Liquidity provision of utility indifference type in decentralized exchanges

Basile Maire; Marcus Wunsch; Masaaki Fukasawa

REVIEW 2 major objections 4 minor 1 cited by

Liquidity provision of utility indifference type in decentralized exchanges

T0 review · 2 major / 4 minor · reviewed 2026-08-09 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A nonzero fee eliminates loss-versus-rebalancing in concentrated-liquidity DEX pools when the external market price is continuous.

desk verdict A rigorous unified treatment of concentrated-liquidity AMMs with a real no-LVR result under continuous prices; the main gap is a sketched existence proof that should be tightened. read the letter →

arxiv 2502.01931 v1 pith:DNQW3JGD submitted 2025-02-04 q-fin.TR q-fin.MF

Masaaki Fukasawa , Basile Maire , Marcus Wunsch This is my paper · ORCID

classification q-fin.TRq-fin.MF MSC 91G8060G44

keywords decentralizedfinanceautomatedmarketmakersarbitrageimpermanentlossloss-versus-rebalancingconcentratedliquidityutilityindifferenceUniswapv3

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper builds a rigorous continuous-time model of liquidity provision in decentralized exchanges whose trading functions come from a concave utility of indifference type, which includes Uniswap v3's concentrated liquidity as a special case. It derives no-arbitrage conditions, identifies the optimal arbitrage trade when they are violated, and proves that with zero fees the liquidity provider's wealth is a concave function of the external price, yielding impermanent loss. The central result is that once a nonzero transaction fee is charged and the external price is continuous, the pool admits well-defined arbitrage-free reserves and the extended loss-versus-rebalancing is nonpositive, so the liquidity provider does not lose to rebalancing regardless of the fee size. It also shows that a pool with many liquidity providers is equivalent to a single representative provider, and that the Uniswap v3 architecture is optimal in how it allocates fee income proportionally to liquidity depth.

What carries the argument

The central object is the utility-indifference trading function φ(x,y,ξ,η)=u_*(x+(1-τH(ξ))ξ, y+(1-τH(η))η), where u_* is a strictly concave utility shifted by a liquidity range. This reduces the pool to an implicit decreasing convex curve Y=f_*(X), with internal price S=-f'_*(X) and fee-spread bid-ask prices A=S/(1-τ), B=(1-τ)S. The free-of-arbitrage condition A≥S*≥B lets the proof compare LP wealth V to the profit-and-loss of trading the reserve position in the external market; convexity of f_* makes the residual nonnegative under continuous S*. Tanaka's formula and the Skorokhod map are the tools that make the continuous-time constructions rigorous.

What would settle it

For a continuous positive price process S*, construct the arbitrage-free reserves from Theorem 3, then compute the extended LVR of Remark 4: V_T - V_0 - ∫_0^T (X_t+X^f_t)dS*_t. The theorem predicts this is always nonpositive; any simulated path or real market dataset with a continuous price series that yields a positive value would falsify Theorem 4.

Watch

Extended reading notes

Core claim

The discovery is Theorem 4: if the pool is free of arbitrage at all times and the external price is positive and continuous, the wealth process minus the profit-and-loss of holding the reserve position in the external market is nondecreasing. With a nonzero fee, the extended loss-versus-rebalancing is therefore nonpositive, meaning liquidity providers face no divergence loss relative to a model-free rebalancing strategy, no matter how small the fee is. The proof uses the convexity of the implicit reserve curve and the fee-induced bid-ask spread; the continuity of the external price is essential, since a jump makes the quadratic covariation term negative and turns LVR positive (Remark 5).

Load-bearing premise

The conclusion that there is no loss-versus-rebalancing under fees rests on the external market price process being continuous; if the price can jump, the paper itself shows the quadratic-covariation term makes LVR positive.

Editorial extensions

If this is right

A nonzero transaction fee in a Uniswap v3-type pool fully super-hedges impermanent loss against the model-free rebalancing strategy when the external price is continuous.
The faster a blockchain produces blocks, the closer the discrete process is to the continuous model, so arbitrage losses should shrink; the paper cites empirical evidence to this effect.
A multi-provider pool with a common fee tier behaves exactly like a single representative constant-product market maker, so wealth dynamics do not depend on whether a provider creates a new pool or joins an existing one.
The Uniswap v3 construction is optimal in the specific sense that the optimal allocation of an incoming order across subpools yields fee income proportional to each subpool's liquidity depth.
With zero fee and concentrated liquidity, impermanent loss is unavoidable and is represented by the positive quadratic-variation term (1/2)∫ f''_*(X)d⟨X⟩.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

