Research and projects
Flagship research
Real, correctly signed factor structure on SGX; spreads and borrow remove it before it becomes a strategy.
The FX volatility premium is real, left-skewed and insurance-like; a leak-free high-IV selection rule roughly doubles the payoff proxy.
The same DV01-neutral 2s10s book earns a net Sharpe of −0.79 under a momentum signal and +0.71 under mean-reversion.
Across 1,000+ specifications and five estimator classes, no model is distinguishable from a univariate AR; shocks and identification add nothing out of sample.
Nothing beats zero on levels; cross-sectionally there is a real stress-regime signal that does not survive transaction costs.
A deterministic single-queue event loop with explicit latency and fill models, built so that a naive quoting strategy loses money for the right reasons.
Further research and systems
The data decisively prefer a forward-looking, interest-rate-smoothing policy rule (log Bayes factors +91 and +46), and the canonical orthogonal-shock assumption is rejected.
BVAR posterior forecasts become Black-Litterman views with the forecast covariance as view uncertainty, so model uncertainty reaches position sizing.
Cross-country yield spreads were not cointegrated out of regime over 2010–2026; the hedge ratio found in one regime is the wrong position in the next.
Gold and yen clear the breakeven win rate for defined-risk spreads; EUR-correlated pairs sit just below it.
Across 5,763 House disclosures, sign-adjusted abnormal returns are roughly zero to slightly negative at every horizon a copy-trader could act on.