Decentralised exchanges nearly doubled their share of spot crypto trading volume between 2024 and 2026, and Uniswap is the largest of them, accounting for 37.9% of the volume of the ten largest decentralized exchanges in the first quarter of 2026. The object of this thesis is its v3 protocol, which lets liquidity providers choose the price range in which to allocate their capital and aggregates single positions into a profile of liquidity over prices. Most of the existing work on liquidity provision in CFMMs is theoretical, while computational studies and empirical evidence are still at an early stage. This thesis implements from scratch the framework for concentrated liquidity introduced by Risk, Tung, and Wang, and extends their analysis to a panel of eleven Ethereum pools over five token pairs and three fee tiers. A static option study on one WETH/USDC snapshot reproduces their pricing of impermanent loss, while the temporal evolution of the liquidity profile is studied through functional principal component analysis. A spectral analysis shows that the profiles are low dimensional but not stable. In most windows across the pools examined, three or four modes explain more than 85% of the variance. The effective rank, the total variance and the mode shapes still change from window to window, and each pool changes in its own way. Most of this change falls in windows that start between March and October 2025. Following the aforementioned authors' work, a shape statistic was then introduced to test whether the observed liquidity profiles match the power law profile implied by a constant elasticity of variance price process. The statistic tests the shape of the profile, not the price process. Near spot the estimates are too noisy to test this shape; away from spot the median exponent settles near a constant on each side, with opposite signs above and below spot. Additionally, a new coverage diagnostic is created in order to relate the support of the spectral analysis to the share of the liquidity distribution it implicitly samples. The processed panel, the code and the results are available in the public GitHub repository, along with a reproducibility guide.

A Geometric and Option-Theoretic Characterisation of Liquidity in Automated Market Makers

COTTINI, GIORGIO
2025/2026

Abstract

Decentralised exchanges nearly doubled their share of spot crypto trading volume between 2024 and 2026, and Uniswap is the largest of them, accounting for 37.9% of the volume of the ten largest decentralized exchanges in the first quarter of 2026. The object of this thesis is its v3 protocol, which lets liquidity providers choose the price range in which to allocate their capital and aggregates single positions into a profile of liquidity over prices. Most of the existing work on liquidity provision in CFMMs is theoretical, while computational studies and empirical evidence are still at an early stage. This thesis implements from scratch the framework for concentrated liquidity introduced by Risk, Tung, and Wang, and extends their analysis to a panel of eleven Ethereum pools over five token pairs and three fee tiers. A static option study on one WETH/USDC snapshot reproduces their pricing of impermanent loss, while the temporal evolution of the liquidity profile is studied through functional principal component analysis. A spectral analysis shows that the profiles are low dimensional but not stable. In most windows across the pools examined, three or four modes explain more than 85% of the variance. The effective rank, the total variance and the mode shapes still change from window to window, and each pool changes in its own way. Most of this change falls in windows that start between March and October 2025. Following the aforementioned authors' work, a shape statistic was then introduced to test whether the observed liquidity profiles match the power law profile implied by a constant elasticity of variance price process. The statistic tests the shape of the profile, not the price process. Near spot the estimates are too noisy to test this shape; away from spot the median exponent settles near a constant on each side, with opposite signs above and below spot. Additionally, a new coverage diagnostic is created in order to relate the support of the spectral analysis to the share of the liquidity distribution it implicitly samples. The processed panel, the code and the results are available in the public GitHub repository, along with a reproducibility guide.
2025
A Geometric and Option-Theoretic Characterisation of Liquidity in Automated Market Makers
AMM
CLMM
Liquidity
File in questo prodotto:
File Dimensione Formato  
Cottini_Giorgio.pdf

accesso aperto

Dimensione 8.62 MB
Formato Adobe PDF
8.62 MB Adobe PDF Visualizza/Apri

The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/20.500.12608/115042