An extension the paper leaves open is heterogeneous fee tiers: the representative-provider and optimality results are proved for a common fee τ, and different fees across subpools would break the unique internal price argument and require a different allocation rule.
Because the proof singles out the quadratic-covariation term d[S*, X+X^f] as the source of LVR under jumps, a natural next step is to decompose LVR for cadlag prices into a continuous part (nonpositive under fees) and a jump part proportional to covarying arbitrage trades, which could guide fee design when price jumps are frequent.
The super-hedge uses the strategy X+X^f, which may have large turnover when S* is continuous but rough; its practical value depends on transaction costs in the external market, a friction the paper does not model.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Read the Fukasawa–Maire–Wunsch paper on utility-indifference AMMs. The headline: it gives a clean unified treatment of concentrated liquidity and fee collection, and the central no-LVR result under continuous prices with a nonzero fee is real and clearly derived. The paper deserves a serious referee, with one caveat about the proof of Theorem 3.

What is new: the Legendre-transform representation of LP value for concentrated liquidity, the Tanaka-formula LVR identity, and the no-LVR theorem under fees. The inf-convolution treatment of multiple LPs is a fresh and useful way to see why Uniswap v3's fee allocation is proportional to liquidity depth. The authors build on their own earlier work and on Milionis et al.; the citations are appropriate and the extension is genuine.

The main soft spot is in Section A.4, Theorem 3. The construction of arbitrage-free reserve processes for general x* and y* is only sketched. The Skorokhod reflection on [0,a] works for the unbounded case, but for the concentrated-liquidity case the proof says 'we can repeat the same argument to concatenate X' after boundary hits. No argument is given that concatenation preserves RCLL and finite variation when the boundary is hit infinitely often. This matters because Theorem 4 is conditional on existence, and the paper uses Theorem 3 to claim that existence. I don't think this is a fatal flaw—likely fillable with a standard excursion argument—but a referee should ask for it.

The continuity assumption on S* is explicitly flagged in Remarks 5 and 6, and the jump case is discussed. That is honest. There are no numerics, but this is a theory paper; fine.

Bottom line: the paper is for people working on AMM pricing and LVR. It should be sent to peer review, with a request for a complete proof of Theorem 3. I would be happy to see it published after that.

Referee Report

2 major / 4 minor

Summary. The paper develops a mathematical framework for constant function market makers of utility indifference type, covering concentrated liquidity as in Uniswap v3 and fees collected in a separate account. Section 3 characterizes no-arbitrage conditions and optimal arbitrage trades. Under zero fees, Theorem 1 represents LP wealth as the Legendre transform of the implicit function and Theorem 2 derives a loss-versus-rebalancing formula via Tanaka's formula. Under positive fees, Theorem 3 asserts the existence of arbitrage-free reserve processes, and Theorem 4 states that if the external price is continuous, the process V - ∫(X+X^f)dS* is nondecreasing, so the extended LVR is nonpositive regardless of fee size. Section 6 aggregates multiple liquidity providers through inf-convolution and argues that the Uniswap v3 fee-sharing rule is optimal in this sense.

Significance. If the results are fully established, they extend the existing LVR and impermanent-loss analysis to concentrated liquidity and to fee-collection architectures of Uniswap v3 type. The use of Tanaka's formula to handle boundary behavior is natural, and the inf-convolution aggregation in Section 6 is a useful conceptual contribution. The paper is transparent about the continuity assumption on S* and flags the jump case in Remarks 5 and 6. No numerical experiments are provided, but for a pure theory paper that is not a deficiency. The main reservation is that the proof of Theorem 3, on which the unconditional reading of the headline no-LVR claim depends, is only sketched for concentrated liquidity.

major comments (2)

[Section A.4 (Theorem 3)] The proof of Theorem 3 is incomplete for the concentrated-liquidity case x* > 0 or y* > 0. After defining sigma as the first time the unbounded construction leaves the admissible band, the text says that in the case X_sigma < 0 one modifies X to be 0 until sigma', and in the case Y_sigma < 0 one modifies X to be x-dagger until sigma', and then "We can repeat the same argument to concatenate X." No argument is given that this concatenation remains RCLL and of finite variation when the boundary is hit infinitely often, nor is it explained how the Skorokhod reflection is restarted after S* re-enters the admissible band. The assertion "In any case, sigma' > 0" is also not generally justified: if S* is already on the relevant side of the boundary at sigma, then sigma' = sigma and the proposed interval has zero length. Because Theorem 4 assumes the LP is free of arbitrage at every time, and because the paper's abstract claims well-definedness of arbitrage-free reserve processes for concentrated liquidity, this gap is load-bearing for the headline conclusion.
[Section A.4 (Theorem 3), jump handling] The proof also dismisses the handling of jumps in X^o with "it is trivally possible to take Delta X^a_t such that the LP is free of arbitrage at time t," but no construction is given. Since X^o is assumed only piecewise constant RCLL, the arbitrageur's jump Delta X^a_t must be chosen so that the post-jump reserves satisfy the no-arbitrage inequalities (6)-(7), and it is not immediate that a finite choice always exists when S* jumps to an extreme value. This is a second unresolved point in the same existence theorem.

