Why
The name is the problem. Impermanent loss suggests something that goes away if you wait, and the mechanism offers no such promise. It is only impermanent if the price returns to where you entered — an event nobody owes you and one that becomes less likely the longer you hold. The honest name is divergence loss, and renaming it changes how the position reads immediately.
What you are actually doing is selling an option, and the payoff shape says so. A constant-product pool automatically sells whichever asset is rising and buys whichever is falling. That is the position of someone who has sold a straddle: you collect a premium — the fees — and you lose whenever the price moves far in either direction. Gains are capped at the fees; losses grow with the square of the move for small moves and without bound for large ones. Once you see it as short volatility, every question becomes a familiar one.
Which reframes the only question worth asking: not "what is the APY" but "what volatility does this price imply, and is realised volatility below it?" Fees are the premium; divergence loss is the cost of being wrong about how much the pair will move. A pool paying 30% on a pair that moves violently can be a losing position, and a pool paying 4% on a tightly correlated pair can be an excellent one. The APY alone cannot distinguish them, which is why it is the number most often quoted.
And it explains why stable-to-stable pools dominate by volume while paying almost nothing. If the two assets barely diverge, the short-volatility cost is near zero and even a tiny fee is profit. The market is not confused; it correctly prices the fee to the volatility. Anywhere the fee looks generous relative to the pair's behaviour, the difference is being paid in emissions, which sends you back to where-yield-comes-from.
The reason this belongs next to the prediction-market work is that a subsidised market maker is the same structure. An LMSR subsidy is a market maker's expected loss, deliberately funded to create liquidity that would not otherwise exist. Both are cases of paying a known cost to guarantee someone can always trade — and in both, the mistake is to book the premium as revenue and the divergence as an accident.
How it works
The payoff, in one table
| Price ratio at exit | Pool value vs. simply holding |
|---|---|
| 1.00× (unchanged) | 0% — the only break-even point |
| 1.25× | −0.6% |
| 1.50× | −2.0% |
| 2× | −5.7% |
| 4× | −20.0% |
| 5× | −25.5% |
Symmetric in both directions — a halving costs the same as a doubling. That symmetry is the signature of a short straddle, and it is why the position has no view: it loses on movement, not on direction.
The two positions, side by side
| Holding both assets | Providing liquidity | |
|---|---|---|
| Income | None | Fees — the premium |
| Upside | Unbounded | Capped at the fees |
| Cost of movement | None | Grows with the size of the move |
| Equivalent to | Long the pair | Short a straddle, plus a fee stream |
| The real question | Will it go up? | Will it move more than the fees pay for? |
The break-even, which is the number to compute
Fees earned over a window, against divergence loss over the same window. Where they cross is the implied volatility the pool is quoting. Then compare it with realised volatility for that pair:
| If realised volatility is... | Then |
|---|---|
| Below the implied break-even | The position earns — you sold vol above its value |
| Above it | The position loses, however good the APY looked |
Why the stable pools make sense
| Pair type | Divergence | Fee needed to break even |
|---|---|---|
| Stable / stable | Near zero | Tiny — and that is what they pay |
| Correlated majors | Moderate | Moderate |
| Volatile / anything | Large | Large — usually topped up with emissions |
Cautions worth keeping
- The table above assumes a constant-product pool. Concentrated liquidity changes the magnitudes sharply — it is more leveraged in both directions, not safer.
- Fees are path-dependent: they accrue with volume, which correlates with the very volatility that costs you. The two are not independent.
- Divergence loss is a real loss the moment you withdraw. The word impermanent has cost people more than the mechanism has.
Where it lands in Jayverse
- DeFi: compute break-even volatility next to the APY. For every jayverse-defi pool, publish fees-earned versus divergence loss over the same window as a required metric alongside the advertised APY, not the APY alone.
- Verex: book the LMSR subsidy as an expected loss, not revenue. The market maker's subsidy is the same short-volatility position as an LP pool, so account for it explicitly as a funded cost of liquidity in market maker accounting.
- Auditor: flag any APY quoted without its offsetting divergence-loss or subsidy cost. Add it as a review check for DeFi pool designs and Verex's market maker, since the name "impermanent loss" already hides the real cost once.
Key expressions
| Expression | 뜻 · 쓰이는 자리 |
|---|---|
| divergence loss | (저자가 제안하는) 가격 괴리 손실 · impermanent loss 대신 쓰자고 주장하는 더 정확한 용어. "The honest name is divergence loss" |
| short a straddle | 스트래들을 매도한 포지션 · 양방향 변동에 모두 손실을 보는 옵션 매도 전략. "Short a straddle, plus a fee stream" |
| break-even | 손익분기점 · 수수료 수입과 손실이 정확히 같아지는 지점을 가리킬 때. "that number... is what the position is really quoting" |
| topped up with emissions | 토큰 발행(에미션)으로 보충되다 · 실제 수수료가 부족해 인센티브 토큰으로 메꿔질 때. "usually topped up with emissions" |
| book ... as revenue | ~을 수익으로 장부에 잡다 · 실제로는 위험 대가인 것을 이익처럼 기록하는 실수. "book the premium as revenue" |
| capped at | ~로 상한이 정해지다 · 이익이나 손실의 최대치가 고정될 때. "Capped at the fees" |
| path-dependent | 경로 의존적인 · 결과가 그 과정(거쳐온 경로)에 따라 달라질 때. "Fees are path-dependent: they accrue with volume" |
| correlate with | ~와 상관관계가 있다 · 두 변수가 같이 움직이는 경향을 말할 때. "which correlates with the very volatility that costs you" |
| without bound | 한계 없이·무한정 · 손실이나 값이 상한 없이 커질 수 있을 때. "without bound for large ones" |
| goes away if you wait | 기다리면 사라진다(는 착각) · impermanent라는 이름이 주는 잘못된 안심을 지적할 때. "something that goes away if you wait" |
| APY | 연 환산 수익률(Annual Percentage Yield) · 유동성 공급 보상률을 나타내지만 변동성 위험을 가리는 숫자로 비판됨. "not 'what is the APY' but 'what volatility does this price imply'" |
| LMSR | 로그마켓스코어링룰(Logarithmic Market Scoring Rule) · 예측시장 자동화 마켓메이커의 보조금 메커니즘, LP 포지션과 같은 구조로 비교됨. "An LMSR subsidy is a market maker's expected loss" |