EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MI-alapú pénzügyi hangulatelemzés tőzsdei előrejelzésekben

David Badon, Virág Vas

Staféta · 05.11.2026 · 11 (11) · 7-23

DOI: 10.52885/wkextx92 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Staféta, online erschienen am 05.11.2026. Verfasst von David Badon und 1 weiteren Autor:innen. Die Arbeit nennt 16 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Behavioural finance research indicates that news sentiment is associated with stock market performance. Smales (2015) found that negative news has a stronger impact on financial markets, while Verma and Verma (2025) reported that economic news indicators outperform social media-based indicators in explaining stock index movements. This study examines whether AI-based Financial Sentiment Analysis (FSA) can improve stock market forecasting. News articles collected from Finviz were analysed using FinBERT in Python to generate sentiment scores, which were then compared with stock returns and trading volume. Granger causality tests were applied to determine whether sentiment indicators can predict subsequent market movements. The findings of the analysis indicate that the predictive power of financial sentiment varies across companies. For some stocks (AMZN, NVDA), a statistically detectable relationship emerges with a lag of 1–3 days, whereas for others, no predictive effect is observed. In some cases (META), a reverse relationship can also be identified.

Zitieren

APA

Badon, D., & Vas, V. (2026). MI-alapú pénzügyi hangulatelemzés tőzsdei előrejelzésekben. Staféta, 11(11), 7-23. https://doi.org/10.52885/wkextx92

Vancouver

Badon D, Vas V. MI-alapú pénzügyi hangulatelemzés tőzsdei előrejelzésekben. Staféta. 2026;11(11):7-23. doi:10.52885/wkextx92

BibTeX
@article{Badon2026,
  author = {Badon, David and Vas, Virág},
  title = {MI-alapú pénzügyi hangulatelemzés tőzsdei előrejelzésekben},
  journal = {Staféta},
  year = {2026},
  volume = {11},
  number = {11},
  pages = {7--23},
  doi = {10.52885/wkextx92}
}
Kostenlos starten →