{"id":"W4376134510","doi":"10.1137/22m1530070","title":"Beating a Benchmark: Dynamic Programming May Not Be the Right Numerical Approach","year":2023,"lang":"en","type":"article","venue":"SIAM Journal on Financial Mathematics","topic":"Financial Markets and Investment Strategies","field":"Economics, Econometrics and Finance","cited_by":10,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Benchmark (surveying); Dynamic programming; Mathematical optimization; Reinforcement learning; Bellman equation; Optimal control; Computer science; Stochastic control; Optimization problem; Mathematics; Artificial intelligence","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002395389,0.0006136707,0.0005243509,0.000501917,0.0003552413,0.001495886,0.0008590131,0.001274721,0.003927783],"category_scores_gemma":[0.01070369,0.0002614813,0.0003571,0.0004389411,0.001069552,0.001773028,0.0009948532,0.00159009,0.0002349906],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001228063,"about_ca_system_score_gemma":0.001147011,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002665512,"about_ca_topic_score_gemma":0.002457683,"domain_scores_codex":[0.9991606,0.0004022483,0.00003793485,0.0001308356,0.000195215,0.00007322658],"domain_scores_gemma":[0.9975252,0.001646522,0.000356613,0.0001554001,0.0002154783,0.0001008077],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00005167719,0.00004964413,0.0007593647,0.000079802,0.00003457617,0.00007191904,0.00003902511,0.8588822,0.0008609885,0.1104427,0.001301618,0.02742652],"study_design_scores_gemma":[0.000005863133,0.00002430382,0.00008805429,0.00001234767,0.000003792935,0.00001059096,0.00001016124,0.9786981,0.0002391385,0.02032643,0.0005770043,0.000004260832],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.04806154,0.0008292391,0.9274684,0.002032421,0.0001132779,0.00008545622,0.00007044443,0.0002598424,0.02107939],"genre_scores_gemma":[0.8793628,0.0004100067,0.1148215,0.0003070551,0.00005373299,0.0001700754,0.00007190007,0.00007603205,0.004726857],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.003927783,"threshold_uncertainty_score":0.01313978,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03803374853306173,"score_gpt":0.2483354425064158,"score_spread":0.210301693973354,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}