{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T06:52:18Z","timestamp":1791183138070,"version":"4.1.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":0,"publisher":"Elsevier BV","license":[{"start":{"date-parts":[[2026,1,1]],"date-time":"2026-01-01T00:00:00Z","timestamp":1767225600000},"content-version":"unspecified","delay-in-days":0,"URL":"https:\/\/www.uspto.gov\/ip-policy\/copyright-policy\/copyright-basics"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>The investment management industry increasingly relies on artificial intelligence and machine learning (ML) to process the growing volume of structured and unstructured data available to equity investors. Existing research, however, remains fragmented: studies typically examine fundamental, technical, or sentiment information in isolation, use small samples or complex model ensembles when they attempt integration, and evaluate performance using statistical accuracy rather than economically meaningful, cost-adjusted portfolio outcomes. This paper presents the conceptual foundation, literature synthesis, methodological design, and empirical results for an explainable machine-learning framework, an XGBoost model paired with SHAP (SHapley Additive exPlanations) that integrates fundamental, technical, and market-sentiment information to predict forward returns of U.S. equities. Sentiment is operationalized through four established, publicly available market-wide indices rather than security-level text. The design uses walk-forward validation, feature-set ablation, cost-adjusted portfolio backtests, and SHAP based attribution across market regimes. The study produced out-of-sample results (2019\u20132025; 82 monthly cross-sections) for 504 U.S. large-cap equities show modest predictive power: the integrated model attains a mean monthly rank information coefficient of .026 (HAC t = 1.36, p = .173), and its improvement over a fundamentals-only baseline (.023) is not statistically significant (one-sided p = .428); neither technical nor sentiment features adds significant value after multiplicity adjustment (Holm-adjusted p = 1.00). The integrated top-decile long-only portfolio earns 27.5% annually net of costs (Sharpe ratio 0.88) versus 17.4% (0.79) for the equal weighted universe, but with higher volatility, no significant alpha (4.6% per year, t = 0.99), and a Sharpe advantage reduces at 50 basis points of one-way cost. Portfolios built on fundamental features, alone or with sentiment, are the exception, earning significant alphas versus SPY (9.9%,t = 2.74; 9.1%, t = 2.27). SHAP attribution is spread across fundamental (about 44%), technical (39%), and sentiment (17%) features, led by firm size, beta, and price-trend measures, and feature rankings are largely stable across market regimes (Spearman \u03c1 = .68-.91). The study is grounded in the Adaptive Markets Hypothesis (Lo, 2004) and Shapley-value theory.\n&lt;p&gt;&lt;\/p&gt;<\/jats:p>","DOI":"10.2139\/ssrn.7560638","type":"posted-content","created":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T06:23:17Z","timestamp":1791181397000},"source":"Crossref","is-referenced-by-count":0,"title":["Explainable Machine Learning for U.S. Equity Selection: Integrating Fundamental,Technical, and Sentiment Signals"],"prefix":"10.2139","author":[{"given":"Chalam Raja","family":"Ankam","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,5]],"date-time":"2026-10-05T06:23:17Z","timestamp":1791181397000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=7560638"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":0,"URL":"https:\/\/doi.org\/10.2139\/ssrn.7560638","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}