defi basics
Impermanent Loss, Explained Without the Math (Then with the Math)
What impermanent loss actually is, when it matters, and how to estimate it before you deposit. With a worked example using ETH/USDC.
2026-05-13
TL;DR
- Impermanent loss (IL) is the gap between holding two assets in your wallet vs. holding them in an LP position.
- IL is real money lost. It’s “impermanent” only in the textbook sense: at the moment of withdrawal, it’s permanent.
- Trading fees can offset IL, but only if volume is high enough relative to volatility.
- Single-asset pools (lending, liquid staking) have zero IL by definition. If IL stresses you out, stay there.
The intuition first
Imagine you have $1000 worth of ETH and $1000 worth of USDC. You can do two things with them:
Option A. Hold them. ETH goes up 50%. You now have $1500 in ETH and $1000 in USDC. Total: $2500.
Option B. Deposit them into a Uniswap-style 50/50 ETH/USDC pool. ETH goes up 50%. The pool automatically rebalances as people trade: they buy USDC from your pool with their now-more-expensive ETH. When you withdraw, you get $1224 in ETH and $1224 in USDC. Total: $2448.
The $52 difference is impermanent loss.
That’s it. There is no other formula or fancy concept. Impermanent loss is what you give up to the pool’s rebalancing mechanism whenever asset prices diverge.
Why does it happen?
A liquidity pool maintains a constant product invariant: the quantity of asset A multiplied by the quantity of asset B stays roughly constant (ignoring fees). When ETH gets more expensive, traders take ETH out of the pool and put USDC in. This continues until the pool’s internal price matches the external market.
The result: the pool always holds less of the asset that went up and more of the asset that went down, exactly the opposite of what you’d want as an investor. If you deposited 50/50 and ETH 5x’d, you’d end up with very little ETH and a lot of USDC, when ideally you’d want all your value in the asset that mooned.
How big is the loss?
For a standard 50/50 pool (e.g., Uniswap V2), the loss depends only on the price ratio change between the two assets. Here’s the table:
| Price change | Impermanent loss |
|---|---|
| 1.25× (one asset +25%) | -0.6% |
| 1.5× | -2.0% |
| 2× | -5.7% |
| 3× | -13.4% |
| 4× | -20.0% |
| 5× | -25.5% |
| 10× | -42.5% |
Read this as: if one asset moves 2× relative to the other, your LP position is worth 5.7% less than just holding the two assets in your wallet.
Importantly, this works in both directions. If ETH halves while USDC stays at $1, that’s still a 2× ratio change, and you still take a 5.7% IL hit relative to having held.
When does IL actually matter?
Here’s the practical question: when do trading fees offset IL?
A pool earns trading fees on every swap. Volatile pairs traded heavily earn enough fees to compensate for IL, sometimes more than enough. Stablecoin pairs (e.g., USDC/USDT) have minimal IL because the price ratio barely moves; even small fees easily cover it.
The risk profile, broadly:
- Stable/stable pools (USDC/USDT, USDC/DAI): IL is essentially zero in normal conditions; only matters during depegs.
- Stable/volatile pools (USDC/ETH): IL is meaningful; you need consistent fee income to come out ahead.
- Volatile/volatile pools (ETH/BTC): IL exists but lower than stable/volatile pairs (the two assets often move together).
- Long-tail pairs (any new token / ETH): IL can be catastrophic; early LPs in many degen pairs have ended up with 90%+ losses while the token holders made money.
A worked example: ETH/USDC at Uniswap V3
Take a hypothetical $10,000 deposit into Uniswap V3 ETH/USDC 0.05% on Ethereum, when ETH is at $2,000.
You deposit: 2.5 ETH + $5,000 USDC.
Three months later, ETH is at $3,000 (50% up). Without doing anything else:
- HODL value: 2.5 ETH × $3,000 + $5,000 = $12,500
- Uniswap V2-style LP value (no fees): ~$12,250 (about $250 IL)
- Uniswap V3 LP value with fees: depends on your concentrated range and the volume that traded through it
If the pool traded $50M in volume during those three months and your position represented 0.5% of the active liquidity, you earned ~$125 in fees. Net: $12,375. Better than no-fees but still $125 below HODL.
