Growing Mimicking Portfolios: Estimating Nontraded Factor Risk Premia

Guanhao Feng; Jingyu He; Jianxin Ma; Cesare Robotti

doi:10.2139/ssrn.5758302

Growing Mimicking Portfolios: Estimating Nontraded Factor Risk Premia

Guanhao Feng, Jingyu He, Jianxin Ma, Cesare Robotti

Research output: Working Papers › Preprint

Abstract

Estimating risk premia for nontraded factors is challenging because these factors are weakly correlated with returns and difficult to proxy with traded assets. We introduce the Mimicking Portfolio Tree (MPT), a data-driven method that constructs characteristic-sorted portfolios to maximize their correlation with nontraded factors. Instead of using predefined bases, MPT hierarchically partitions stocks by optimally selecting characteristics and split thresholds, improving factor spanning. Empirically, MPT generates mimicking portfolios with substantially higher correlations and more reliable risk-premia estimates. Applying MPT to 69 nontraded factors over four decades, we find only about one-third earn statistically significant premia.

Original language	English
DOIs	https://doi.org/10.2139/ssrn.5758302
Publication status	Published - 17 Nov 2025

Bibliographical note

Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s)

Research Keywords

Characteristic-based portfolios
Machine Learning
Mimicking Portfolios
Nontraded Factors
Risk Premium

Access Output

10.2139/ssrn.5758302

Other Access Options

Check CityUHK Library

Permanent Link

Right click to copy link

Cite this

@techreport{44a8c74851b047138e0fb9acbc0a808e,

title = "Growing Mimicking Portfolios: Estimating Nontraded Factor Risk Premia",

abstract = "Estimating risk premia for nontraded factors is challenging because these factors are weakly correlated with returns and difficult to proxy with traded assets. We introduce the Mimicking Portfolio Tree (MPT), a data-driven method that constructs characteristic-sorted portfolios to maximize their correlation with nontraded factors. Instead of using predefined bases, MPT hierarchically partitions stocks by optimally selecting characteristics and split thresholds, improving factor spanning. Empirically, MPT generates mimicking portfolios with substantially higher correlations and more reliable risk-premia estimates. Applying MPT to 69 nontraded factors over four decades, we find only about one-third earn statistically significant premia.",

keywords = "Characteristic-based portfolios, Machine Learning, Mimicking Portfolios, Nontraded Factors, Risk Premium",

author = "Guanhao Feng and Jingyu He and Jianxin Ma and Cesare Robotti",

note = "Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s)",

year = "2025",

month = nov,

day = "17",

doi = "10.2139/ssrn.5758302",

language = "English",

type = "WorkingPaper",

}

TY - UNPB

T1 - Growing Mimicking Portfolios

T2 - Estimating Nontraded Factor Risk Premia

AU - Feng, Guanhao

AU - He, Jingyu

AU - Ma, Jianxin

AU - Robotti, Cesare

N1 - Full text of this publication does not contain sufficient affiliation information. With consent from the author(s) concerned, the Research Unit(s) information for this record is based on the existing academic department affiliation of the author(s)

PY - 2025/11/17

Y1 - 2025/11/17

N2 - Estimating risk premia for nontraded factors is challenging because these factors are weakly correlated with returns and difficult to proxy with traded assets. We introduce the Mimicking Portfolio Tree (MPT), a data-driven method that constructs characteristic-sorted portfolios to maximize their correlation with nontraded factors. Instead of using predefined bases, MPT hierarchically partitions stocks by optimally selecting characteristics and split thresholds, improving factor spanning. Empirically, MPT generates mimicking portfolios with substantially higher correlations and more reliable risk-premia estimates. Applying MPT to 69 nontraded factors over four decades, we find only about one-third earn statistically significant premia.

AB - Estimating risk premia for nontraded factors is challenging because these factors are weakly correlated with returns and difficult to proxy with traded assets. We introduce the Mimicking Portfolio Tree (MPT), a data-driven method that constructs characteristic-sorted portfolios to maximize their correlation with nontraded factors. Instead of using predefined bases, MPT hierarchically partitions stocks by optimally selecting characteristics and split thresholds, improving factor spanning. Empirically, MPT generates mimicking portfolios with substantially higher correlations and more reliable risk-premia estimates. Applying MPT to 69 nontraded factors over four decades, we find only about one-third earn statistically significant premia.

KW - Characteristic-based portfolios

KW - Machine Learning

KW - Mimicking Portfolios

KW - Nontraded Factors

KW - Risk Premium

U2 - 10.2139/ssrn.5758302

DO - 10.2139/ssrn.5758302

M3 - Preprint

BT - Growing Mimicking Portfolios

ER -

Growing Mimicking Portfolios: Estimating Nontraded Factor Risk Premia

Abstract

Bibliographical note

Research Keywords

Access Output

Other Access Options

Permanent Link

Fingerprint

Cite this