{"paper_id":"2404.09090","title":"DEX Specs: A Mean Field Approach to DeFi Currency Exchanges","authors":["Erhan Bayraktar","Asaf Cohen","April Nellis"],"paper_date":"2024-04-13","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2404.09090v1","page_url":"https://thequant.space/flowcharts/dex-specs-a-mean-field-approach-to-defi-currency-exchanges/","scores":{"empirical_rigor":7,"math_complexity":7.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a mean‑field game model of heterogeneous liquidity providers in Uniswap v3 and extends it with a Stackelberg game that includes MEV bots performing just‑in‑time liquidity attacks.","quote":"We investigate the behavior of liquidity providers (LPs) by modeling a decentralized cryptocurrency exchange (DEX) based on Uniswap v3","summary":"We investigate the behavior of liquidity providers (LPs) by modeling a decentralized cryptocurrency exchange (DEX) based on Uniswap v3. LPs with heterogeneous characteristics choose optimal liquidity positions subject to uncertainty regarding the size of exogenous incoming transactions and the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Each LP has a fixed amount of capital (valued in token B) that it can contribute to the pool.","type":"model","stated":"explicit","quote":"Furthermore, we will assume that each LP has a fixed amount of capital (as valued in terms of token B) that it is able to contribute to the pool."},{"assumption":"LPs value their capital using the market exchange rate m* which is assumed impervious to manipulation.","type":"model","stated":"explicit","quote":"We assume that LPs value their total capital using the market exchange rate, which we denote as m*, because it is impervious to manipulation from within the pool."},{"assumption":"LPs have a belief parameter δ about future market exchange rate movements and model m* dynamics with a geometric Brownian motion.","type":"statistical","stated":"explicit","quote":"we assume that, based on δ, the LP believes that the market exchange rate will evolve according to m*_t = m*_{t-1} e(α+δ/1000−σ^2)Δt+σΔW_t."},{"assumption":"The distribution of LP types Θ is separable, complete, atomless, and each type is drawn independently.","type":"model","stated":"explicit","quote":"Assumption 3.1. The space of characteristics/types, supp(Θ), is a separable and complete metric space. We also impose that Θ is atomless."},{"assumption":"The time horizon for the LP optimization problem is fixed at T = 2700 blocks.","type":"model","stated":"explicit","quote":"We will also hold the time horizon of the problem fixed for all players at T = 2700."},{"assumption":"Bots consider a fixed gas fee G in their profit calculation, while LPs do not.","type":"execution","stated":"explicit","quote":"Note that we choose to have Bots to consider gas fees (G, which we set as fixed for now) while LPs do not."}]},"datasets":{"status":"available","items":[{"name":"Etherscan transaction data","provider":"Etherscan","kind":"crypto-onchain","access":"public","period":null,"universe":null,"quote":"All transaction data is obtained from the Etherscan API...","known_id":"etherscan"},{"name":"Coinmetrics market exchange rates","provider":"Coinmetrics","kind":"market-prices","access":"public","period":null,"universe":null,"quote":"exchange rate data was obtained from the Coinmetrics API..."},{"name":"GitHub code and datasets","provider":"GitHub","kind":"synthetic","access":"public","period":null,"universe":null,"quote":"Our code and the necessary datasets can be found on Github."}],"known":[{"id":"etherscan","name":"Etherscan","provider":"Etherscan","kind":"crypto-onchain","access":"public","quote":"1.5 Data All transaction data is obtained from the Etherscan API and exchange rate data was obtained from the Coinmetrics API and the CoinGecko API.","mentions":2,"used_in":2},{"id":"coingecko","name":"CoinGecko","provider":"CoinGecko","kind":"market-prices","access":"public","quote":"1.5 Data All transaction data is obtained from the Etherscan API and exchange rate data was obtained from the Coinmetrics API and the CoinGecko API.","mentions":1,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"unclear"},"transaction_costs":{"status":"available","modeled":"unclear","headline_net_of_costs":null,"detail":null,"quote":null,"source":"conflict","rule":"R4b","candidates":{"front_matter":"n/a","evidence_card":"partial","evidence_card_quoted":true},"conflict_kind":"applicability","card_detail":"Bots incur a fixed gas fee G (average $10‑$30 per swap) while LPs are assumed not to pay gas fees.","card_quote":"the average fee for a Uniswap currency swap or liquidity adjustment hovers between $10 -$30","meaning":"the platform classifier and the paper-text extraction disagree, so no value is asserted"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"The calibrated mean‑field equilibrium reproduces observed pool liquidity and exchange‑rate dynamics.","kind":"predictive-accuracy","quote":"the equilibrium strategy resulting from this mean-field game produces pool exchange rate dynamics and liquidity evolution consistent with observed pool behavior."},{"finding":"When LPs anticipate bot attacks, their collective profits increase by approximately 3%.","kind":"performance","quote":"when LPs choose their liquidity positions while anticipating Bot attacks, they are able to increase their collective profits by approximately 3%."},{"finding":"Including bots improves the MAPE between simulated pool and market exchange rates from 0.1357% to 0.0732%, close to the observed 0.063%.","kind":"predictive-accuracy","quote":"the MAPE between the simulated pool exchange rate and the market exchange rate improved from the previous value of 0.1357% to 0.0732%, which is very close to the observed MAPE"},{"finding":"Risk‑averse LPs concentrating liquidity in anticipation of bots reduce overall MAPE by 46.1% across simulations.","kind":"performance","quote":"This led to a 21.54% increase in the accuracy of the predicted liquidity distribution and a 46.1% decrease in the overall MAPE between the market and pool exchange rates throughout our simulations."}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":17,"citations":{"cited_by_in_corpus":2,"by_independent_authors":2,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":2,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":1,"extracted_at":"2026-11-14T18:11:22Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}