This is the typical pattern for major pairs in normal markets: LP positions slightly underperform HODL but still beat sitting in cash. The point of being an LP isn’t to outperform, it’s to earn yield on a position you wanted to hold anyway, accepting some downside vs. perfect HODL.
When LP positions clearly win
LP positions can outperform HODL when:
- Range is tight and price stays inside. Uniswap V3 lets you concentrate liquidity in a price band. If the band captures most trading volume, fees can substantially exceed IL.
- Pool incentives. Many pools layer reward token emissions on top of fees. With a $10k position earning 15% APY in protocol tokens, IL of 2-3% is easily covered.
- Mean-reverting markets. If the price oscillates without trending, the pool earns fees from each oscillation while ending close to where it started.
LP positions clearly underperform HODL when one asset trends strongly. If you LP’d ETH/USDC at $1,500 and ETH ran to $4,000 over a year, you would have done much better just holding ETH.
The avoid-IL alternatives
If IL bothers you, you have options:
- Single-asset lending pools. Aave, Compound, Morpho, Sky Lending. You deposit one asset, earn interest, withdraw the same asset. No IL by construction. Lower yields, but the math is clean.
- Liquid staking tokens. Lido stETH, Rocket Pool rETH, Coinbase cbETH. Single-asset, earns staking rewards, no IL.
- Pendle PT/YT positions. More complex; isolates the yield from the principal. No traditional IL but its own math.
We tag every pool on our site with its IL risk (yes, no, or unknown) and exposure type (single or multi). Use those filters to avoid IL entirely if that’s your preference. Browse single-asset USDC pools for the most straightforward starting point.
Bottom line
IL is real money you can lose. It’s also a reasonable cost to pay for fee income if you’re depositing assets you wanted to hold anyway, into a pool with strong volume. It’s a terrible cost to pay if you’re chasing high APY in a low-volume long-tail pair.
Before depositing into any LP position, ask yourself: “Would I be happy holding both these assets in this 50/50 ratio for the next six months?” If yes, the LP position is reasonable. If no, if you really want to be all-in on one of the assets, just hold that asset directly.
Author's Q&A
› Does impermanent loss apply to liquid staking tokens like stETH?
Only if you're providing stETH as half of a liquidity pair (e.g. stETH/ETH on Curve). Just holding stETH in your wallet is a single-asset position - no IL by definition. The slight discount stETH sometimes trades at versus ETH is a separate phenomenon (liquidity discount on secondary markets), not impermanent loss.
› Why is IL worse in concentrated liquidity than in V2-style pools?
Concentrated liquidity amplifies both fees and IL by the same factor. If you concentrate 10× more tightly than a full-range V2 pool, you earn roughly 10× more fees per dollar - but the IL when price moves outside your range is also amplified, and once out of range you earn zero. The math says you should win if implied volatility from fees exceeds realized volatility from price movement; in practice on volatile pairs that's a difficult bet.
› How do I estimate IL before depositing?
Pick a worst-case price move (say, ETH drops 30% relative to USDC) and use the standard formula: IL ≈ 2√r/(1+r) − 1, where r is the price ratio. For a 30% drop r=0.7, IL ≈ −1.1%; for a 50% drop, IL ≈ −5.7%; for a 2× move, IL ≈ −5.7% (same magnitude by symmetry). Compare that worst-case IL to the trading fees you expect to earn over the same period. If fees < IL, the position is a bet against volatility.
› Is IL ever zero or positive?
IL is zero when prices return to the ratio you deposited at (which is rare in practice). IL cannot be positive in the strict definition - by construction, holding the assets directly always produces ≥ the LP value. What can be positive is total return: fees + IL. The whole point of LP-ing is collecting enough fees to more than offset the negative IL.
› If IL is so bad, why does anyone LP?
Three reasons. First, in low-volatility pairs (stablecoin/stablecoin, stETH/ETH) IL is structurally small and fees can compound meaningfully. Second, in high-volume pairs IL is offset by trading fees over time - Uniswap V3 ETH/USDC at 0.05% fee tier has been profitable for active LPs through most of 2024-2026. Third, reward emissions in protocol tokens can subsidize IL during a protocol's growth phase. The trap is treating advertised APY as net return without subtracting expected IL.