Out-of-Sample Predictability of Firm-Specific Stock Price Crashes: A Machine Learning Approach

40 Pages Posted: 28 Mar 2022 Last revised: 19 May 2024

See all articles by Devrimi Kaya

Devrimi Kaya

University of Erlangen-Nuremberg-Friedrich Alexander Universität Erlangen Nürnberg

Doron Reichmann

Ruhr University of Bochum - Department of Finance and Banking

Milan Reichmann

Leipzig University

Date Written: May 19, 2024

Abstract

We use machine learning methods to predict firm-specific stock price crashes and evaluate the out-of-sample prediction performance of various methods compared to traditional regression approaches. Using a wide set of numerical and textual data from 10-K filings, our results show that a logistic regression with financial data inputs performs reasonably well and sometimes outperforms newer classifiers such as random forests and neural networks. However, consistent with concurrent work, we find that a stochastic gradient boosting model systematically outperforms the logistic regression, and forecasts using suitable combinations of financial and textual data inputs yield significantly higher prediction performance. Overall, the evidence suggests that machine learning methods can help predict stock price crashes which should offer significant value to investors.

Keywords: Stock Price Crash Risk; Textual Disclosures; Machine Learning; Natural Language Processing

JEL Classification: G11, G14, G32, M41

Suggested Citation

Kaya, Devrimi and Reichmann, Doron and Reichmann, Milan, Out-of-Sample Predictability of Firm-Specific Stock Price Crashes: A Machine Learning Approach (May 19, 2024). Available at SSRN: https://ssrn.com/abstract=4043938 or http://dx.doi.org/10.2139/ssrn.4043938

Devrimi Kaya

University of Erlangen-Nuremberg-Friedrich Alexander Universität Erlangen Nürnberg ( email )

Lange Gasse 20
Nuremberg, Bavaria 90403
Germany

Doron Reichmann (Contact Author)

Ruhr University of Bochum - Department of Finance and Banking ( email )

Germany

Milan Reichmann

Leipzig University ( email )

Leipzig, DE
Germany

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
301
Abstract Views
904
Rank
190,145
PlumX Metrics