day volume by its liquidity to get daily turnover, and multiply by the share of the fee tier that LPs keep to get the daily fee rate. Second, read the break-even turnover from the table above for the price move the LP fears and the planned holding period. Third, the pair itself: ETH and USDC do not move together, so the full formula applies. Fourth, upkeep: a concentrated position that leaves its range earns nothing, and every range reset is a paid transaction. A pool that fails the second check pays its LPs a loss on average, as the Uniswap v3 evidence shows (Loesch et al., 2021).

Losses That Are Not Impermanent Loss

Impermanent loss measures a gap against holding, and several larger losses sit outside it. A pool whose liquidity is pulled by its creator loses principal outright, as in the liquidity drains on DEX pools ↗ that on-chain checks can detect. A stablecoin depeg, a contract bug or a compromised token approval can wipe out a position with no price ratio involved. The math carries over to other settings. Pools for tokenized bonds and funds follow the same formula, as laid out in the guide to real-world asset liquidity pools ↗. Scam pools and failed contracts leave no gap to measure, only a balance of zero. The ratio sets the loss, the fees set the offset, and the contract sets everything else.

Conclusion

Impermanent loss is a function of one ratio, the price at withdrawal divided by the price at deposit. The formula gives a number for every move, the table shows that moderate moves cost little, and the loss returns to zero when the ratio does. Fees are the other half of the ledger, and the measured pool shows both outcomes: a calm month to 7 October where fees won, and a falling month to 15 November where they fell short even though trading tripled.

The difference between a good pool and a bad one stays hidden until prices move. A pool trading 2.6% of its value a day looks healthy in a quiet month and fails the break-even test the moment ETH moves far; in November the same pool traded 8.3% a day and still did not cover a 24% fall. Concentrated ranges sharpen the problem, because the same bet earns more fees and loses more, and the loss arrives before a slow-reacting LP can respond.

Before depositing, you should divide the pool's 30-day volume by its liquidity and compare the result with the break-even table for the move you fear. Check the pair, the fee tier and the range upkeep as well. A pool that passes keeps paying you for risk; one that fails charges you for it.

Quick Stats

  • 5.72% — impermanent loss when one token doubles or halves against the other
  • $46.68M — 30-day volume of the Uniswap V2 ETH/USDC pool to 15 November 2026
  • 8.3% — share of that pool's value traded each day
  • 11.97% — peak loss for an LP who entered ETH/USDC on 1 September 2025
  • 49.5% — Uniswap v3 LPs with negative returns in an early study of 17 pools
  • 99.991% — share of ETH-USDC LP return variance driven by market price

Data current as of November 2026.

FAQ

?Is impermanent loss the same as losing money?

No. Impermanent loss compares a pool position with holding the same tokens, so an LP can lose money in dollars while the loss against holding is small. In the measured window ETH fell 24.2%, so the position lost $626.58 in dollars, of which only $40.55 was impermanent loss; a holder lost the other $586.03 as well. A real dollar loss needs falling prices, a failed contract or a withdrawal at a bad moment.

?Is providing liquidity better than holding ETH and USDC?

It depends on volume and on the move. In the 30 days to 7 October 2026, a calm month, LP fees on the Uniswap V2 ETH/USDC pool covered the loss about 3.1 times over. In the 30 days to 15 November, when ETH fell 24.2%, they covered about 0.7 times it. Across 14 months, estimated fee income at a normal pace covered about a third of the 9.20% loss. In an early study of 17 Uniswap v3 pools, fees of $199.3M fell short of $260.1M in impermanent loss, so holding won on average.

?Does a higher fee tier protect liquidity providers from impermanent loss?

No. The fee tier changes income, not the loss, which depends only on the price ratio. A higher tier pays more per trade, and a lower tier can make up for it in volume. The measured 0.05% ETH/USDC pool traded 131% of its value a day in the 30 days to 15 November 2026. Compare turnover times fee with the break-even table instead of picking the highest tier.

