{"paper_id":"2506.19856","title":"Supervised Similarity for Firm Linkages","authors":["Ryan Samson","Adrian Banner","Luca Candelori","Sebastien Cottrell","Tiziana Di Matteo","Paul Duchnowski","Vahagn Kirakosyan","Jose Marques","Kharen Musaelian","Stefano Pasquali","Ryan Stever","Dario Villani"],"paper_date":"2025-06-09","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2506.19856v1","page_url":"https://thequant.space/flowcharts/supervised-similarity-for-firm-linkages/","scores":{"empirical_rigor":6,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a new proxy for firm linkages called Characteristic Vector Linkages (CVLs) and applies Quantum Cognition Machine Learning (QCML) to improve similarity-based momentum spillover trading.","quote":"We introduce a novel proxy for firm linkages, Characteristic Vector Linkages (CVLs).","summary":"We introduce a novel proxy for firm linkages, Characteristic Vector Linkages (CVLs). We use this concept to estimate firm linkages, first through Euclidean similarity, and then by applying Quantum Cognition Machine Learning (QCML) to similarity learning. We demonstrate that both methods can be used…","strategy":"QCML similarity CVL momentum spillover strategy"},"assumptions":{"status":"available","items":[{"assumption":"All input characteristics are demeaned by GICS Industry Code to avoid industry bias.","type":"data","stated":"implied","quote":"All input characteristics are demeaned by GICS Industry Code to avoid characteristics linking firms based on industry biases."},{"assumption":"Accounting values are averaged over the most recent 4 quarters.","type":"data","stated":"implied","quote":"Accounting values are averaged over the most recent 4 quarters, unless otherwise stated."},{"assumption":"Transaction costs are not removed; results are not meant as a stand‑alone strategy.","type":"execution","stated":"explicit","quote":"We do not remove transaction costs as these forecasts are not intended to represent a stand-alone investment strategy."},{"assumption":"QCML model parameters are kept static after initial training for simplicity.","type":"model","stated":"explicit","quote":"for simplicity we keep the parameters static after initial training."},{"assumption":"Portfolio returns are smoothed over 21 days and evaluated over Jan 2014–Jun 2024.","type":"market","stated":"implied","quote":"We compute daily returns to the smoothed portfolios, and analyze portfolio performance from January 2014 through June 2024."}]},"datasets":{"status":"available","items":[{"name":"Bloomberg market data","provider":"Bloomberg","kind":"market-prices","access":"commercial","period":null,"universe":null,"quote":"We use market data sourced from Bloomberg and accounting data sourced from S&P Capital IQ:","known_id":"bloomberg"},{"name":"S&P Capital IQ accounting data","provider":"S&P Capital IQ","kind":"fundamentals","access":"commercial","period":null,"universe":null,"quote":"We use market data sourced from Bloomberg and accounting data sourced from S&P Capital IQ:","known_id":"capital-iq"}],"known":[{"id":"bloomberg","name":"Bloomberg","provider":"Bloomberg L.P.","kind":"market-prices","access":"commercial","quote":"We use market data sourced from Bloomberg and accounting data sourced from S&P Capital IQ: Accounting Liquidity: (0.5 * Total Assets + 0.25 * Cash + 0.25 * Current Assets - 0.5 * Intangible Assets), scaled by Total Assets [28] Accruals: Sign flipped 4 quarter change of (Total Assets - Working Capital - Total Liabilities - Long Term Investments + Long Term Debt), scaled by Total Assets [34, 40] Book to Price: Common E","mentions":2,"used_in":2},{"id":"capital-iq","name":"S&P Capital IQ","provider":"S&P Global","kind":"fundamentals","access":"commercial","quote":"We use market data sourced from Bloomberg and accounting data sourced from S&P Capital IQ: Accounting Liquidity: (0.5 * Total Assets + 0.25 * Cash + 0.25 * Current Assets - 0.5 * Intangible Assets), scaled by Total Assets [28] Accruals: Sign flipped 4 quarter change of (Total Assets - Working Capital - Total Liabilities - Long Term Investments + Long Term Debt), scaled by Total Assets [34, 40] Book to Price: Common E","mentions":1,"used_in":1}],"sample":{"start":"2014-01","end":"2024-06","market":"US equity markets","asset_class":"equities","frequency":"daily"},"data_resolution":"daily"},"transaction_costs":{"status":"available","modeled":"no","headline_net_of_costs":false,"detail":null,"quote":"We do not remove transaction costs as these forecasts are not intended to represent a stand-alone investment strategy.","source":"front_matter","rule":"R3","candidates":{"front_matter":"no","evidence_card":"no","evidence_card_quoted":true},"card_detail":null,"card_quote":"We do not remove transaction costs as these forecasts are not intended to represent a stand-alone investment strategy.","meaning":"a trading or execution result is reported without costs"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"QCML 252‑day input signal achieves a Sharpe of 1.10 versus 0.73 for Euclidean, a >50% improvement.","kind":"performance","quote":"the 252-day input return QCML signal shows a Sharpe of 1.10 compared to 0.73 for the Euclidean signal (a greater than 50% improvement in Sharpe)"},{"finding":"QCML combined input signal Sharpe 1.42 exceeds Euclidean combined Sharpe 1.24.","kind":"performance","quote":"The QCML combined input return signal gives a Sharpe of 1.42, an improvement over the Euclidean combined input return signal, which has a Sharpe of 1.24"},{"finding":"QCML half‑life (90.9 days) is 2.5× longer than Euclidean half‑life (36.3 days).","kind":"performance","quote":"QCML signal has a half-life (90.9 days) 2.5 times longer than the Euclidean signal (36.3 days)"},{"finding":"Euclidean similarity yields Sharpe ratios from 0.73 to 1.37 depending on horizon.","kind":"performance","quote":"The Euclidean similarity approach, despite its simplicity, generates risk controlled portfolio returns with Sharpe ratios ranging from 0.73 to 1.37 depending on the input horizon"}],"reported_performance":{"strategy":"QCML similarity CVL momentum spillover strategy","sharpe":1.42,"sharpe_benchmark":null,"annual_return_pct":null,"total_return_pct":null,"max_drawdown_pct":null},"out_of_sample":{"method":"holdout","headline_out_of_sample":true}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":28,"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":"2026-11-14T17:36:10Z"},"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/"}