{"paper":{"title":"Deep Reinforcement Learning Framework for Diversified Portfolio Management Across Global Equity Markets","license":"http://creativecommons.org/licenses/by-nc-nd/4.0/","headline":"Deep reinforcement learning for portfolio allocation shows competitive performance mainly in the Euro Stoxx 50 but delivers no statistically significant excess returns over buy-and-hold across global equity markets.","cross_cats":["cs.AI","cs.LG","cs.NE","q-fin.TR"],"primary_cat":"q-fin.PM","authors_text":"Kamil Kashif, Robert \\'Slepaczuk","submitted_at":"2026-05-17T07:50:37Z","abstract_excerpt":"This study develops and evaluates a deep reinforcement learning framework for dynamic portfolio allocation across global equity markets. The Soft Actor-Critic algorithm is used to learn continuous portfolio weights within a Markov Decision Process, incorporating transaction costs, turnover penalties, and diversification constraints into the reward function. Five model configurations are compared, varying in reward formulation, policy structure (flat versus hierarchical Dirichlet), portfolio constraints, and temporal encoder (LSTM versus Transformer), and evaluated via walk-forward optimization"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"RL strategies achieve competitive risk-adjusted performance primarily in the Euro Stoxx 50, where statistically significant abnormal returns are observed, but the central hypothesis is only partially confirmed: no strategy achieves statistically significant excess returns relative to Buy and Hold under HAC-robust inference across all markets.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"The sixteen walk-forward out-of-sample folds spanning 2003-2026 provide a sufficiently unbiased test of out-of-sample performance without the RL agent overfitting to the specific market regimes present in the training windows.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"A Soft Actor-Critic reinforcement learning framework for dynamic global equity allocation shows competitive risk-adjusted returns mainly in Euro Stoxx 50 but no consistent statistically significant outperformance versus buy-and-hold across all three markets.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Deep reinforcement learning for portfolio allocation shows competitive performance mainly in the Euro Stoxx 50 but delivers no statistically significant excess returns over buy-and-hold across global equity markets.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"38dbaabd31edc40221cf9ccec6e038977d2ab54bd79dba7823cb057e8daafcce"},"source":{"id":"2605.17307","kind":"arxiv","version":1},"verdict":{"id":"7ba134ef-1d95-435a-895d-acc1ef36f31e","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-19T23:01:11.162588Z","strongest_claim":"RL strategies achieve competitive risk-adjusted performance primarily in the Euro Stoxx 50, where statistically significant abnormal returns are observed, but the central hypothesis is only partially confirmed: no strategy achieves statistically significant excess returns relative to Buy and Hold under HAC-robust inference across all markets.","one_line_summary":"A Soft Actor-Critic reinforcement learning framework for dynamic global equity allocation shows competitive risk-adjusted returns mainly in Euro Stoxx 50 but no consistent statistically significant outperformance versus buy-and-hold across all three markets.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"The sixteen walk-forward out-of-sample folds spanning 2003-2026 provide a sufficiently unbiased test of out-of-sample performance without the RL agent overfitting to the specific market regimes present in the training windows.","pith_extraction_headline":"Deep reinforcement learning for portfolio allocation shows competitive performance mainly in the Euro Stoxx 50 but delivers no statistically significant excess returns over buy-and-hold across global equity markets."},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2605.17307/integrity.json","findings":[],"available":true,"detectors_run":[{"name":"doi_title_agreement","ran_at":"2026-05-19T23:31:20.156251Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"doi_compliance","ran_at":"2026-05-19T23:13:08.639839Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"claim_evidence","ran_at":"2026-05-19T22:01:57.794290Z","status":"completed","version":"1.0.0","findings_count":0},{"name":"ai_meta_artifact","ran_at":"2026-05-19T21:33:23.756706Z","status":"skipped","version":"1.0.0","findings_count":0}],"snapshot_sha256":"b6031aa13e6d09008ffe1bfdd095d01160ad83ef22cb77c00793eb54f75c4bb5"},"references":{"count":48,"sample":[{"doi":"","year":2021,"title":"and Consoli, Sergio and Piras, Luca and Podda, Alessandro Sebastian and Recupero, Diego Reforgiato , title =","work_id":"1728ec53-efb7-4829-9132-22525a9d871d","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Sensors , volume =","work_id":"98971898-657c-434e-9df9-c8ee8bc0bce6","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"Bui, Quynh and. Applying. Physica A: Statistical Mechanics and its Applications , volume =. 2022 , publisher =","work_id":"9ab7c52d-120f-4e05-abc9-0d9e6f225feb","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1952,"title":"The Journal of Finance , volume =","work_id":"cbfe7d50-afe4-4410-829d-6a1f39565925","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1992,"title":"Financial Analysts Journal , volume =","work_id":"7350b074-7001-489c-afcf-c478ca80b02f","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":48,"snapshot_sha256":"17dfeecbdc4feefd31cabd0f3930b0e7da19ccd196352b28c1bab68abaa55d07","internal_anchors":2},"formal_canon":{"evidence_count":1,"snapshot_sha256":"bcd971435e4d3e9c97ed7f38e992c53422a2153b7a687afb62175e5748c1da95"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}