{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,10,18]],"date-time":"2026-10-18T12:39:41Z","timestamp":1792327181940,"version":"4.4.0"},"posted":{"date-parts":[[2026]]},"group-title":"SSRN","reference-count":43,"publisher":"Elsevier BV","content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"abstract":"<jats:p>Based on the \u201cvirtue of complexity\u201d framework of Kelly et al. (2024a), we propose a novel two-step method to optimize input variables in machine learning for predicting stock returns. This method first selects new variables according to the marginal effect of out-of-sample R2 and Sharpe ratio, and then uses principal component analysis (PCA) to reduce the dimension of the current variable set. For comparison, the results under only the variable selection method and the dimension reduction method are also presented. The proposed methods are applied to aggregate US equity market return prediction and it is found that the two-step method performs best when the training window is long enough and the penalty is large enough. In our empirical analysis, the out-of-sample R2and Sharpe ratio are improved by up to 176.34% and 71.95% respectively compared to the benchmark under the new method. Using only the variable selection method provides better robustness and can improve model performance across different training windows, but the effect is relatively limited. The method proposed in our study improves the efficiency of input variable selection and the out-of-sample performance of the machine learning model.<\/jats:p>","DOI":"10.2139\/ssrn.6546054","type":"posted-content","created":{"date-parts":[[2026,4,8]],"date-time":"2026-04-08T23:40:47Z","timestamp":1775691647000},"source":"Crossref","is-referenced-by-count":0,"title":["Enhanced Prediction of Stock Returns: Optimizing Variable Selection in Machine Learning"],"prefix":"10.2139","author":[{"given":"Jiaxu","family":"Huo","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5400-749X","authenticated-orcid":true,"given":"Chunping","family":"Zhong","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-5313-8044","authenticated-orcid":true,"given":"Gaoxiu","family":"Qiao","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"78","reference":[{"issue":"6","key":"ref1","doi-asserted-by":"crossref","first-page":"716","DOI":"10.1109\/TAC.1974.1100705","article-title":"A new look at the statistical model identification","volume":"19","author":"H Akaike","year":"1974","journal-title":"IEEE transactions on automatic control"},{"issue":"3","key":"ref2","doi-asserted-by":"crossref","first-page":"1593","DOI":"10.1111\/jofi.13226","article-title":"Integrating factor models","volume":"78","author":"D Avramov","year":"2023","journal-title":"The Journal of Finance"},{"issue":"3686164","key":"ref3","first-page":"20","article-title":"Predicting corporate bond returns: Merton meets machine learning","author":"T G Bali","year":"2020","journal-title":"Georgetown McDonough School of Business Research Paper"},{"issue":"48","key":"ref4","doi-asserted-by":"crossref","first-page":"30063","DOI":"10.1073\/pnas.1907378117","article-title":"Benign overfitting in linear regression","volume":"117","author":"P Bartlett","year":"2020","journal-title":"Proceedings of the National Academy of Sciences"},{"key":"ref5","first-page":"15849","volume":"116","author":"M Belkin","year":"2019","journal-title":"Reconciling modern machine-learning practice and the classical bias-variance trade-off"},{"key":"ref6","first-page":"541","article-title":"To understand deep learning we need to understand kernel learning","author":"M Belkin","year":"2018","journal-title":"International Conference on Machine Learning"},{"key":"ref7","author":"Pmlr"},{"key":"ref8","doi-asserted-by":"crossref","DOI":"10.3386\/w31689","author":"A Didisheim","year":"2023","journal-title":"Complexity in factor pricing models (No. w31689)"},{"key":"ref9","author":"M F Dixon","year":"2019","journal-title":"The four horsemen of machine learning in finance"},{"issue":"1","key":"ref10","doi-asserted-by":"crossref","first-page":"259","DOI":"10.1111\/jofi.13415","article-title":"Test assets and weak factors","volume":"80","author":"S Giglio","year":"2025","journal-title":"The Journal of