{"paper_id":"2502.04097","title":"Impermanent loss and Loss-vs-Rebalancing II","authors":["Abe Alexander","Guillaume Lambert","Lars Fritz"],"paper_date":"2025-02-06","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2502.04097v3","page_url":"https://thequant.space/flowcharts/impermanent-loss-and-loss-vs-rebalancing-ii/","scores":{"empirical_rigor":3,"math_complexity":7.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes to analyze the relationship between impermanent loss (IL) and loss‑versus‑rebalancing (LVR) in automated market makers, identifying three time‑regime behaviors and studying how fees alter these dynamics.","quote":"This paper examines the relationship between impermanent loss (IL) and loss-versus-rebalancing (LVR) in automated market makers (AMMs). We find there are three relevant regimes: (i) very short times where LVR and IL are identical; (ii) intermediate time where LVR and IL show distinct distribution functions but are connected via the central limit theorem exhibiting the same expectation value; (iii) long time behavior","summary":"This paper examines the relationship between impermanent loss (IL) and loss-versus-rebalancing (LVR) in automated market makers (AMMs). Our main focus is on statistical properties, the impact of fees, the role of block times, and, related to the latter, the continuous time limit. We find there are…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Asset prices follow a geometric Brownian motion (GBM) or Brownian motion for analytical convenience.","type":"model","stated":"explicit","quote":"Fur-thermore, we introduce the concept of Brownian motion and geometric Brow-nian motion to discuss the price dynamics of an asset pair. In contrast, Brownian motion can result in negative prices.","support":"full"},{"assumption":"There is an infinite external liquidity source (oracle) that the AMM price tracks with a one‑step lag (block time).","type":"market","stated":"explicit","quote":"We assume that the price of the AMM follows an external infinite liquidity source/oracle at a distance of one time step (block time is a natural candidate for this). Additionally, the arbitrageur does not pay any transaction fees.","support":"full"},{"assumption":"Arbitrageurs do not pay any transaction fees in the baseline arbitrage analysis.","type":"execution","stated":"explicit","quote":"Additionally, the arbitrageur does not pay any transaction fees. This is an unrealistic assumption but for a general modeling we take that approach.","support":"full"},{"assumption":"No additional uninformed trading activity is present.","type":"market","stated":"explicit","quote":"Furthermore, we assume there is no additional uninformed trading activity.","support":"full"},{"assumption":"A natural time scale Δt (e.g., block time) exists for price changes.","type":"statistical","stated":"explicit","quote":"We assume that there is a natural time scale ∆t for a price change, for instance the block time.","support":"full"},{"assumption":"Simulations use discrete GBM updates with volatility σ = 0.001 (or 0.004 in later experiments) and a base fee of 0.02%.","type":"other","stated":"explicit","quote":"The price evolution follows a discrete geometric Brownian motion with volatility σ = 0.001. The base fee is set to 0.02%.","support":"partial"}]},"datasets":{"status":"available","items":[{"name":"Synthetic AMM simulation","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"We have made a simulation of the key metrics, shown in Figure. 12. The AMM is initialized with 1000 tokens of asset A and an equivalent value of asset B at an initial price of 100.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R4a","candidates":{"front_matter":"n/a","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Base fee set to 0.02% (0.0002) applied to trades in the simulation; fees reduce average LVR substantially more than average IL.","card_quote":"The base fee is set to 0.02%. The first is that LVR in absence of fees accounted still follows a relatively narrow distribution around an average value.","meaning":"the paper reports no trading or execution result; the cost the extractor found is a model parameter"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"Three regimes are identified: very short times where IL and LVR are identical, intermediate times where they share the same expectation value but have distinct distributions, and long times where both expectations and distributions differ.","kind":"theoretical","quote":"We find there are three relevant regimes: (i) very short times where LVR and IL are identical; (ii) intermediate time where LVR and IL show distinct distribution functions but are connected via the central limit theorem exhibiting the same expectation value; (iii) long time behavior where both the distribution functions and averages are distinct. 12 IL, on the other hand, is not sensitive to the elementary step","support":"full"},{"finding":"In the intermediate regime, IL and LVR have the same expectation value (⟨IL⟩ ∝ p²σ²T) despite hugely different distribution shapes.","kind":"statistical","quote":"12 IL, on the other hand, is not sensitive to the elementary step within the period ∆t, only to the difference from start to finish, called ∆P = pf −p. Therefore, we can estimate it using the same formula we used for LVR and find that for a period of length T we get ⟨IL(T)⟩∝∆P 2 = p2σ2T , (7) as an expectation value, just like LVR, see Appendix. We find that LVR has a normal shaped distribution while IL has a","support":"full"},{"finding":"Fees reduce the average LVR substantially more than the average IL while having relatively little effect on the shape of the LVR distribution.","kind":"performance","quote":"The first is that LVR in absence of fees accounted still follows a relatively narrow distribution around an average value. However, the average LVR, is largely reduced compared to the case in the absence of fees (more on that later).","support":"full"},{"finding":"LVR distribution tends toward a normal (Gaussian) shape, whereas IL distribution exhibits a power‑law singularity for small IL.","kind":"statistical","quote":"We find that LVR has a normal shaped distribution while IL has a distribu-tion that becomes power-law singular for small IL. However, both distributions have the same average value (shown analytically in the Appendix), and the distribution function of LVR can be constructed from the sum of many values of infinitesimal IL drawn from its distribution function in accordance with the central limit theorem.","support":"full"},{"finding":"For infinitesimal price changes, IL and LVR are mathematically identical, but over longer periods they diverge significantly.","kind":"theoretical","quote":"6 Conclusion This paper has explored the intricate relationship between impermanent loss (IL), loss-versus-rebalancing (LVR), and the role of fees in automated market makers (AMMs). For infinitesimal price changes, IL and LVR are mathematically identical, despite their different interpretations.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":34,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"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":2,"extracted_at":"2026-11-14T19:52:30Z"},"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/"}