?What is loss-versus-rebalancing and how does it differ from impermanent loss?

Loss-versus-rebalancing (LVR) measures what arbitrageurs extract by trading against stale pool prices. It is a running cost that grows with volatility and pool depth, so fees must offset it. Impermanent loss compares only start and end positions and reverses if the price returns. Columbia and Chicago researchers show that hedging market exposure leaves LVR behind as the cost an LP still carries.

?What happens to a concentrated position when the price leaves its range?

The position stops earning fees and converts entirely into one token. If the price falls below the range, it holds only the token that fell; above the range, only the token that rose. Fees resume if the price re-enters the range. Full-range positions avoid this because they stay active at every price, at the cost of lower fees per dollar.

?Do lenders on a protocol like Aave suffer impermanent loss?

No. Impermanent loss comes from a pool that rebalances two tokens along a price curve. A lender deposits one asset and earns interest paid by borrowers, so no pool ratio shifts. Lenders carry other risks instead, such as bad debt from failed liquidations and smart contract bugs. The loss against holding two tokens belongs to AMM liquidity providers, who use one of several ways to earn in decentralized finance (DeFi) ↗.

?Do farming rewards cancel out impermanent loss?

Only if the reward tokens keep their value. Rewards are paid in a separate token ↗, and the LP bears that token's price risk on top of the pool's. A reward that pays 20% a year in a token that falls 40% leaves the LP behind. Treat reward income as unreliable until sold, and compare the sold amount with the break-even loss.

?Can a liquidity provider deposit only one token to avoid impermanent loss?

A constant-product pool must hold both tokens, so a single-token deposit is split by a swap and the exposure remains. Bancor V3 offered single-sided staking with instant 100% impermanent loss protection from May 2022, but paused that protection on 19 June 2022. Uniswap v3 allows one-sided positions outside the current price, which act like limit orders rather than loss protection.

References / Sources

Platform & Company Data

Pool data, prices and protocol documentation used for the measured examples.

  • DexPaprika: Uniswap V2 and V3 ETH/USDC pool data (dexpaprika.com, Nov 2026)
  • Ethereum RPC: Uniswap V2 and V3 ETH/USDC swap events and reserves (rpc.mevblocker.io, Nov 2026)
  • DefiLlama: ETH daily prices, coins API (defillama.com, Nov 2026)
  • CoinPaprika: ETH Market Data (coinpaprika.com, Nov 2026)
  • Uniswap: Understanding Returns, Pools and Concentrated Liquidity (developers.uniswap.org, 2026)
  • Uniswap: Protocol Fees after the UNIfication vote of December 2025 (developers.uniswap.org, 2026)
  • Balancer: Weighted pools and impermanent loss (docs-v2.balancer.fi, 2026)
  • Bancor: Ratification of emergency actions, 19 June (gov.bancor.network, Jun 2022)
Academic & Technical

Papers and whitepapers behind the formulas and the LP return evidence.

  • Loesch et al.: Impermanent Loss in Uniswap v3 (arxiv.org, Nov 2021)
  • Milionis et al.: Automated Market Making and Loss-Versus-Rebalancing (moallemi.com, 2022)
  • Aigner and Dhaliwal: Uniswap, Impermanent Loss and Risk Profile (ssrn.com, 2021)
  • Adams et al.: Uniswap v2 Core (app.uniswap.org, 2020)
  • Adams et al.: Uniswap v3 Core (blog.uniswap.org, Mar 2021)
Share this guide
Written by Bartek Hagan

Original CoinPaprika guide.

Report a correction
All guides

Cryptocurrencies are highly volatile and involve significant risk. You may lose part or all of your investment.

All information on Coinpaprika is provided for informational purposes only and does not constitute financial or investment advice. Always conduct your own research (DYOR) and consult a qualified financial advisor before making investment decisions.

Coinpaprika is not liable for any losses resulting from the use of this information.