{"paper_id":"2509.02941","title":"Non-Linear and Meta-Stable Dynamics in Financial Markets: Evidence from High Frequency Crypto Currency Market Makers","authors":["Igor Halperin"],"paper_date":"2025-09-03","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2509.02941v1","page_url":"https://thequant.space/flowcharts/non-linear-and-meta-stable-dynamics-in-financial-markets-ev/","scores":{"empirical_rigor":0,"math_complexity":0,"quadrant":"Philosophers"},"proposes":{"status":"available","statement":"The paper proposes a non‑parametric analysis of high‑frequency cryptocurrency price data using the Kramers‑Moyal expansion to directly estimate drift and diffusion functions and compute the underlying market potential without assuming a specific parametric model.","quote":"Instead of assuming a specific parametric model, we employ a non-parametric approach using the Kramers-Moyal expansion, well known in statistical physics, see e.g. [4]. We find clear evidence of highly non-linear drift functions whose non-linearity gradually weakens with a decreased sampling frequency.","summary":"This work builds upon the long-standing conjecture that linear diffusion models are inadequate for complex market dynamics. Specifically, it provides experimental validation for the author’s prior arguments that realistic market dynamics are governed by higher-order (cubic and higher)…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The drift of the price diffusion depends only on the current price (or log‑price) state variable.","type":"model","stated":"implied","quote":"This framework, motivated by a first-principles analysis of money flows and their price impact, posits that the drift is a non-linear function of a state variable 𝑥such as the current asset price or its log-price. For such time scales the assumption that the drift of price diffusion depends only on the price itself might be too heroic: many other factors (macro-economic, political, etc.) impact the behavior of price","support":"full"},{"assumption":"The stochastic term is Gaussian white noise with zero mean and delta‑correlation.","type":"statistical","stated":"explicit","quote":"The term 𝜉𝑡is a Gaussian white noise process with zero mean and delta-correlation, ⟨𝜉𝑡𝜉𝑡′⟩= 𝛿(𝑡−𝑡′).","support":"full"},{"assumption":"The Langevin equation provides a valid representation of the log‑price dynamics.","type":"model","stated":"explicit","quote":"We model the evolution of the log-price, denoted by 𝑥(𝑡), using the Langevin equation ¤𝑥(𝑡) = 𝜇(𝑥) + 𝜎(𝑥)𝜉𝑡 (1) Here, ¤𝑥(𝑡) is the time derivative of the log-price. The Langevin equation (1) is equivalent to the Itˆo’s stochastic differential equation (SDE) commonly used in finance.","support":"full"},{"assumption":"High‑frequency Uniswap v3 transaction data are sufficiently rich and stationary for non‑parametric Kramers‑Moyal estimation.","type":"data","stated":"implied","quote":"To this end, we turn to the high-frequency world of cryptocurrency markets, which serve as an ideal real-world laboratory for such study. This design creates a highly state-dependent liquidity profile, making it a fertile ground for observing the non-linear dynamics we seek to uncover [5].","support":"full"},{"assumption":"Extreme observations (top/bottom 0.5%) are trimmed to avoid distortion of the estimation.","type":"data","stated":"explicit","quote":"To mitigate the influence of extreme data points that can distort non-parametric estimation, we trim the data by removing the top 0.5% and bottom 0.5% of observations based on their log-price values.","support":"full"},{"assumption":"Potential is obtained by numerical integration of the estimated drift.","type":"model","stated":"explicit","quote":"From the estimated physical drift rate, we then compute the underlying potential 𝑈(𝑥) by numerical integration. The potential 𝑈(𝑥) is computed using Eq.(3) by numerically integrating the estimated physical drift rate 𝜇(𝑥).","support":"full"}]},"datasets":{"status":"available","items":[{"name":"USDC‑WETH log‑price on Arbitrum (2024)","provider":"Uniswap v3 (Arbitrum network)","kind":"crypto-onchain","access":"public","period":"01‑01‑2024 to 12‑31‑2024","universe":"USDC‑WETH pair","quote":"The first dataset contains the history of the USDC-WETH log-price on Arbitrum from 01-01-2024 to 12-31-2024, with about 300k transaction records.","support":"full"},{"name":"USDC‑WETH log‑price on Arbitrum (2025 H1)","provider":"Uniswap v3 (Arbitrum network)","kind":"crypto-onchain","access":"public","period":"01‑01‑2025 to 06‑30‑2025","universe":"USDC‑WETH pair","quote":"The second dataset is for the same USDC-WETH pair for the period 01-01-2025 to 06-30-2025.","support":"full"},{"name":"WBTC‑WETH log‑price on Arbitrum (2024)","provider":"Uniswap v3 (Arbitrum network)","kind":"crypto-onchain","access":"public","period":"01‑01‑2024 to 12‑31‑2024","universe":"WBTC‑WETH pair","quote":"The third dataset is for the WBTC-WETH pair for the period 01-01-2024 to 12-31-2024.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"unclear"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"Strong non‑linearity of the drift is observed, leading to a double‑well potential for the USDC‑WETH pool in two‑month windows for most periods.","kind":"theoretical","quote":"The most striking feature of these graphs is a strong non-linearity of the drift (the left column) which translates into a double-well potential 𝑈(𝑥) for two out of three time periods analyzed in these plots (see the right column in Fig. 1).","support":"full"},{"finding":"When using shorter (one‑month) windows, the double‑well disappears and a single‑well, non‑harmonic potential is obtained.","kind":"theoretical","quote":"One can see that the double well potential observed for longer time windows is gone and replaced by a (non-harmonic) single-well potential.","support":"full"},{"finding":"The non‑linearity of the drift weakens as the sampling frequency decreases.","kind":"theoretical","quote":"We find clear evidence of highly non-linear drift functions whose non-linearity gradually weakens with a decreased sampling frequency.","support":"full"},{"finding":"The WBTC‑WETH pool exhibits single‑well but still non‑quadratic potentials with a very wide basin, indicating strong non‑linearity.","kind":"theoretical","quote":"However, in all cases for this dataset, the potential has a very wide basin, suggesting a strong non-linearity.","support":"full"},{"finding":"Overall, the analysis provides direct, model‑free experimental evidence supporting the hypothesis of non‑linear market dynamics governed by higher‑order drift terms.","kind":"theoretical","quote":"The results presented in this work offer a direct, model-free confirmation of our central hypothesis. 10 5 Conclusion This work presented a direct, non-parametric analysis of high-frequency cryptocurrency price data, pro-viding strong experimental evidence for the theories of non-linear market dynamics previously proposed by the author.","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":24,"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":"2026-11-14T19:48:16Z"},"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/"}