Impermanent loss is the inventory-value shortfall of a liquidity position relative to holding its original assets outside the pool, measured before adding fees or rewards and under the stated pool assumptions. A position can gain dollars while still underperforming that holding benchmark. Trading fees may offset the gap, but a displayed fee yield does not establish that they will. The essential calculation compares final assets, fees, rewards, and costs with a consistent alternative over the same period.
The analytical question is therefore not simply whether a pool pays an attractive yield. It is whether the compensation is sufficient for the inventory changes and other costs required to earn it. The answer depends on the chosen benchmark, the pool's design, and the path of trading activity.
Begin With The Assets You Would Otherwise Hold
Suppose a hypothetical investor supplies one ETH, valued at $2,000, and 2,000 stable units worth one dollar each. The initial position is $4,000. The comparison portfolio holds those exact same quantities without entering the pool.
This benchmark matters. Comparing the pool only with the original $4,000 asks whether it made money in dollar terms. Comparing it with the unchanged asset quantities asks whether providing liquidity improved the result relative to holding them. Both questions are legitimate, but they can produce different answers.
The worked example assumes a constant-product pool, equal starting asset values, no fees in the trading invariant, no intermediate deposits or withdrawals, and arbitrage that aligns the pool with the external price. The Uniswap v2 whitepaper provides a primary description of constant-product market making. The calculations here isolate a simplified version to make the inventory effect visible.
Under these assumptions, the reserve product associated with the position is one ETH multiplied by 2,000 stable units, or 2,000. When the external ETH price changes, arbitrage changes the pool's asset mix. The position ends up with fewer units of the asset that appreciated relative to the other asset.
That inventory adjustment is the economic service the provider supplies: making assets available to traders across prices. The fee is compensation for participating in that process, not a return earned while retaining exactly the original portfolio.
A Price Doubling Produces A Dollar Profit And A Relative Shortfall
Assume ETH rises from $2,000 to $4,000. The modeled position now contains approximately 0.707107 ETH and 2,828.427 stable units. Each side is worth $2,828.427, so the combined value is $5,656.854.
The holding portfolio is worth one ETH at $4,000 plus 2,000 stable units: $6,000. The liquidity position has gained $1,656.854 from its initial value, but it is $343.146 behind holding.
| Comparison | Calculation | Result |
|---|---|---|
| Liquidity position's dollar gain | 5,656.854 − 4,000 | $1,656.854 |
| Liquidity position's return on initial capital | 5,656.854 / 4,000 − 1 | 41.42% |
| Holding portfolio's dollar gain | 6,000 − 4,000 | $2,000 |
| Shortfall relative to holding | 5,656.854 / 6,000 − 1 | −5.72% |
| Dollar shortfall as share of initial capital | 343.146 / 4,000 | 8.58% |
The last two rows explain a common comparison error. A 5.72% shortfall measured against the final holding portfolio is not the same quantity as 5.72% of initial capital. A quoted fee yield based on deposited capital cannot be compared with that percentage without aligning the denominators.
More generally, if the relative asset price changes by a factor of r, the fee-free constant-product value relative to holding is:
2 × sqrt(r) / (1 + r)
Subtracting one gives the relative shortfall. This expression is a result for the stated model, not a universal formula for every liquidity position.
| Relative Price Factor | Price Change | Modeled Shortfall Relative To Holding |
|---|---|---|
| 0.25 | −75% | −20.00% |
| 0.50 | −50% | −5.72% |
| 1.00 | 0% | 0.00% |
| 1.50 | +50% | −2.02% |
| 2.00 | +100% | −5.72% |
| 4.00 | +300% | −20.00% |
The symmetry applies to relative price ratios, not equal percentage moves up and down. Doubling and halving are reciprocal changes. A 50% increase and a 50% decline are not.
The formula also concerns the price of one pool asset relative to the other. It is not necessarily driven by either asset's dollar price alone. If two hypothetical assets both double in dollar terms while their relative price is unchanged, this model has no relative inventory shortfall from that move. The position and the unchanged holding portfolio both appreciate. If one doubles while the other triples, their relative price changes, and the inventory comparison must use that ratio.
This distinction helps explain why historically correlated assets may appear attractive for liquidity provision while still carrying a meaningful failure mode. Correlation is an observation about co-movement, not an obligation to keep moving together. A wrapper, redemption claim, or pegged token can lose its expected relationship with the other asset. A model that assumes the relationship persists is then answering a narrower question than the provider faces. Separate the ordinary relative-price scenario from a scenario in which one asset's economic claim deteriorates.
Calculate The Fee Hurdle In Dollars
Return to the price-doubling example. Assume the provider receives $180 of fees valued at the end of the period and incurs $35 of total transaction and management costs. For clarity, fees are treated as a separate cash receipt, without reinvestment or changes to the modeled inventory.
The final net result is $5,656.854 + $180 − $35 = $5,801.854. The position remains $198.146 behind the $6,000 holding benchmark. Earning fees and beating the benchmark are different findings.
To match holding in this particular scenario, gross fees would need to cover the $343.146 inventory shortfall plus $35 of costs: $378.146. That is approximately 9.45% of the initial $4,000 over the modeled period. It is not an annual yield forecast, and changing the ending price changes the hurdle.
This is where APR, APY, and the source of crypto yield become relevant. An annualized fee display may extrapolate a short observation window. A reward display may value tokens at their current price. Neither automatically matches the actual cash flows and final portfolio used in an opportunity-cost comparison.
If fees remain inside the pool, the analysis must incorporate that mechanism instead of adding the same fees again outside the position. If rewards are sold, use actual net proceeds. If they are retained, show their quantity and valuation separately. The goal is to prevent accounting choices from manufacturing an apparent outperformance.