Finance"},{"key":"ref11","author":"A Goyal","year":"2021","journal-title":"A comprehensive look at the empirical performance of equity premium prediction ii"},{"key":"ref12","doi-asserted-by":"crossref","first-page":"659","DOI":"10.1007\/s11222-016-9646-1","article-title":"Correlation and variable importance in random forests","volume":"27","author":"B Gregorutti","year":"2017","journal-title":"Statistics and Computing"},{"issue":"5","key":"ref13","doi-asserted-by":"crossref","first-page":"2223","DOI":"10.1093\/rfs\/hhaa009","article-title":"Empirical asset pricing via machine learning","volume":"33","author":"S Gu","year":"2020","journal-title":"The Review of Financial Studies"},{"key":"ref14","first-page":"1157","article-title":"An introduction to variable and feature selection","volume":"3","author":"I Guyon","year":"2003","journal-title":"Journal of machine learning research"},{"issue":"5","key":"ref15","doi-asserted-by":"crossref","first-page":"1980","DOI":"10.1093\/rfs\/hhaa017","article-title":"Factor timing","volume":"33","author":"V Haddad","year":"2020","journal-title":"The Review of Financial Studies"},{"issue":"5","key":"ref16","doi-asserted-by":"crossref","first-page":"551","DOI":"10.1016\/0893-6080(90)90005-6","article-title":"Universal approximation of an unknown mapping and its derivatives using multilayer feedforward networks","volume":"3","author":"K Hornik","year":"1990","journal-title":"Neural networks"},{"key":"ref17","author":"A Jacot","year":"2018","journal-title":"Neural Tangent Kernel: Convergence and Generalization in Neural Networks. arXiv e-prints"},{"issue":"5","key":"ref18","doi-asserted-by":"crossref","first-page":"1721","DOI":"10.1111\/jofi.12060","article-title":"Market expectations in the cross -section of present values","volume":"68","author":"B Kelly","year":"2013","journal-title":"The Journal of Finance"},{"key":"ref19","doi-asserted-by":"crossref","DOI":"10.3386\/w31502","author":"B Kelly","year":"2023","journal-title":"Financial Machine Learning (No. w31502)"},{"key":"ref20","author":"B Kelly","year":"2024","journal-title":"Large (and Deep) Factor Models"},{"key":"ref21","author":"B Kelly","year":"2022","journal-title":"The virtue of complexity everywhere"},{"issue":"1","key":"ref22","doi-asserted-by":"crossref","first-page":"459","DOI":"10.1111\/jofi.13298","article-title":"The virtue of complexity in return prediction","volume":"79","author":"B Kelly","year":"2024","journal-title":"The Journal of Finance"},{"issue":"541","key":"ref23","doi-asserted-by":"crossref","first-page":"287","DOI":"10.1080\/01621459.2021.1928514","article-title":"Variable selection via Thompson sampling","volume":"118","author":"Y Liu","year":"2023","journal-title":"Journal of the American Statistical Association"},{"issue":"509","key":"ref24","doi-asserted-by":"crossref","first-page":"270","DOI":"10.1080\/01621459.2014.908125","article-title":"Regularization methods for high-dimensional instrumental variables regression with an application to genetical genomics","volume":"110","author":"W Lin","year":"2015","journal-title":"Journal of the American Statistical Association"},{"issue":"1","key":"ref25","doi-asserted-by":"crossref","first-page":"171","DOI":"10.1016\/j.jfineco.2005.12.002","article-title":"The empirical risk -return relation: A factor analysis approach","volume":"83","author":"S C Ludvigson","year":"2007","journal-title":"Journal of financial economics"},{"key":"ref26","first-page":"1","article-title":"Degree-heterogeneous Latent Class Analysis for High-dimensional Discrete Data","author":"Z Lyu","year":"2025","journal-title":"Journal of the American Statistical Association"},{"issue":"4","key":"ref27","doi-asserted-by":"crossref","first-page":"667","DOI":"10.1002\/cpa.22008","article-title":"The generalization error of random features regression: Precise asymptotics and the double descent curve","volume":"75","author":"S Mei","year":"2022","journal-title":"Communications on Pure and Applied Mathematics"},{"key":"ref28","author":"B Moritz","year":"2016","journal-title":"Tree-based conditional portfolio sorts: The relation between past and future stock returns"},{"issue":"12","key":"ref29","doi-asserted-by":"crossref","DOI":"10.1088\/1742-5468\/ac3a74","article-title":"Deep