{"paper_id":"2409.05144","title":"QuantFactor REINFORCE: Mining Steady Formulaic Alpha Factors with Variance-bounded REINFORCE","authors":["Junjie Zhao","Chengxi Zhang","Min Qin","Peng Yang"],"paper_date":"2024-09-08","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2409.05144v3","page_url":"https://thequant.space/flowcharts/quantfactor-reinforce-mining-steady-formulaic-alpha-factors/","scores":{"empirical_rigor":6,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes QuantFactor REINFORCE (QFR), a new reinforcement learning algorithm based on REINFORCE for mining formulaic alpha factors.","quote":"Based on the above discussions, this paper proposes a new RL-based method for formulaic alpha factors mining, named QuantFactor REINFORCE (QFR).","summary":"Alpha factor mining aims to discover investment signals from the historical financial market data, which can be used to predict asset returns and gain excess profits. Powerful deep learning methods for alpha factor mining lack interpretability, making them unacceptable in the risk-sensitive real…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The underlying MDP has deterministic transitions following a Dirac distribution.","type":"model","stated":"explicit","quote":"we observe that the underlying MDPs has deterministic transitions, which satisfies the Dirac distribution."},{"assumption":"Reward is only obtained after completing a full trajectory.","type":"model","stated":"explicit","quote":"trajectory feedback, i.e., non-zero reward can only be obtained after completing a full trajectory"},{"assumption":"The transition model is unbiased.","type":"statistical","stated":"explicit","quote":"Assuming the transition model is unbiased, we have"},{"assumption":"Execution price equals expected price (zero slippage) due to TWAP/VWAP usage.","type":"execution","stated":"explicit","quote":"we assume the difference between the expected and actual execution price is zero because daily trading using TWAP/VWAP strategies effectively neutralizes this kind of price slippage"},{"assumption":"The selected stock universe (CSI300, CSI500, CSI1000, NDX, DJI, SPX) is representative of diverse market conditions.","type":"market","stated":"implied","quote":"The six indices are constructed from stocks traded on diverse exchanges, including the Shanghai Stock Exchange, the Shenzhen Stock Exchange, the Nasdaq Stock Market, and the New York Stock Exchange."}]},"datasets":{"status":"available","items":[{"name":"Real asset price dataset Y","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":null,"quote":"Input: Real asset price dataset Y, Real asset feature dataset X = {X′l}"},{"name":"CSI300 constituents","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"CSI300 constituents","quote":"The trading strategy is a top-50 strategy. On each trading day, the CSI300 constituents are first sorted according to the mined"},{"name":"Six index constituents (CSI300, CSI500, CSI1000, NDX, DJI, SPX)","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"Various global equity indices","quote":"The six indices are constructed from stocks traded on diverse exchanges, including the Shanghai Stock Exchange, the Shen‑zhen Stock Exchange, the Nasdaq Stock Market, and the New York Stock Exchange."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":null,"detail":null,"quote":"we assume the difference between the expected and actual execution price is zero because daily trading using TWAP/VWAP strategies effectively neutralizes this kind of price slippage","source":"front_matter","rule":"R3","candidates":{"front_matter":"no","evidence_card":"no","evidence_card_quoted":true},"card_detail":null,"card_quote":"we assume the difference between the expected and actual execution price is zero because daily trading using TWAP/VWAP strategies effectively neutralizes this kind of price slippage","meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"QFR boosts correlation with returns by 3.83% and achieves stronger excess returns than prior alpha factor mining methods.","kind":"performance","quote":"Evaluations on real assets data indicate the proposed algorithm boosts correlation with returns by 3.83%, and a stronger ability to obtain excess returns compared to the latest alpha factors mining methods"},{"finding":"QFR outperforms all other algorithms in convergence speed and stability, improving the metric by 3.83% over the best PPO baseline.","kind":"performance","quote":"QFR outperforms all other algorithms in convergence speed and stability and improves by 3.83% on the metric compared to the best-performing PPO algorithm"},{"finding":"In backtesting on CSI300, QFR dominates baseline methods in 9 out of 12 quarters, showing superior returns under varied market conditions.","kind":"performance","quote":"QFR outperforms the baseline methods under most market conditions. Among 12 backtested quarters, the QFR-driven investment strategy failed to dominate in only 3 quarters."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":21,"citations":{"cited_by_in_corpus":9,"by_independent_authors":7,"how_cited":{"used_as_baseline":0,"uses_method_or_data":1,"extends":1,"discusses_findings":0,"mentions":5,"unclassified":2},"independent_papers":[{"paper":{"paper_id":"2610.14959","title":"AlphaRLM: Aligning Language Models for Literature-Grounded Symbolic Factor Discovery","paper_date":"2026-10-10","authors":["Jie Huang","Yuecheng Hong","Yao Chen","Hao Zhong","Linfeng Xiao","Xudong Liu"],"empirical_rigor":8,"page_url":"https://thequant.space/flowcharts/alpharlm-aligning-language-models-for-literature-grounded-symbolic-factor/"},"how":"extends","quote":null},{"paper":{"paper_id":"2602.14670","title":"FactorMiner: A Self-Evolving Agent with Skills and Experience Memory for Financial Alpha Discovery","paper_date":"2026-02-16","authors":["Yanlong Wang","Jian Xu","Hongkang Zhang","Shao-Lun Huang","Danny Dongning Sun","Xiao-Ping Zhang"],"empirical_rigor":8,"page_url":"https://thequant.space/flowcharts/factorminer-a-self-evolving-agent-with-skills-and-experience-memory-for/"},"how":"uses_method_or_data","quote":null}],"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":"2026-11-14T18:02:49Z"},"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/"}