A contrastive learning framework that blends dynamic stock graphs with static company relations improves next-day stock movement classification by roughly two accuracy points over four baselines.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Dynamic Graph Representation with Contrastive Learning for Financial Market Prediction: Integrating Temporal Evolution and Static Relations
A contrastive learning framework that blends dynamic stock graphs with static company relations improves next-day stock movement classification by roughly two accuracy points over four baselines.