News-Driven Stock Jumps, Financial Graphs, and Cost-Aware Portfolio Learning
Summary
This compilation presents three quantitative finance topics. First, it reports that news frequency and tone are associated with stock return jump likelihood and jump size characteristics. News frequency appears especially important for jump probability, and the reported relationship strengthens over the period studied. Larger firms and firms with greater analyst coverage, media visibility, or institutional ownership show greater sensitivity to news. These are reported empirical associations; the document does not provide the underlying data or methods for independently assessing them.
The second section surveys financial graph applications, including graph construction from stock relationships, filtering dense connections, clustering, price prediction, and portfolio optimization such as hierarchical risk parity. The final section describes machine learning for portfolio construction with trading costs directly included in the objective, so the model learns asset weights while accounting for gradual position changes. The account says this approach improved the executable efficient frontier out of sample, but gives no detailed evaluation design or numerical results. Because these topics are summarized together, each receives limited methodological detail.
Key ideas
- News frequency and absolute tone are associated with stock return jump likelihood and jump size characteristics.
- The reported influence of news on jump likelihood increases over the sample period.
- Firm size, media visibility, analyst coverage, and institutional ownership relate to sensitivity to news-driven jumps.
- Financial graphs can represent stock relationships for clustering, prediction, and portfolio construction.
- Including trading costs in machine-learned portfolio weights can favor gradual trading and more persistent signals.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.