Paper: arXiv 2403.00009

Authors: Cyril Bachelard, Apostolos Chalkis, Vissarion Fisikopoulos, Elias Tsigaridas

Abstract

The present article explores the application of randomized control techniques in empirical asset pricing and performance evaluation. It introduces geometric random walks, a class of Markov chain Monte Carlo methods, to construct flexible control groups in the form of random portfolios adhering to investor constraints. The sampling-based methods enable an exploration of the relationship between academically studied factor premia and performance in a practical setting. In an empirical application, the study assesses the potential to capture premias associated with size, value, quality, and momentum within a strongly constrained setup, exemplified by the investor guidelines of the MSCI Diversified Multifactor index. Additionally, the article highlights issues with the more traditional use case of random portfolios for drawing inferences in performance evaluation, showcasing challenges related to the intricacies of high-dimensional geometry.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 8.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced mathematical concepts like geometric random walks and high-dimensional convex geometry (MCMC methods), while its empirical application is explicitly grounded in real-world data (MSCI index constraints) and computational implementation for portfolio analysis.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>Explore RCT methods for factor premia & performance eval"] --> B{"Methodology & Data"};
  B --> C["Geometric Random Walk<br>(MCMC method) for random portfolios"];
  B --> D["Data:<br>Equity Universe & Constraints<br>(e.g., MSCI Multifactor)"];
  C --> E["Computational Process:<br>Construct flexible control groups"];
  D --> E;
  E --> F{"Empirical Application & Outcomes"};
  F --> G["Findings:<br>Identify premia for size, value, quality, momentum<br>under constraints"];
  F --> H["Highlight challenges:<br>High-dimensional geometry limits<br>traditional inference"];

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes geometric random walks, a class of continuous MCMC methods, to generate random portfolios under high‑dimensional constraints for asset‑pricing and performance analysis.

Source sentence
Specifically, we introduce geometric random walks, a class of continuous MCMC methods tailored for high-dimensional constrained scenarios from computational geometry, as an instrument to create a randomized control which is useful to evaluate asset pricing anomalies within constrained setups.

Assumes 5

  • The distribution of no‑skill performance can be approximated through sampling of random portfolios. statistical
    Source sentence
    However, it relies on the assumption that the distribution of no-skill performance, crucial for hypothesis system evaluation, can be objectively approximated through sampling.
  • Analytical solutions for random‑portfolio performance are generally unavailable, requiring numerical techniques. model
    Source sentence
    In most cases of practical relevance, analytical solutions are not feasible, and we must resort to numerical techniques to obtain the desired results.
  • Stocks must have a continuous five‑year price history without gaps larger than two weeks. data
    Source sentence
    Furthermore, to be included in our study, stocks need a consistent price history of five years without any gaps larger than two weeks.
  • Only liquid stocks with median daily trading volume over the prior 252 days above USD 1.5 million are retained. execution
    Source sentence
    As a threshold, we require a median trading volume over the previous 252 trading days to be above USD 1.5 million.
Show 1 more
  • All returns are computed from discrete daily total returns using closing prices in USD. data
    Source sentence
    returns using closing prices denoted in USD, covering the period 01.01.1995 to 31.12.2022.

Data

  • CRSP daily stock file · Center for Research in Securities Prices (CRSP) market-prices commercial
    1995-01-01 to 2022-12-31 · U.S. large‑ and mid‑cap equities covering ~85% of market cap
    Source sentence
    We suggest utilizing data from the Center for Research in Securities Prices (CRSP) as an alternative data source. We select a subset of stocks listed on the NYSE, AMEX, and NASDAQ indexes, focusing solely on ordinary common shares of companies incorporated in the U.S.
  • MSCI USA index constituents (proxy) · MSCI (proxied via CRSP filtering) benchmark proprietary
    Large and mid‑cap U.S. equities
    Source sentence
    we found that a good approximation to mimick the MSCI universe is to filter for the largest stocks listed on the three exchanges such that in aggregate they cover 85% of total market capitalization.

Recognised sources: JKP Global Factor Data CRSP

Costs

Unclear The classifier was not confident.

Finds 4

  • Random portfolios constructed under realistic constraints are unsuitable for formal hypothesis testing of skill but can be useful for visual and descriptive performance analysis. negative
    Source sentence
    We explained that commonly employed random portfolios do not offer a statistically acceptable experimental design for performance comparisons and cautioned the reader against such a use case with the aid of an illustrative example.
  • A backtest of a nonsensical “letter Z” strategy outperformed the S&P 500 by 6.2% annualized, illustrating spurious skill signals from random portfolios. performance
    Source sentence
    the results show an annualized outperformance of 6.2% (based on daily geometric returns) and the Sharpe ratio test of Ledoit and Wolf
  • Factor loadings of the letter‑Z strategies are significant, with positive exposures to size, value, profitability and investment, and negative to momentum, indicating that apparent outperformance is driven by known factor tilts. statistical
    Source sentence
    one finds that the majority of factor loadings are statistically significant. Coefficients are predominantely positive for size and value as well as for the two quality‑related factors –profitability and investment– while all coefficients for momentum are negative.
  • The authors provide an open‑source C++/R implementation (GeomScale) that reproduces the MSCI Diversified Multi‑Factor index guidelines and enables reproducible experiments. theoretical
    Source sentence
    an efficient open‑source implementation in C++ with an interface in R

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 14 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.