{"paper_id":"2404.07223","title":"Stock Recommendations for Individual Investors: A Temporal Graph Network Approach with Mean-Variance Efficient Sampling","authors":["Youngbin Lee","Yejin Kim","Javier Sanz-Cruzado","Richard McCreadie","Yongjae Lee"],"paper_date":"2024-03-27","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2404.07223v3","page_url":"https://thequant.space/flowcharts/stock-recommendations-for-individual-investors-a-temporal-g/","scores":{"empirical_rigor":6,"math_complexity":6.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a new model called Portfolio Temporal Graph Network Recommender (PfoTGNRec) for stock recommendation that incorporates individual preferences, diversification, and temporal dynamics.","quote":"In response, we propose a new model, Portfolio Temporal Graph Network Recommender PfoTGNRec, which can handle time-varying collaborative signals and incorporates diversification-enhancing sampling.","summary":"Recommender systems can be helpful for individuals to make well-informed decisions in complex financial markets. While many studies have focused on predicting stock prices, even advanced models fall short of accurately forecasting them. Additionally, previous studies indicate that individual…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Daily adjusted closing prices are used as temporal features for stocks.","type":"data","stated":"explicit","quote":"We use daily adjusted closing prices for the temporal features of stocks, retrieved from an open source Python package yfinance."},{"assumption":"Abnormal transactions and stocks with highly unstable price movements are removed from the dataset.","type":"data","stated":"explicit","quote":"To exclude abnormal transactions, we remove stocks with highly unstable price movements."},{"assumption":"Stocks halted in trading (no price changes for 30 consecutive days) are excluded.","type":"data","stated":"explicit","quote":"to eliminate stocks that have been halted in trading, we remove stocks with no price changes for 30 consecutive days."},{"assumption":"The dataset is split chronologically with an 8:1:1 ratio preserving temporal order.","type":"model","stated":"explicit","quote":"For the data split, we utilize a chronological approach based on interaction timestamps to partition the dataset into training, validation, and testing sets. This division follows a ratio of 8:1:1, which preserves the temporal order of interactions."}]},"datasets":{"status":"available","items":[{"name":"Individual investor transaction dataset","provider":"National Bank of Greece","kind":"other","access":"proprietary","period":"January 2018 to November 2022","universe":"92 stocks, 8,337 users","quote":"We conduct experiments using individual investor transaction dataset, provided by National Bank of Greece [29]. This dataset includes real transaction data of users and represents a snapshot of the Greek market. The data spans from January 2018 to November 2022 comprising user buy orders during this period."}],"known":[{"id":"yahoo-finance","name":"Yahoo Finance","provider":"Yahoo","kind":"market-prices","access":"public","quote":"We use daily adjusted closing prices for the temporal features of stocks, retrieved from an open source Python package yfinance.","mentions":3,"used_in":1}],"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":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":"PfoTGNRec achieves substantially higher recommendation accuracy than static models, e.g., NDCG@5 of 0.4535 versus 0.0416 for the best static baseline.","kind":"performance","quote":"the best-performing static model, BPR, achieves an NDCG@5 of 0.0416. In contrast, the best-performing dynamic model, TGN, achieves an NDCG@5 of 0.4535."},{"finding":"The proposed model demonstrates superior overall performance compared to state-of-the-art baselines on real-world individual trading data.","kind":"performance","quote":"On real-world individual trading data, our approach demonstrates superior performance compared to state-of-the-art baselines, including cutting-edge dynamic embedding models and existing stock recommendation models."},{"finding":"The hyperparameter λMV has the most significant impact on both recommendation and portfolio metrics, decreasing NDCG@5 sharply as it increases.","kind":"performance","quote":"has the most significant impact on both metrics. NDCG@5 decreases sharply with increasing"}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"available","supports":[],"contradicts":[],"qualifies":[],"compared_with":16,"citations":{"cited_by_in_corpus":2,"by_independent_authors":2,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":1,"mentions":0,"unclassified":1},"independent_papers":[{"paper":{"paper_id":"2510.15993","title":"Aligning Language Models with Investor and Market Behavior for Financial Recommendations","paper_date":"2025-10-14","authors":["Fernando Spadea","Oshani Seneviratne"],"empirical_rigor":7,"page_url":"https://thequant.space/flowcharts/aligning-language-models-with-investor-and-market-behavior-f/"},"how":"discusses_findings","quote":"As highlighted by [ref] and [CITED] , individuals frequently disregard theoretically optimal financial advice if it conflicts with their personal preferences"}],"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/youngandbin/PfoTGNRec","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":"2026-11-14T18:12:19Z"},"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/"}