Quantitative Researcher
Quantbox Research
05.2025 - 07.2025
- Examined sell-side analyst prediction data to identify analysts with consistent predictive skill using forecast-result correlation based metrics- Bayesian shrinkage, t-statistic, and developed systematic portfolios using this analysis.
- Conducted extensive strategy experiments, using varying holding periods, adaptive feedback and decay-weighted trends.
- Performed case-by-case analysis of analyst coverage patterns, and stock-level anomalies to construct a highly refined framework.
- Iteratively backtested strategies against benchmarks (Sharpe, Sortino, win ratio), achieving consistent high performance while maintaining high trade volumes and avoiding risky portfolios through deep data-driven filters.
- Extended the framework with a Graph Attention Network (GAT) on the analyst coverage network, revealing hidden graph-based features and integrating them into the existing model to enhance the performance of our framework.