Decentralised Finance and Automated Market Making: Predictable Loss and Optimal Liquidity Provision

DeFi: Data-Driven Characterisation of Uniswap V3 Ecosystem & an Ideal Crypto Law for Liquidity Pools

“…Each version of Uniswap has its own dedicated subgraph, which has a precise endpoint for querying data and a database schema to expose the available fields. Our terminology follows the Uniswap v3 schema 6 and we download pools data via the related subgraph. Our first aim is to identify the pools most representative of the Uniswap ecosystem, which we interpret as having significant liquidity consumption and provision events, but also showing high interconnectedness.…”

“…Flows of liquidity between Uniswap v2 pools are studied in [17], while [18] shows the difficulty of earning significant returns by providing liquidity in Uniswap v3. Interestingly, [6] fully characterises the wealth of LTs and LPs in Uniswap v3, and shows that LPs suffer a "Predictable Loss".…”

“…However, concentrated liquidity also implies that LPs collect LTs' fees only while the exchange rate of executed trades lies between two ticks over which they are indeed providing liquidity. For a full characterisation of the wealth of LTs and LPs in Uniswap v3, we point to the mathematical analyses pursued in [6,7].…”

Inefficiency of CFMs: hedging perspective and agent-based simulations

“…It has been shown empirically [7] and experimentally [19] that liquidity providers (LPs) in constant function markets (CFMs) (in particular, Uniswap v3) lose money on average. Indeed, from e.g.…”

“…Indeed, from e.g. [7,Table 2], the average LP transaction in the ETH/USDC pool (from May 2021-August 2022) resulted in a position loss of −1.64%, and fee income of 0.155% of the size of the initial trade, with an average hold time of 6.1 days.…”