What is it about?
We test whether AI can forecast major stock markets during extreme turbulence (2010–2022). Using decision trees and LSTM neural networks, we predict equity index movements and compare them with a standard linear regression benchmark. The goal is to see which tools stay accurate after shocks like COVID-19 and the Russia–Ukraine war.
Featured Image
Why is it important?
Market shocks can break models trained in calm times. This paper is timely because it benchmarks machine-learning forecasts against linear regression across three sub-periods that include COVID-19 and the Russia–Ukraine war. It shows AI methods can outperform traditional approaches in high-volatility conditions, and that predictive accuracy differs across shock episodes—useful for risk management and algorithmic trading design.
Perspectives
What stood out to me is that “volatile” is not one thing: models behaved differently after COVID-19 than after the outbreak of war. Stress-testing across distinct shock windows feels more informative than reporting a single average error. The comparison also underlines a practical lesson: before adding complexity, always ask whether it improves decisions (signals, timing, portfolio rules) under the regimes investors actually face.
Tibor Bareith
ELTE Centre for Economic and Regional Studies Hungary
Read the Original
This page is a summary of: Analysis of the performance of predictive models during Covid-19 and the Russian-Ukrainian war, Pénzügyi Szemle = Public Finance Quarterly, June 2023, Corvinus University of Budapest,
DOI: 10.35551/pfq_2023_2_7.
You can read the full text:
Contributors
The following have contributed to this page
