{"paper_id":"2509.04812","title":"Deep Learning for Conditional Asset Pricing Models","authors":["Hongyi Liu"],"paper_date":"2025-09-05","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2509.04812v1","page_url":"https://thequant.space/flowcharts/deep-learning-for-conditional-asset-pricing-models/","scores":{"empirical_rigor":6,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a pseudo‑Siamese Network for Asset Pricing (SNAP) model that uses deep learning to estimate conditional deep alphas, betas and factor risk premia.","quote":"We propose a new pseudo-Siamese Network for Asset Pricing (SNAP) model, based on deep learning approaches, for conditional asset pricing.","summary":"We propose a new pseudo-Siamese Network for Asset Pricing (SNAP) model, based on deep learning approaches, for conditional asset pricing. Our model allows for the deep alpha, deep beta and deep factor risk premia conditional on high dimensional observable information of financial characteristics…","strategy":"pseudo-SNAP model"},"assumptions":{"status":"available","items":[{"assumption":"Firm characteristics for month t are matched with data available at the end of month t‑1.","type":"data","stated":"explicit","quote":"For every specific month t’s return, we match it with firm characteristics available at the end of month t −1 under the assumption given by Green, Hand, and Zhang (2017)."},{"assumption":"Macroeconomic predictors are transformed to be stationary before use.","type":"statistical","stated":"explicit","quote":"We implement the suggested standard transformation code to have stationary macroeconomic series as suggested in McCracken and Ng (2016)."}]},"datasets":{"status":"available","items":[{"name":"CRSP monthly equity returns","provider":"Center for Research in Security Prices (CRSP)","kind":"market-prices","access":"public","period":"January 1970 to December 2019","universe":"All US stocks with delisting returns, common stocks on NYSE, AMEX, NASDAQ","quote":"We collect monthly equity return data for all stocks from the Center for Research in Security Prices (CRSP), including delisting returns","known_id":"crsp"},{"name":"Macro predictors","provider":"FRED‑MD database","kind":"macro","access":"public","period":"January 1970 to December 2019","universe":"124 macroeconomic variables","quote":"In order to capture investors’ expectations about future business cycle conditions from macroeconomic activities, we have 124 macroeconomic predictors in total which are from the FRED‑MD database as defined in McCracken and Ng (2016).","known_id":"fred"}],"known":[{"id":"crsp","name":"CRSP","provider":"Center for Research in Security Prices","kind":"market-prices","access":"commercial","quote":"8 4 Empirical Analysis 4.1 Data We collect monthly equity return data for all stocks from the Center for Research in Security Prices (CRSP), including delisting returns (Shumway and Warther, 1999) from January 1970 to December 2019.","mentions":3,"used_in":3},{"id":"compustat","name":"Compustat","provider":"S&P Global","kind":"fundamentals","access":"commercial","quote":"These characteristics are constructed from CRSP and Compustat database.","mentions":3,"used_in":1}],"sample":{"start":"1970-01","end":"2019-12","market":"U.S. stocks","asset_class":"equities","frequency":"monthly"},"data_resolution":"weekly-monthly"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"no","evidence_card":"no","evidence_card_quoted":false},"meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The pseudo‑SNAP model outperforms benchmark models (FF3, FF5, LASSO, ElasticNet, ridge, feed‑forward network) in out‑of‑sample prediction and achieves an out‑of‑sample Sharpe ratio above 2 with the smallest performance decay.","kind":"performance","quote":"Our pseudo‑SNAP model outperforms the benchmark models ... in terms of out‑of‑sample prediction and out‑of‑sample Sharpe ratio. In particular, our model yields a stellar out‑of‑sample Sharpe ratio above 2 and achieves the smallest performance decay compared to other benchmarks."},{"finding":"The hypothesis that deep mispricing errors (deep alphas) are zero is strongly rejected.","kind":"statistical","quote":"The mispricing testing significantly rejects the null hypothesis that the mispricing errors are zero."},{"finding":"Arbitrage portfolios constructed from deep mispricing errors generate significant alphas across CAPM, Fama‑French 3‑factor and 5‑factor models.","kind":"performance","quote":"We construct alpha‑weighted arbitrage portfolios and find the significant alphas across three classical asset pricing models (CAPM, Fama‑French 3‑factor and 5‑factor model)."},{"finding":"LSTM networks help store long‑term dependency, enhancing out‑of‑sample prediction and yielding the smallest out‑of‑sample performance decay among benchmarks.","kind":"performance","quote":"We find that LSTM network not only helps store the long‑term dependency to enhance out‑of‑sample prediction but also keeps the smallest out‑of‑sample performance decay compared to the benchmark models."}],"reported_performance":{"strategy":"pseudo-SNAP model","sharpe":3.15,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":21,"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":"2026-11-14T17:28:02Z"},"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/"}