minor comments (4)

[Section A.4] There is a typo "trivally" for "trivially" and "concatanate" for "concatenate" in the proof of Theorem 3.
[Section A.4] The displayed line "dφ = dψ + dη = d log S* (1−τ)S" appears to omit operators; it should read dφ = dψ + dη = d log S* − d log S, consistent with the definition of φ.
[Section 5, Theorem 4 and Remark 4] The integral in (13) is written with the right-continuous integrand X_t + X^f_t, but the Stieltjes integration-by-parts formula actually defines the left-continuous version because S* is continuous. The notation could be clarified to avoid ambiguity.
[Section 6.3] The statement that the derived properties "are consistent with the description of Uniswap v3 in Section 3 of [5]" is quite terse; giving the precise correspondence would help the reader verify the claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central theorems are derived from stated assumptions and external mathematical results, not from their own conclusions.

full rationale

The paper's central derivation chain is self-contained. Theorem 1 is proven in A.2 via the Legendre transform v(p) = inf_x {xp + f_*(x)} and Lemma 1; Theorem 2 is proven in A.3 via Tanaka's formula; Theorem 4 is proven in A.5 using only the convexity of f, the fee-collection inequality (12) dX^f ≥ τ/(1−τ)dX^↑, dY^f ≥ τ/(1−τ)dY^↑, and the no-arbitrage bounds A_− ≥ S* ≥ B_−. No parameter is fitted to data, and no 'prediction' is a renamed input. The self-citations [9,10] are presented as extensions, and their prior results are not used in place of a proof here; the appendix contains the relevant arguments. The 'optimality' claim in Remark 7 is explicitly definitional ('in the sense that...'), so labeling the inf-convolution allocation as optimal is a transparent naming, not a hidden equivalence. The only notable weakness is that Theorem 3's proof in A.4 for general concentrated liquidity is sketched: the concatenation after boundary hits ('We can repeat the same argument to concatanate X') is not fully detailed. That is a rigor gap in existence, not evidence that a conclusion is assumed by construction.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters are fitted to data; the model parameters x*, y*, and tau are inputs that define the liquidity range and fee. The proofs rely on standard convex analysis, Tanaka's formula, and the Skorokhod map, all cited from prior literature. No new entities or forces are introduced.

assumptions (5)

domain assumption u is strictly concave, three times continuously differentiable, increasing, with limits at 0 and infinity.
Section 2 defines the utility-indifference class; these properties ensure the implicit function f is strictly convex and the ask/bid price functions are well-defined.
domain assumption External price S* is a positive continuous semimartingale (Theorem 2) or positive continuous (Theorem 4).
Continuity is used to apply Tanaka's formula and to make the no-LVR result hold; Remark 5 shows jumps reintroduce LVR.
domain assumption Under tau > 0, reserve processes are of finite variation and Xo is piecewise constant RCLL (Theorem 3).
This ensures Stieltjes integration and makes the construction of the arbitrage-free reserve process via the Skorokhod map possible.
standard math Existence and properties of the Skorokhod map on [0,a] (Kruk et al. [12]).
Used in the proof of Theorem 3 to construct Xa so that B <= S* <= A.
standard math Tanaka's formula and Ito's formula for continuous semimartingales (Revuz-Yor [20]).
Used in the proof of Theorem 2 to handle the local time terms arising from the reflection at the boundaries of the liquidity range.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Liquidity provision of utility indifference type in decentralized exchanges." pith.science (2026). https://pith.science/paper/DNQW3JGD

