{"paper_id":"2403.07998","title":"Pairs Trading Using a Novel Graphical Matching Approach","authors":["Khizar Qureshi","Tauhid Zaman"],"paper_date":"2024-03-12","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2403.07998v1","page_url":"https://thequant.space/flowcharts/pairs-trading-using-a-novel-graphical-matching-approach/","scores":{"empirical_rigor":8,"math_complexity":6,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a matching‑based pair selection method that uses graph‑theoretic maximum weight matchings to choose cointegrated pairs with no shared assets.","quote":"We propose techniques from graph theory such as matchings, to determine subsets of pairs with no common assets.","summary":"Pairs trading, a strategy that capitalizes on price movements of asset pairs driven by similar factors, has gained significant popularity among traders. Common practice involves selecting highly cointegrated pairs to form a portfolio, which often leads to the inclusion of multiple pairs sharing…","strategy":"matching-based pairs trading"},"assumptions":{"status":"available","items":[{"assumption":"All daily log‑normal price changes have the same mean µ1 and noise standard deviation σ1.","type":"model","stated":"explicit","quote":"To further simplify our analysis, we will assume all daily log-normal price changes have the same mean µ1 and noise standard deviation σ1"},{"assumption":"All cointegrated spreads have the same standard deviation σ.","type":"model","stated":"explicit","quote":"and all cointegrated spreads have the same standard deviation σ."},{"assumption":"A unit amount is invested in each pair.","type":"execution","stated":"explicit","quote":"For each portfolio, we assume a unit amount is invested in each pair."},{"assumption":"Covariance is zero between any pairs that do not share a common stock.","type":"statistical","stated":"explicit","quote":"We assume the covariance is zero between any pairs which do not share a common stock."},{"assumption":"Baseline portfolio selects exactly 250 pairs based on p‑values.","type":"model","stated":"explicit","quote":"For the baseline portfolio we assume 250 pairs are selected for trading based on their p-values."},{"assumption":"Matching portfolio has m1 = m2 = 0 (no shared‑stock pairs).","type":"model","stated":"explicit","quote":"The matching portfolio has m1 = m2 = 0 and the baseline portfolio has m1 = 0."}]},"datasets":{"status":"available","items":[{"name":"S&P 500 stock price data","provider":"S&P","kind":"market-prices","access":"public","period":"2010-2023","universe":"500 stocks","quote":"We estimated the mean and variance for the daily change in log‑prices of all stocks on the S&P 500 using price data from 2010 to 2023."},{"name":"Monthly pair‑selection data (lowest p‑value pairs)","provider":"S&P","kind":"market-prices","access":"public","period":"January 2021 to August 2023","universe":"250 pairs with smallest p‑values each month","quote":"We kept the variance for the 250 pairs with the lowest p‑value each month. We then took the median value of this set of variances across all stocks and months in our dataset to obtain σ2."}],"known":[],"sample":{"start":"2017","end":"2023","market":"S&P 500 stocks","asset_class":"equities","frequency":null},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":null,"detail":"Annual transaction cost of 1% per stock, i.e., daily fee of 0.00397% applied to gross returns.","quote":"We model the annual transaction costs for each stock at 1%, translating to a daily fee of 1%/252 days = 0.00397%, consistent with the findings in SEC (2019).","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Annual transaction cost of 1% per stock, i.e., daily fee of 0.00397% applied to gross returns.","card_quote":"We model the annual transaction costs for each stock at 1%, translating to a daily fee of 1%/252 days = 0.00397%, consistent with the findings in SEC (2019).","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The matching‑based strategy achieves a gross Sharpe ratio of 1.23, far above the baseline’s 0.48 and the market’s 0.59.","kind":"performance","quote":"Notably, our matching‑based strategy showcases a marked improvement in risk‑adjusted performance, evidenced by a gross Sharpe ratio of 1.23, a significant enhancement over the baseline value of 0.48 and market value of 0.59."},{"finding":"Matching portfolios reduce variance by eliminating the covariance term that accounts for about 50% of baseline variance.","kind":"theoretical","quote":"Most of this difference comes from the covariance of the pairs with shared stocks, which represents 50% of the variance in the baseline portfolio. This covariance term does not exist in the matching portfolio, producing a much lower portfolio variance."},{"finding":"Net returns of matching approaches are positive (~66%) while baseline approaches are negative after transaction costs.","kind":"performance","quote":"Overall, the matching approaches returned near 66%, while the baseline approaches returns -26.15% and -13.40%."},{"finding":"Matching portfolio turnover is about one‑third of the baseline, leading to lower transaction‑cost drag.","kind":"performance","quote":"The matching method has a median turnover one third as large as the baseline method."},{"finding":"Matching strategies achieve annualized net returns of ~7.6% with Sharpe ratio 1.23, comparable to the S&P 500’s 10.79% return but with superior risk‑adjusted metrics.","kind":"performance","quote":"The net annualized returns of the matching‑based strategies are 7.64% and 7.25%, slightly below the S&P 500’s value of 10.79%."}],"reported_performance":{"strategy":"matching-based pairs trading","sharpe":1.23,"sharpe_benchmark":0.48,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":52,"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":"2026-11-14T18:13:36Z"},"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/"}