EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

CMLF Stock Return Prediction Model Enhanced with Fine-Grained Statistical Feature Fusion

Haoyu Chen

ITM Web of Conferences · 2026 · 89 · 03008

DOI: 10.1051/itmconf/20268903008 ↗

Kurz gefasst

Fachartikel in der Zeitschrift ITM Web of Conferences, online erschienen am 14.08.2026. Verfasst von Haoyu Chen. Die Arbeit nennt 10 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

The Comparative Multi-granularity Learning Framework (CMLF) can be effectively used in the context of stock price trend prediction research, to overcome the problem of mismatch between fine-grained data and coarse-grained prediction targets by incorporating stock data at varying temporal resolutions. The current CMLF method, however, mainly aims at identifying the temporal patterns of fine-grained data by modelling a single time-step secondary input with a two-layer Gated Recurrent Unit (GRU) network, but ignores the statistical properties -captured in intraday fine- grained data. The absence of this results in loss of valuable information limiting the overall predictive ability of the model. To overcome this shortcoming, this paper uses extreme values and the average statistics during the fine-grained stage of data processing, which is the effective combination of such statistical characteristics with the initial temporal representations. Consequently, the model feature extraction ability at each time step is improved. Experimental results reveal that the enhanced model has a greater ability to use intraday fine-grained information, and it has a better predictive efficiency in multi-granularity stock trend forecasting tasks.

Zitieren

APA

Chen, H. (2026). CMLF Stock Return Prediction Model Enhanced with Fine-Grained Statistical Feature Fusion. ITM Web of Conferences, 89, 03008. https://doi.org/10.1051/itmconf/20268903008

Vancouver

Chen H. CMLF Stock Return Prediction Model Enhanced with Fine-Grained Statistical Feature Fusion. ITM Web of Conferences. 2026;89:03008. doi:10.1051/itmconf/20268903008

BibTeX
@article{Chen2026,
  author = {Chen, Haoyu},
  title = {CMLF Stock Return Prediction Model Enhanced with Fine-Grained Statistical Feature Fusion},
  journal = {ITM Web of Conferences},
  year = {2026},
  volume = {89},
  pages = {03008},
  doi = {10.1051/itmconf/20268903008}
}
Kostenlos starten →