Insights
Selected research and market commentary.
Notes from the research team, intended for sophisticated readers. The full set of work, including longer technical papers, is available to qualified investors on request.
- Research note11 min
Sentiment Factors and Deep Sequence Models: What Survives Honest Evaluation
Combining low-signal-to-noise sentiment factors with high-capacity sequence models is a tempting research direction. We summarize what we have found does and does not survive realistic forward testing — and why the productive use of deep learning in this setting is narrower than the literature suggests.
- Research note8 min
On the Persistence of Cross-sectional Momentum in Single-stock Options
We examine whether implied-volatility momentum effects identified in the equity literature survive the higher transaction costs and capacity constraints of the listed single-stock options market.
- Methodology10 min
Regime Detection without Overfitting: A Sober View
Most published regime-switching results fail in forward testing. We sketch a framework for evaluating regime models against the appropriate null hypothesis, and argue regimes are best used as a risk overlay rather than as an alpha source.
Additional notes — including longer technical papers and topics we are not yet ready to publish externally — are available to qualified investors. Request access via the contact form.