{"paper_id":"2503.05254","title":"Modeling metaorder impact with a Non-Markovian Zero Intelligence model","authors":["Adele Ravagnani","Fabrizio Lillo"],"paper_date":"2025-03-07","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2503.05254v2","page_url":"https://thequant.space/flowcharts/modeling-metaorder-impact-with-a-non-markovian-zero-intellig/","scores":{"empirical_rigor":4,"math_complexity":7.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper introduces the Non‑Markovian Zero Intelligence (NMZI) model, a variant of the Zero Intelligence model where the probability of a limit order being a buy or sell depends on an exponentially weighted mid‑price return.","quote":"We propose a novel explainable model for small tick assets, the Non-Markovian Zero Intelligence, which is a variant of the well-known Zero Intelligence model.","summary":"Devising models of the limit order book that realistically reproduce the market response to exogenous trades is extremely challenging and fundamental in order to test trading strategies. We propose a novel explainable model for small tick assets, the Non-Markovian Zero Intelligence, which is a…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The probability that a limit order is a sell/buy is a function of the exponentially weighted mid‑price return, representing past price dynamics.","type":"model","stated":"explicit","quote":"The main modification is that the probability of limit orders' signs (buy/sell) is not constant but is a function of the exponentially weighted mid-price return, representing the past price dynamics, and can be interpreted as the reaction of traders with reservation prices to the price trend.","support":"full"},{"assumption":"The model focuses on small‑tick assets.","type":"market","stated":"explicit","quote":"We propose a novel explainable model for small tick assets, the Non-Markovian Zero Intelligence, which is a variant of the well-known Zero Intelligence model.","support":"full"},{"assumption":"Two main parameters govern the model: a time‑scale (memory) and the strength of traders' reaction to the price trend.","type":"model","stated":"explicit","quote":"In our model, the reaction of these traders occurs with a time scale that corresponds to the memory we employ in the evaluation of the price dynamics, and its strength is tuned by a parameter that enters the definition of the probability that a limit order is a sell/buy. Throughout the paper, we study how our results and market impact dynamics depend on these two parameters.","support":"full"},{"assumption":"Limit orders, market orders and cancellations follow independent Poisson processes in the baseline ZI model.","type":"model","stated":"explicit","quote":"In the original model, limit orders, market orders and cancellations in the LOB are modeled as three inde-pendent Poisson processes.","support":"full"},{"assumption":"Metaorders are executed at a constant speed (constant trading rate).","type":"execution","stated":"explicit","quote":"As expected, the larger the total volume, the stronger the increase of the mid-price during the execution of the metaorder.","support":"full"},{"assumption":"The simulated order‑book dynamics are calibrated on real LOBSTER data for Tesla on a single day.","type":"data","stated":"explicit","quote":"We use the LOBSTER dataset1 for the small-tick asset Tesla on January 5, 2015, which has a tick-size equal to 0.01$. After the pre-processing2, we end up with 170, 416 events.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"LOBSTER dataset (Tesla, Jan 5 2015)","provider":"LOBSTER","kind":"order-book","access":"public","period":"2015-01-05","universe":"Tesla","quote":"We use the LOBSTER dataset1 for the small-tick asset Tesla on January 5, 2015, which has a tick-size equal to 0.01$. After the pre-processing2, we end up with 170, 416 events.","support":"full","known_id":"lobster"}],"known":[{"id":"lobster","name":"LOBSTER","provider":"LOBSTER","kind":"order-book","access":"commercial","quote":"We use the LOBSTER dataset1 for the small-tick asset Tesla on January 5, 2015, which has a tick-size equal to 0.01$.","mentions":2,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"l2-book"},"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":"The NMZI model predicts a concave mid‑price path during metaorder execution and a reversion after execution, matching empirical observations.","kind":"theoretical","quote":"With numerical simulations and analytical arguments, we show that the model predicts a concave price path during a metaorder execution and to a price reversion after the execution ends, as empirically observed. During the execution of a metaorder, the concavity of the market impact emerges.","support":"full"},{"finding":"The impact coefficient estimated from simulations with metaorder execution is (5.063 ± 0.011) ticks, consistent with the ZI‑only estimate of ≈4.9 ticks.","kind":"statistical","quote":"By fitting the difference between the mid-price