{"paper_id":"2505.14420","title":"SAE-FiRE: Enhancing Earnings Surprise Predictions Through Sparse Autoencoder Feature Selection","authors":["Huopu Zhang","Yanguang Liu","Miao Zhang","Zirui He","Mengnan Du"],"paper_date":"2025-05-20","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2505.14420v2","page_url":"https://thequant.space/flowcharts/sae-fire-enhancing-earnings-surprise-predictions-through-sp/","scores":{"empirical_rigor":6.5,"math_complexity":7,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes the SAE‑FiRE framework, a sparse autoencoder‑based feature selection method for earnings surprise prediction from financial documents.","quote":"We propose the SAE‑FiRE framework for earnings surprise prediction that effectively suppresses feature noise in financial documents.","summary":"Predicting earnings surprises from financial documents, such as earnings conference calls, regulatory filings, and financial news, has become increasingly important in financial economics. However, these financial documents present significant analytical challenges, typically containing over 5,000…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Earnings surprise direction is predicted as a binary supervised learning task using only the textual content of the most recent financial documents.","type":"data","stated":"explicit","quote":"We frame the prediction of the direction of the next earnings surprise (ES) as a supervised learning task using only the textual content of the most recent financial documents."},{"assumption":"Financial documents are truncated to at most 20,000 tokens for activation extraction.","type":"execution","stated":"explicit","quote":"For every financial document (truncated to T = 20,000 tokens), we extract each token’s residual‑stream activations"},{"assumption":"All experiments use a fixed random seed of 42 for reproducibility.","type":"execution","stated":"explicit","quote":"all random seeds were fixed to 42 for reproducibility."},{"assumption":"Baseline models (logistic regression, XGBoost, MLP) are trained with standard hyper‑parameters as implemented in scikit‑learn and PyTorch.","type":"model","stated":"explicit","quote":"regression and XGBoost baselines are trained with scikit‑learn and XGBoost, while MLPs employ Py‑Torch with three hidden layers of size 256 and dropout of 0.5 and ReLU activations."},{"assumption":"The binary classification setup simplifies the nuanced nature of earnings surprises that vary in magnitude.","type":"statistical","stated":"explicit","quote":"our experimental setup focuses on binary classification, which simplifies the nuanced nature of earnings surprises that often vary in magnitude and impact."}]},"datasets":{"status":"available","items":[{"name":"Conference Call transcripts","provider":null,"kind":"text-news","access":"unknown","period":null,"universe":null,"quote":"We employ three main datasets in our analysis: (i) Conference Call, (ii) 10‑Q Report, and (iii) FNSPID Nasdaq News."},{"name":"10‑Q Report","provider":null,"kind":"text-news","access":"unknown","period":null,"universe":null,"quote":"We employ three main datasets in our analysis: (i) Conference Call, (ii) 10‑Q Report, and (iii) FNSPID Nasdaq News."},{"name":"FNSPID Nasdaq News","provider":null,"kind":"text-news","access":"unknown","period":null,"universe":null,"quote":"We employ three main datasets in our analysis: (i) Conference Call, (ii) 10‑Q Report, and (iii) FNSPID Nasdaq News.","known_id":"fnspid"}],"known":[{"id":"fnspid","name":"FNSPID","provider":"Dong et al.","kind":"text-news","access":"public","quote":"We employ three main datasets in our analysis: (i) Conference Call, (ii) 10 -Q Report, and (iii) FNSPID Nasdaq News.","mentions":6,"used_in":1}],"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":"forecast-accuracy","findings":[{"finding":"SAE‑FiRE models achieve the highest F1‑scores and AUC values across all three datasets, e.g., on Conference Calls F1‑scores of 0.759, 0.757, 0.743 and AUC of 0.673, 0.668, 0.657.","kind":"performance","quote":"On Conference Calls, the three SAE‑FiRE models achieve the highest F1‑scores (0.759, 0.757, 0.743) and AUC values (0.673, 0.668, 0.657)."},{"finding":"Using top‑k selected SAE features improves performance versus using all 16K features (e.g., 0.743 F1 and 0.657 AUC vs 0.737 F1 and 0.642 AUC with all features).","kind":"predictive-accuracy","quote":"SAE‑FiRE with selective top‑k feature retention achieves 0.743 F1 (0.657 AUC) on Conference Calls using the Gemma 2‑2B variant."},{"finding":"Larger LLM (Gemma 2‑9B 131K) yields better metrics than smaller (Gemma 2‑2B 16K): accuracy 0.801 vs 0.793, AUC 0.668 vs 0.657, weighted F1 0.757 vs 0.743.","kind":"performance","quote":"the model using 9B 131K SAE bringing accuracy from 0.793 to 0.801, AUC from 0.657 to 0.668, and weighted average F1 from 0.743 to 0.757."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":31,"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":1,"extracted_at":"2026-11-14T17:38:08Z"},"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/"}