High Volume Can Increase Fees And Adverse Inventory Selection
Consider a second, purely illustrative fee calculation. A provider has $100,000 of a uniformly shared $5 million pool, a 2% share. The pool processes $500,000 in daily volume at an assumed 0.3% fee, with no protocol deduction. Total fees are $1,500, and the provider's simple proportional share is $30 for that day.
This arithmetic says nothing about whether the day's inventory changes were favorable. It also assumes the provider's share remains constant and participates in every relevant trade. Concentrated positions and different fee distributions require a more detailed calculation.
The quality of order flow matters. Repeated trades around a stable price can generate fees with a small ending inventory gap. Trades that repeatedly exploit a pool's stale relative price can impose a different economic burden. Two pools with the same volume and fee rate can therefore produce different provider outcomes.
The research paper Automated Market Making and Loss-Versus-Rebalancing analyzes adverse selection against a dynamic rebalancing benchmark. That benchmark differs from simply retaining the original assets. Its model helps separate fee income from costs associated with informed arbitrage, but its assumptions do not make it a universal profit estimator for live positions.
The practical implication is to ask what trading activity requires the provider to give up. Volume and usable DEX liquidity describe different aspects of a market. For a provider, volume is also an incomplete description of the compensation problem. The same turnover can arrive under very different price paths and inventory demands.
A Return To The Starting Price Does Not Erase Every Cost
Under the simplified fee-free model, if the relative price returns exactly to its starting point and the position remains unchanged, the terminal inventory comparison returns to equality. This is the narrow basis for describing the relative loss as potentially reversible.
Actual economic experience can contain costs that do not reverse. The provider may have paid to enter and exit, repositioned a range, sold accumulated rewards, or withdrawn after a price move. A later market recovery does not automatically undo those separate decisions and expenses.
There is also a benchmark issue. A static holding comparison evaluates starting assets against ending assets. It does not describe every opportunity available to a different strategy during the intervening path. A positive result against one benchmark is useful evidence, but it is not proof that the position dominated all alternatives.
The countercase deserves equal attention. If relative prices remain close, fees are substantial, and operational costs are modest, liquidity provision can outperform holding in the stated comparison. The framework should be capable of recognizing that result. It is designed to identify the conditions of success, rather than assume that every inventory adjustment is a reason to reject the activity.
What would change a skeptical assessment? A complete record showing fee receipts, inventory values, costs, and matched benchmark cash flows would be persuasive. A high displayed APR without those components would not establish the same conclusion.
Concentrated Positions Need Their Own Inventory Model
A narrow position can earn fees on less capital around a chosen price region, but the full-range formula above does not describe its complete risk. Its inventory can change more sharply as the market approaches a boundary, and its exposure beyond that boundary can differ substantially from the starting mix.
Uniswap's liquidity-provider risk guidance identifies out-of-range positions, smart-contract risks, and token-related risks as separate considerations. A fee estimate is therefore only part of the assessment. Time outside the active region and the costs of managing the position also matter.
Repeatedly resetting a narrow range introduces a sequence of portfolio changes. A calculation that applies one terminal full-range impermanent-loss percentage to the entire history may miss the effects of those changes. Each deposit, withdrawal, and repositioning should enter the cash-flow record at its actual time.
Execution costs belong in that record too. Price impact, slippage, and network costs can affect the swaps used to create or rebalance a position. These costs may be especially material when the position is small or when rebalancing happens during thin market conditions.
Nor should all reported protocol income be attributed to the provider. The distinction between protocol revenue and value reaching stakeholders applies to fee analysis: identify which part of a trading charge reaches the position after the system's distribution rules. Gross activity is not the same as a particular participant's receipt.
Build A Matched Cash-Flow Comparison
A defensible evaluation begins with quantities, then adds valuations. Record the assets deposited, the units later withdrawn, fees claimed or retained, rewards sold or held, and all separate costs. For every contribution or withdrawal, create an equivalent cash flow in the comparison portfolio at the same time.
This prevents later deposits from being mistaken for investment gains and prevents a changing capital base from distorting the measured yield. It also makes the comparison reproducible: another reader can change a valuation assumption without reconstructing the entire history.
| Component | Liquidity Position | Holding Benchmark |
|---|---|---|
| Initial assets | Actual deposited quantities | Same starting quantities |
| Later cash flows | Actual additions and withdrawals | Matched timing and economic amounts |
| Ending inventory | Actual assets attributable to position | Assets remaining under benchmark rule |
| Fees and rewards | Count once, using disclosed values | Include only if benchmark genuinely earns them |
| Costs | Entry, management, exit, and relevant charges | Comparable costs under benchmark assumptions |
A protocol's headline TVL cannot supply these missing records. Large deposits may demonstrate substantial participation, but they do not reveal whether a particular position outperformed its matched alternative.
The final finding should identify both the result and its sensitivity. A position might beat holding before management costs and lag afterward. It might outperform when rewards are valued at their sale proceeds but underperform under a later price assumption. Those distinctions explain the result rather than weaken it.
Impermanent loss becomes a useful analytical concept when it is attached to a specific portfolio comparison. The decision-relevant output is the dollar fee hurdle, the uncertainty around earning it, and the residual risks that fees do not measure. Keeping those quantities explicit makes the economics of liquidity provision far clearer than a yield percentage alone.
Max Clark
Max Clark writes about cryptocurrencies, with particular interests in altcoins and NFTs. His experience in finance informs his coverage of monetary systems and decentralized finance. Outside his writing, he enjoys sports and adventures.