{"paper_id":"2508.13174","title":"AlphaEval: A Comprehensive and Efficient Evaluation Framework for Formula Alpha Mining","authors":["Hongjun Ding","Binqi Chen","Jinsheng Huang","Taian Guo","Zhengyang Mao","Guoyi Shao","Lutong Zou","Luchen Liu","Ming Zhang"],"paper_date":"2025-08-10","paper_type":"dataset-benchmark","paper_url":"https://arxiv.org/pdf/2508.13174v1","page_url":"https://thequant.space/flowcharts/alphaeval-a-comprehensive-and-efficient-evaluation-framewor/","scores":{"empirical_rigor":6,"math_complexity":4,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes AlphaEval, a unified, backtest‑free, parallelizable framework for evaluating automated alpha mining models.","quote":"we propose AlphaEval, a unified, parallelizable, and backtest-free evaluation framework","summary":"Formula alpha mining, which generates predictive signals from financial data, is critical for quantitative investment. Although various algorithmic approaches-such as genetic programming, reinforcement learning, and large language models-have significantly expanded the capacity for alpha discovery…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The risk‑free rate is assumed to be zero for simplicity.","type":"market","stated":"explicit","quote":"The risk-free rate is assumed to be zero for simplicity."}]},"datasets":{"status":"available","items":[{"name":"A‑share dataset","provider":"Qlib","kind":"market-prices","access":"public","period":"2010-01-01 – 2024-12-31 (train/valid/tes","universe":null,"quote":"We get the A-share dataset and S&P 500 dataset from Qlib. The time ranges for Train/Valid/Test on A-Share dataset are 2010-01-01 – 2019-12-31/2020-01-01 – 2020-12-31/2021-01-01 – 2024-12-31.","known_id":"qlib"},{"name":"S&P 500 dataset","provider":"Qlib","kind":"market-prices","access":"public","period":"2010-01-01 – 2020-12-31 (train/valid/tes","universe":null,"quote":"The time ranges for Train/Valid/Test on S&P 500 dataset are 2010-01-01 – 2015-12-31/2016-01-01 – 2016-12-31/2017-01-01 – 2020-12-31.","known_id":"qlib"}],"known":[{"id":"qlib","name":"Qlib datasets","provider":"Microsoft","kind":"market-prices","access":"public","quote":"We use both A-share and U.S. stock datasets provided by Qlib as our evaluation benchmarks.","mentions":3,"used_in":2}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"unclear","headline_net_of_costs":null,"detail":null,"quote":null,"source":"conflict","rule":"R4b","candidates":{"front_matter":"no","evidence_card":"n/a","evidence_card_quoted":false},"conflict_kind":"applicability","meaning":"the platform classifier and the paper-text extraction disagree, so no value is asserted"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"AlphaEval scores are highly consistent with precision backtesting outcomes while providing broader diagnostic insight and faster evaluation.","kind":"performance","quote":"The results show that AlphaEval scores are highly consistent with precision backtesting outcomes while offering broader diagnostic insight and significantly faster evaluation."},{"finding":"AlphaEval demonstrates superior alpha selection performance compared to conventional single‑metric filtering approaches such as IC alone.","kind":"performance","quote":"AlphaEval demonstrates superior alpha selection performance compared to conventional single-metric filtering approaches (e.g., by IC alone)."},{"finding":"AlphaEval reduces relative evaluation time by more than 25% thanks to its backtest‑free, parallelizable design.","kind":"performance","quote":"AlphaEval achieves a significant speedup, reducing relative evaluation time by more than 25%."},{"finding":"Relative Rank Entropy (RRE) exhibits a strong negative correlation with annualized turnover, indicating more stable alphas lead to lower trading activity.","kind":"statistical","quote":"RRE exhibits a strong and statistically significant negative linear correlation with turnover."},{"finding":"Alphas with Perturbation Fidelity Score (PFS) ≥ 0.9 show significantly lower maximum drawdown, confirming PFS as a reliable risk‑screening metric.","kind":"statistical","quote":"Alphas with higher Perturbation Fidelity Score (PFS ≥0.9) show significantly lower maximum drawdown (MaxDD), indicating greater robustness to market perturbations."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":29,"citations":{"cited_by_in_corpus":5,"by_independent_authors":3,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":3,"unclassified":2},"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/dhj74/AlphaEval","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":1,"extracted_at":"2026-11-14T17:30:04Z"},"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/"}