at the end and at the beginning of the execution, i.e. mT −m0, as a function of Q and referring to Eq. (2), we obtain a coefficient equal to (5.063±0.011) ticks. This is consistent with ˆk estimated without any execution and the coefficients estimated with a metaorder execution and different trading intervals.","support":"partial"},{"finding":"Larger total metaorder volume leads to a stronger increase of the mid‑price during execution.","kind":"performance","quote":"As expected, the larger the total volume, the stronger the increase of the mid-price during the execution of the metaorder.","support":"full"},{"finding":"Increasing the memory parameter (1/β₂) lengthens the concave regime and enhances the reversion mechanism.","kind":"theoretical","quote":"For fixed values of ∆and α, as 1/β2 increases so do ¯R∗and τ ∗. This means that considering a longer memory in the evaluation of the price trend leads to a longer concave regime and so, to a better reproduction of the behavior of the market impact we observe in real markets.","support":"full"},{"finding":"In the stationary regime, both the immediate impact and the reversion component stabilize to constant values, yielding a linear impact.","kind":"theoretical","quote":"In the stationary regime, the overall child MOs’ impact stabilizes to a stationary value. In the stationary regime, the impact stabilizes to a constant value8.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[{"paper":{"paper_id":"2602.19590","title":"Metaorder modelling and identification from public data","paper_date":"2026-02-23","authors":["Ezra Goliath","Tim Gebbie"],"empirical_rigor":7.5,"page_url":"https://thequant.space/flowcharts/metaorder-modelling-and-identification-from-public-data/"},"question":"Do metaorders exhibit a concave impact trajectory during execution?","this_paper_finding":"The NMZI model predicts a concave mid‑price path during metaorder execution and a reversion after execution, matching empirical observations.","other_paper_finding":"The execution profile of synthetic metaorders is concave, indicating non‑linear impact during execution.","other_paper_quote":"Fig-ure 3 shows that the execution profile of the synthetic metaorders is concave, indicating that the impact is not linear in ϕ.","note":"Both report a concave impact/execution profile; A predicts a concave mid‑price path with post‑execution reversion, B confirms non‑linear concave execution.","confidence":0.644},{"paper":{"paper_id":"2303.08867","title":"A Bayesian theory of market impact","paper_date":"2023-03-15","authors":["Louis Saddier","Matteo Marsili"],"empirical_rigor":4,"page_url":"https://thequant.space/flowcharts/a-bayesian-theory-of-market-impact/"},"question":"What is the magnitude of market impact per metaorder?","this_paper_finding":"The impact coefficient estimated from simulations with metaorder execution is (5.063 ± 0.011) ticks, consistent with the ZI‑only estimate of ≈4.9 ticks.","other_paper_finding":"The impact is independent of the meta‑order execution protocol in the regime t≪ν⁻².","other_paper_quote":"The independence of this law on the execution protocol is a consequence of the fact that the scaling regime of interest t ≪ν−2 lies in a region where the contribution from the meta-order to the order flow is statistically undetectable.","note":"Both find impact ≈5 ticks: A predicts protocol‑independent impact; B measures 5.063±0.011 ticks, matching the ZI‑only ≈4.9‑tick estimate.","confidence":0.516}],"contradicts":[{"paper":{"paper_id":"2303.10043","title":"Optimal liquidation with temporary and permanent price impact, an application to cryptocurrencies","paper_date":"2023-03-17","authors":["Hugo E. Ramirez","Julián Fernando Sanchez"],"empirical_rigor":6,"page_url":"https://thequant.space/flowcharts/optimal-liquidation-with-temporary-and-permanent-price-impact-an-application-to/"},"question":"How does the price impact of trades depend on order size?","this_paper_finding":"In the stationary regime, both the immediate impact and the reversion component stabilize to constant values, yielding a linear impact.","other_paper_finding":"Power functional forms fit the calibrated temporary and permanent impact data better than linear forms, with lower residuals and higher R‑squared.","other_paper_quote":"both the total sum of the residuals and the average of the residuals are lower for the power functional form than for the linear form. Likewise, the R2 is higher in all cases for the power functional form","note":"A finds power‑law impact fits better (lower residuals, higher R²) whereas B reports a constant linear impact, showing a disagreement.","confidence":0.467}],"qualifies":[],"compared_with":13,"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":true,"code_url":"https://github.com/adeleravagnani/non-markovian-zero-intelligence-lob-model","note":null},"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":"2026-11-14T19:51:51Z"},"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/"}