{"paper_id":"2309.07708","title":"Market-GAN: Adding Control to Financial Market Data Generation with Semantic Context","authors":["Haochong Xia","Shuo Sun","Xinrun Wang","Bo An"],"paper_date":"2023-09-14","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2309.07708v2","page_url":"https://thequant.space/flowcharts/market-gan-adding-control-to-financial-market-data-generati/","scores":{"empirical_rigor":7.5,"math_complexity":4,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes a Contextual Market Dataset and Market-GAN, a novel controllable generative model for financial market data with semantic context.","quote":"we proposed the Contextual Market Dataset with market dynamics, stock ticker, and history state as context, leveraging a market dynamics modeling method that combines linear regression and Dynamic Time Warping clustering to extract market dynamics; ii) we present Market-GAN, a novel architecture incorporating a Generat","summary":"Financial simulators play an important role in enhancing forecasting accuracy, managing risks, and fostering strategic financial decision-making. Despite the development of financial market simulation methodologies, existing frameworks often struggle with adapting to specialized simulation context.…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The context variables H, d, l are disentangled in the representation space and can be conditioned independently on X.","type":"model","stated":"explicit","quote":"With the assumption that H, d, l are disentangled in the representation space and thus independently conditioned on X, the objective can be represented as p(X|H, d, l) ∝p(H, d, l|X) = p(H|X) · p(d|X) · p(l|X)."},{"assumption":"Financial time‑series features follow the Markov property.","type":"statistical","stated":"implied","quote":"As most downstream machine learning tasks assume that financial time-series features follow the Markov property, our model also learns the auto-regressive transaction distribution p(Xt|X0:t−1)."},{"assumption":"The financial market is non‑stationary, causing context distribution shifts over time.","type":"market","stated":"implied","quote":"Given the non-stationary nature of the financial market, the context distribution p(Ht,ϵ, dt, lt) may shift over time, leading to Out Of Distribution (OOD) problem"}]},"datasets":{"status":"available","items":[{"name":"Contextual Market Dataset","provider":null,"kind":"market-prices","access":"unknown","period":"2000-2023","universe":"Dow Jones Industrial Average (29 stocks)","quote":"We evaluate Market-GAN with the Dow Jones Industrial Average data from 2000 to 2023"},{"name":"Dow Jones Industrial Average daily OHLC","provider":null,"kind":"market-prices","access":"unknown","period":"January 2000 to June 2023","universe":"29 stocks (excluding DOW due to insufficient data) from the DJI index","quote":"We utilized the daily OHLC features of 29 stocks (excluding DOW due to insufficient data) from the DJI index, from January 2000 to June 2023."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"n/a","evidence_card":"unclear","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"forecast-accuracy","findings":[{"finding":"Market‑GAN with C‑TimesBlock achieves the lowest alignment loss Ld (0.023) compared to real data (0.055) and outperforms baselines.","kind":"performance","quote":"... Market‑GAN with C‑TimesBlock outperforms baseline experts on Ld and Ll significantly."},{"finding":"Market‑GAN (CTB) attains the third lowest fidelity loss LD among five methods, balancing alignment and fidelity.","kind":"performance","quote":"Market‑GAN with CTB got the 3rd lowest LD among the 5 compared methods."},{"finding":"Market‑GAN generates data with zero violations of market facts, unlike baseline methods.","kind":"performance","quote":"Market‑GAN guarantees 0 violation of the facts on data with its constrained generation pipeline while baseline methods don’t capture the fact without…"},{"finding":"Overall, Market‑GAN surpasses four state‑of‑the‑art time‑series generative models on the proposed holistic evaluation metrics.","kind":"performance","quote":"showcase superior performance in comparison to 4 state-of-the-art time-series generative models."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":18,"citations":{"cited_by_in_corpus":8,"by_independent_authors":7,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":4,"unclassified":4},"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":"2026-11-14T18:22:12Z"},"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/"}