{"paper_id":"2509.05107","title":"Painting the market: generative diffusion models for financial limit order book simulation and forecasting","authors":["Alfred Backhouse","Kang Li","Jakob Foerster","Anisoara Calinescu","Stefan Zohren"],"paper_date":"2025-09-05","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2509.05107v1","page_url":"https://thequant.space/flowcharts/painting-the-market-generative-diffusion-models-for-financi/","scores":{"empirical_rigor":9,"math_complexity":8,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a diffusion-based generative model with inpainting that converts limit order book data into a structured image format to generate future LOB states.","quote":"We introduce a novel approach, converting LOB data into a structured image format, and applying diffusion models with inpainting to generate future LOB states.","summary":"Simulating limit order books (LOBs) has important applications across forecasting and backtesting for financial market data. However, deep generative models struggle in this context due to the high noise and complexity of the data. Previous work uses autoregressive models, although these experience…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The model uses only Level‑2 order‑book data and cannot use Level‑3 message data.","type":"data","stated":"explicit","quote":"Unlike LOBS5, our model cannot use Level-3 message data and considers only Level-2 order book data."},{"assumption":"Only the top 10 price levels of the order book are needed, as they contain the majority of useful information.","type":"model","stated":"explicit","quote":"They use the top 10 levels of the book, claiming that 80% of the information is contained in the first level of the book and 90% of orders are never filled so more than 10 levels adds unnecessary noise"},{"assumption":"Experiments are limited to regular trading hours.","type":"market","stated":"explicit","quote":"We restrict the data to regular trading hours"}]},"datasets":{"status":"available","items":[{"name":"LOBSTER","provider":"LOBSTER","kind":"order-book","access":"public","period":"1 July 2022 to 30 December 2022 (102 tra","universe":"Alphabet (GOOG) and Intel (INTC) stocks","quote":"We focus on the same examples of a small-tick (Alphabet, GOOG) and large-tick (Intel, INTC) stocks used in LOB-Bench","known_id":"lobster"}],"known":[{"id":"lobster","name":"LOBSTER","provider":"LOBSTER","kind":"order-book","access":"commercial","quote":"5 Experiments 5.1 Data and experimental setup We focus on the same examples of a small-tick (Alphabet, GOOG) and large-tick (Intel, INTC) stocks used in LOB-Bench [27], using the LOBSTER dataset [15].","mentions":1,"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":"forecast-accuracy","findings":[{"finding":"The model achieves state‑of‑the‑art performance on GOOG under Wasserstein loss and is competitive under L1 loss.","kind":"performance","quote":"For GOOG, our model is competitive under L1 loss and achieves state-of-the-art performance under Wasserstein loss."},{"finding":"The model performs significantly better on GOOG than on INTC, likely due to differences in tick size and volatility.","kind":"performance","quote":"Our model performs significantly better on GOOG than INTC, both in perceived visual quality and in comparison to other models later on."},{"finding":"Diffusion models can generate entire LOB sequences in parallel, allowing fast inference even with few sampling steps.","kind":"performance","quote":"diffusion models are able to generate entire LOB sequences in parallel, perhaps even with very few sampling steps, leading to high throughput for longer prediction horizons."},{"finding":"Increasing the number of inference steps yields only marginal performance gains, indicating efficient sampling.","kind":"performance","quote":"sample speed can be increased with minimal performance degradation however the model struggles with robustness and generalisability across different stocks and domains."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":34,"citations":{"cited_by_in_corpus":1,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":1,"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":"2026-11-14T17:27:53Z"},"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/"}