Machine learning and sentiment analysis in behavioural investing – evidence from Poland
Bibliographic Data
| ID | 21390774 |
|---|---|
| Authors | Łukasz Kołodziejczyk (0000-0002-4704-4188, SGH Warsaw School of Economics, corresponding author) |
| Year | 2026 |
| Volume | 88 |
| Issue | 1 |
| Pages | 219-239 |
| Publication date | 2026-03-30 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Ruch Prawniczy Ekonomiczny i Socjologiczny (JOURNAL) |
| Journal identifiers | ISSN: 0035-9629 • E-ISSN: 2543-9170 |
| Publisher | Adam Mickiewicz University Poznan (PUBLISHER) |
| DOI | 10.14746/rpeis.2026.88.1.12 |
| OpenAlex | W7160995226 |
| Language | EN |
| References cited | 49 |
The recent popularity of behavioural finance, combined with machine learning, offers an opportunity to challenge the duopoly of fundamental and technical analysis in stock selection. Behavioural analysis – an indirect method of stock evaluation based on the direct analysis of investor behaviour – offers a novel approach to investing. Its most popular instrument, sentiment analysis, has been shown to be useful in investing, although the relation between investor sentiment and stock prices is not yet clear, and behavioural investing is not well described in the literature. Moreover, there are only a few papers concerning investor sentiment in the Polish equity market, which creates a research gap, compared to other developed markets. The goal of this paper is to examine this relation and to test the efficacy of sentiment-based methods in behavioural investing. The relationship between investor sentiment and stock price movements was examined using Matthews and Pearson correlation coefficients, and the efficacy of sentiment-based investment strategies was compared with the buy-and-hold approach. By leveraging threads from the Bankier. pl forum, this paper confirms a positive but modest correlation between investor sentiment and returns of WIG20 stocks, consistent with prior findings. While previous papers found low correlations and modest investment gains, this paper recognizes a statistically significant and stronger relationship between cumulative sentiment and cumulative returns (29.7%) compared to daily sentiment and fluctuations (5.4%, 2.7%, 3.0%). Moreover, this study tests behavioural methods in investment strategies, where sentiment-based trading outperformed the buy-and-hold approach in two-thirds of cases. The theoretical profitability is eliminated by including transaction costs, underscoring limited practical utility, and suggesting future research
Behavioral economics · Equity (law) · Investment (military) · Market sentiment · Popularity · Profitability index · Sentiment analysis · Stock (firearms) · Technical analysis · Complex Systems and Time Series Analysis · Financial Markets and Investment Strategies · Stock Market Forecasting Methods
| Citation velocity | historical |
|---|---|
| Highly cited | No |