double descent: Where bigger models and more data hurt","author":"P Nakkiran","year":"2021","journal-title":"Journal of Statistical Mechanics: Theory and Experiment"},{"key":"ref30","article-title":"Forecasting Chinese crude oil futures volatility: New evidence based on dual feature processing of large-scale variables","author":"G Qiao","year":"2024","journal-title":"Journal of Forecasting"},{"key":"ref31","article-title":"Random features for large-scale kernel machines","author":"A Rahimi","year":"2007","journal-title":"Advances in neural information processing systems"},{"issue":"539","key":"ref32","doi-asserted-by":"crossref","first-page":"1270","DOI":"10.1080\/01621459.2020.1847121","article-title":"Variational Bayes for high-dimensional linear regression with sparse priors","volume":"117","author":"K Ray","year":"2022","journal-title":"Journal of the American Statistical Association"},{"issue":"5","key":"ref33","doi-asserted-by":"crossref","first-page":"465","DOI":"10.1016\/0005-1098(78)90005-5","article-title":"Modeling by shortest data description","volume":"14","author":"J Rissanen","year":"1978","journal-title":"Automatica"},{"key":"ref34","first-page":"461","author":"G Schwarz","year":"1978","journal-title":"Estimating the dimension of a model. The annals of statistics"},{"issue":"512","key":"ref35","doi-asserted-by":"crossref","first-page":"1607","DOI":"10.1080\/01621459.2014.984812","article-title":"High-dimensional variable selection with reciprocal l 1-regularization","volume":"110","author":"Q Song","year":"2015","journal-title":"Journal of the American Statistical Association"},{"key":"ref36","author":"D J Sutherland","year":"2015","journal-title":"On the error of random Fourier features"},{"issue":"548","key":"ref37","doi-asserted-by":"crossref","first-page":"3008","DOI":"10.1080\/01621459.2023.2298028","article-title":"Clustering High-Dimensional Noisy Categorical Data","volume":"119","author":"Z Tian","year":"2024","journal-title":"Journal of the American Statistical Association"},{"issue":"123","key":"ref38","first-page":"1","article-title":"Benign overfitting in ridge regression","volume":"24","author":"A Tsigler","year":"2023","journal-title":"Journal of Machine Learning Research"},{"issue":"4","key":"ref39","doi-asserted-by":"crossref","first-page":"1455","DOI":"10.1093\/rfs\/hhm014","article-title":"A comprehensive look at the empirical performance of equity premium prediction","volume":"21","author":"I Welch","year":"2008","journal-title":"The Review of Financial Studies"},{"issue":"545","key":"ref40","doi-asserted-by":"crossref","first-page":"343","DOI":"10.1080\/01621459.2022.2118601","article-title":"Subspace estimation with automatic dimension and variable selection in sufficient dimension reduction","volume":"119","author":"J Zeng","year":"2024","journal-title":"Journal of the American Statistical Association"},{"issue":"528","key":"ref41","doi-asserted-by":"crossref","first-page":"1708","DOI":"10.1080\/01621459.2018.1527227","article-title":"Optimal sparse singular value decomposition for high-dimensional high-order data","volume":"114","author":"A Zhang","year":"2019","journal-title":"Journal of the American Statistical Association"},{"issue":"1","key":"ref42","doi-asserted-by":"crossref","first-page":"486","DOI":"10.1016\/j.ijforecast.2021.12.013","article-title":"Forecasting crude oil market volatility using variable selection and common factor","volume":"39","author":"Y Zhang","year":"2023","journal-title":"International Journal of Forecasting"},{"issue":"512","key":"ref43","doi-asserted-by":"crossref","first-page":"1770","DOI":"10.1080\/01621459.2015.1036994","article-title":"Reinforcement learning trees","volume":"110","author":"R Zhu","year":"2015","journal-title":"Journal of the American Statistical Association"}],"container-title":[],"original-title":[],"deposited":{"date-parts":[[2026,10,18]],"date-time":"2026-10-18T11:40:08Z","timestamp":1792323608000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.ssrn.com\/abstract=6546054"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026]]},"references-count":43,"URL":"https:\/\/doi.org\/10.2139\/ssrn.6546054","relation":{},"subject":[],"published":{"date-parts":[[2026]]},"subtype":"preprint"}}