@misc{pith2026250201931,
  author       = {Pith},
  title        = {Pith review of: Liquidity provision of utility indifference type in decentralized exchanges},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DNQW3JGD}},
  note         = {Machine review of arXiv:2502.01931}
}

read the original abstract

We present a mathematical formulation of liquidity provision in decentralized exchanges. We focus on constant function market makers of utility indifference type, which include constant product market makers with concentrated liquidity as a special case. First, we examine no-arbitrage conditions for a liquidity pool and compute an optimal arbitrage strategy when there is an external liquid market. Second, we show that liquidity provision suffers from impermanent loss unless a transaction fee is levied under the general framework with concentrated liquidity. Third, we establish the well-definedness of arbitrage-free reserve processes of a liquidity pool in continuous-time and show that there is no loss-versus-rebalancing under a nonzero fee if the external market price is continuous. We then argue that liquidity provision by multiple liquidity providers can be understood as liquidity provision by a representative liquidity provider, meaning that the analysis boils down to that for a single liquidity provider. Last, but not least, we give an answer to the fundamental question in which sense the very construction of constant function market makers with concentrated liquidity in the popular platform Uniswap v3 is optimal.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Optimal Dynamic Fees in Automated Market Makers
q-fin.TR 2025-06 conditional novelty 6.0 of 10

In a constant-function market maker, optimal dynamic fees balance arbitrage deterrence against noise-trader attraction, and a fee that is linear in inventory and external price is a near-optimal approximation.

Reference graph

Works this paper leans on

21 extracted references · 18 canonical work pages · cited by 1 Pith paper

[1]

and Chitra, T.: Replicating marke t makers, Digital Finance 5, 367–387 (2023)

Angeris, G., Evans, A. and Chitra, T.: Replicating marke t makers, Digital Finance 5, 367–387 (2023)

work page 2023
[2]

Inf-convolution of risk measu res and optimal risk transfer, Finance Stochast

Barrieu, P ., El Karoui, N. Inf-convolution of risk measu res and optimal risk transfer, Finance Stochast. 9, 269–298 (2005)

work page 2005
[3]

and Feinstein, Z: Axioms for Automated Marke t Makers: A Mathematical Framework in FinT ech and Decentralized Finan ce (2024), arXiv:2210.01227v3

Bichuch, M. and Feinstein, Z: Axioms for Automated Marke t Makers: A Mathematical Framework in FinT ech and Decentralized Finan ce (2024), arXiv:2210.01227v3

arXiv 2024
[4]

3 (2024)10.1137 /23M1602103

Cartea, ´A., Drissi, F ., and Monga, M.: Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Prov ision, SIAM Journal on Financial Mathematics 15, Iss. 3 (2024)10.1137 /23M1602103

work page 2024
[5]

Cartea, ´A., Drissi, F ., and Monga, M.: Predictable Losses of Liquidi ty Pro- vision in Constant Function Markets and Concentrated Liqui dity Markets, Applied Mathematical Finance, 30:2, 69-93 (2023)

work page 2023
[6]

Financial Cryptography and Data Security

Cohen, S., Sabat´ e-Vidales, M., Siska, D., and Szpruch, L., In: Essex, A., et al. Financial Cryptography and Data Security. FC 2023 Internat ional Workshops, FC 2023. Lecture Notes in Computer Science, vol 13953. Sprin ger, Cham (2024)

work page 2024
[7]

Evans, A.: Liquidity provider returns in geometric mean markets, Cryptoe- conomic Systems, 1(2) (2021). 15

work page 2021
[8]

Financial Cryptography and Data Se- curity

Evans, A., Angeris, G., and Chitra, T.: Optimal Fees for G eometric Mean Market Makers, In: Bernhard, M., et al. Financial Cryptography and Data Se- curity. FC 2021 International Workshops, FC2021. Lecture Notes in Computer Science, vol 12676. Springer, Berlin, Heidelberg (2021)

work page 2021

Show all 21 references

[9]

Fukasawa, M., Maire, B., and Wunsch, M.: W eighted varian ce swaps hedge against Impermanent Loss (2023), Quantitative Finance 23:6, 901-911

2023
[10]

Fukasawa, M., Maire, B., and Wunsch, M.: Model-free Hed ging of Imper- manent Loss in Geometric Mean Market Makers (2023), to appear in Applied Mathematical Finance

2023
[11]

and Canidio, A.: Measuring arbitrage losse s and proﬁtability of AMM iquidity (2024), WWW ’24: Companion Proceedings of the ACM Web Conference 2024, 1761-1767

Fritsch, R. and Canidio, A.: Measuring arbitrage losse s and proﬁtability of AMM iquidity (2024), WWW ’24: Companion Proceedings of the ACM Web Conference 2024, 1761-1767

2024
[12]

Kruk, L., Lehoczky , J., Ramanan, K., and Shreve, S.: An e xplicit formula for the Skorokhod map on [0, a]. Ann. Probab. 35 (2007), no. 5, 1740–1768

