Relative Valuation with Machine Learning

Abstract

We use machine learning for relative valuation and peer firm selection. In out‐of‐sample tests, our machine learning models substantially outperform traditional models in valuation accuracy. This outperformance persists over time and holds across different types of firms. The valuations produced by machine learning models behave like fundamental values. Overvalued stocks decrease in price and undervalued stocks increase in price in the following month. Determinants of valuation multiples identified by machine learning models are consistent with theoretical predictions derived from a discounted cash flow approach. Profitability ratios, growth measures, and efficiency ratios are the most important value drivers throughout our sample period. We derive a novel method to express valuation multiples predicted by our machine learning models as weighted averages of peer firm multiples. These weights are a measure of peer–firm comparability and can be used for selecting peer‐groups.

Private Firm Valuation Using Multiples: Can Artificial Intelligence Algorithms Learn Better Peer Groups?

“…Gradient Boosting Decision Trees were proposed for relative valuation and peer firm selection tasks to answer the question of stocks being overvalued or undervalued relative to their peers [49]. The authors argue that the choice of peers is often highly subjective, peer selection is loosely based on the industry criterion or only a limited number of selection criteria (variables), and empirical evidence suggests practitioners strategically selecting peers to achieve desired valuation results.…”

Instance-based Explanations for Gradient Boosting Machine Predictions with AXIL Weights

“…This study builds upon our prior work in [6], where an earlier version of the AXIL algorithm was used to select "peer firms" for the purpose of relative valuation. In that setting the objective was to find firms that are similar to each other, but only insofar as that similarity relates to valuation.…”

“…When AXIL weights are fitted and applied to the same dataset, the resulting matrix is symmetric. 6 Each entry in the matrix can be interpreted as a measure of similarity between two instances (see Figure 2a).…”

Digital Transformation, Operating Capacity and Firm Performance: Empirical Evidence from the Chinese Stock Market

“…This paper adopts the Word2Vec technology method based on text analysis and machine learning on the expression information of keywords in annual statistical reports, which can better accommodate the text context environment according to the context content of the vocabulary, realize the expansion of similar words based on the particular corpus of finance and economics, and avoid the subjectivity of manual judgment. This method has been proven feasible in empirical research based on text analysis (Siano & Wysocki 2021;Geertsema & Lu 2023).…”

“…It establishes a theoretical model of the relationship between digital transformation, firm operating capability and firm performance. Specifically, first, this paper uses a Word2Vec-based machine learning approach to construct digital transformation indicators to form an unbalanced panel data of A-share listed companies in China's Shanghai and Shenzhen markets between 2015 and 2021 (Siano & Wysocki 2021;Geertsema & Lu 2023;Ma et al 2023); then, it empirically tests the relationship between digital transformation, corporate operating capacity and firm performance. We find that: (1) the higher the degree of digital transformation, the stronger the firm's operating capacity.…”