What impermanent loss actually is
It is not a fee and not a hack. It is the gap between two outcomes: the value of your LP position, and the value of the two tokens if you had never deposited them. When price moves, the pool sells you out of the asset that rose and into the one that fell. Withdraw at that moment and you hold less of the winner than you started with.
Concentration multiplies it
In a full-range pool your liquidity is spread from zero to infinity, so the rebalancing is gentle. In a concentrated range the same money is packed into a narrow band, and the position flips composition far faster. On identical price moves:
| Range | Price −10% | Price −20% | Price +20% |
|---|---|---|---|
| ±5% | −4.13% | −10.17% | −7.79% |
| ±10% | −2.83% | −9.08% | −6.57% |
| ±20% | −1.43% | −6.36% | −4.34% |
Read the first column: a 10% drop costs a ±5% position nearly three times what it costs a ±20% one. That is the price of the higher fee rate a narrow range earns — the two move together, and no setting gives you one without the other.
The formula
L = V / (2√P₀ − P₀/√Pb − √Pa)
holding = x₀·P₁ + y₀ tokens you deposited, valued at the new price
position = x₁·P₁ + y₁ what the pool gives back at the new price
IL = position / holding − 1Where x and y are the base and quote amounts the CLMM math implies at each price. For a full range this reduces to the familiar 2√r/(1+r)−1 — we check ours against it: −0.4526% versus −0.4525% on a 21% move.
What actually reduces it
- A wider range, at the cost of a lower fee rate. Use the range optimizer to see the trade in numbers.
- Correlated pairs. SOL/JitoSOL barely diverges; SOL/USDC diverges with every SOL move.
- Fees, given time. The gap is a stock, fees are a flow — the longer a position sits in range, the more likely fees have covered it.
- A delta-neutral hedge — which cuts drawdown and also cuts returns. We published the delta-neutral LP backtest, including the part that argues against it.