2007
[13]

and W ang

Lee, C., Tung, S. and W ang. T.: Growth rate of liquidity p rovider’s wealth in G3Ms (2024), arXiv:2403.18177

2024 arXiv
[14]

and Treccani, A.: Blockchain And Distributed Ledgers: Mathemat- ics, T echnology, And Economics, 2021, WSPC

Lipton, A. and Treccani, A.: Blockchain And Distributed Ledgers: Mathemat- ics, T echnology, And Economics, 2021, WSPC

2021
[15]

Maire, B., and Wunsch, M.: Market Neutral Liquidity Pro vision, Ledger, 9 (2024)

2024
[16]

Milionis, J., Moallemi, C.C., Roughgarden, T., and Zha ng, A.L.: Automated Market Making and Loss-V ersus-Rebalancin g (2022), arXiv:2208.06046v2

2022 arXiv
[17]

Milionis, J., Moallemi, C.C., and Roughgarden, T.: Aut omated Mar- ket Making and Arbitrage Proﬁts in the Presence of Fees (2023 ), arXiv:2305.14604

2023 arXiv
[18]

Financial Innovation 8 (2022), no

Mohan, V ., Automated market makers and decentralized e xchanges: a DeFi primer. Financial Innovation 8 (2022), no. 20

2022
[19]

and Y en, J.: An arbi trage driven price dynamics of Automated Market Makers in the presence of fees

Najnudel, J., Tung, S., Y amazaki, K. and Y en, J.: An arbi trage driven price dynamics of Automated Market Makers in the presence of fees. Frontiers of Mathematical Finance 3(4): 560-571 (2024)

2024
[20]

and Y or, M: Continuous Martingales and Brownian Motion , 1999, Springer

Revuz, D. and Y or, M: Continuous Martingales and Brownian Motion , 1999, Springer

1999
[21]

and W ang, T

Tung, S. and W ang, T. A mathematical framework for model ling CLMM dynamics in continuous-time, preprint. 16 A Proofs A.1 Proof of Lemma 1 Note that Xt = 0 can happen only if x∗> 0 since f (0+) = ∞. In this case, 0 ∈ Ξ and r = s(0). If Xt = 0, then Yt = f∗(Xt) > 0 and r = s(0...

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

and Chitra, T.: Replicating marke t makers, Digital Finance 5, 367–387 (2023)

Angeris, G., Evans, A. and Chitra, T.: Replicating marke t makers, Digital Finance 5, 367–387 (2023)

work page 2023

[2] [2]

Inf-convolution of risk measu res and optimal risk transfer, Finance Stochast

Barrieu, P ., El Karoui, N. Inf-convolution of risk measu res and optimal risk transfer, Finance Stochast. 9, 269–298 (2005)

work page 2005

[3] [3]

and Feinstein, Z: Axioms for Automated Marke t Makers: A Mathematical Framework in FinT ech and Decentralized Finan ce (2024), arXiv:2210.01227v3

Bichuch, M. and Feinstein, Z: Axioms for Automated Marke t Makers: A Mathematical Framework in FinT ech and Decentralized Finan ce (2024), arXiv:2210.01227v3

arXiv 2024

[4] [4]

3 (2024)10.1137 /23M1602103

Cartea, ´A., Drissi, F ., and Monga, M.: Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Prov ision, SIAM Journal on Financial Mathematics 15, Iss. 3 (2024)10.1137 /23M1602103

work page 2024

[5] [5]

Cartea, ´A., Drissi, F ., and Monga, M.: Predictable Losses of Liquidi ty Pro- vision in Constant Function Markets and Concentrated Liqui dity Markets, Applied Mathematical Finance, 30:2, 69-93 (2023)

work page 2023

[6] [6]

Financial Cryptography and Data Security

Cohen, S., Sabat´ e-Vidales, M., Siska, D., and Szpruch, L., In: Essex, A., et al. Financial Cryptography and Data Security. FC 2023 Internat ional Workshops, FC 2023. Lecture Notes in Computer Science, vol 13953. Sprin ger, Cham (2024)

work page 2024

[7] [7]

Evans, A.: Liquidity provider returns in geometric mean markets, Cryptoe- conomic Systems, 1(2) (2021). 15

work page 2021

[8] [8]

Financial Cryptography and Data Se- curity

Evans, A., Angeris, G., and Chitra, T.: Optimal Fees for G eometric Mean Market Makers, In: Bernhard, M., et al. Financial Cryptography and Data Se- curity. FC 2021 International Workshops, FC2021. Lecture Notes in Computer Science, vol 12676. Springer, Berlin, Heidelberg (2021)

work page 2021