Research · iQuant Labs

The ideas behind the tools.

One readable note per idea — what each tool measures, why it works, and where it fails. Written the way the site is built: plain language, mechanisms over mystique, losses included.
Method We tested topological crash indicators. Here's what survived. Persistent homology and market trees are the frontier of quantitative risk. We audited both against our own gauges before shipping — one earned a place, one earned a caveat. July 2026 · 6 min read Regime Regime investing: what the risk-averse actually get Regime strategies usually earn less than buy-and-hold. The case for them is what happens to your worst year — and the arithmetic of drawdowns most investors never do. July 2026 · 5 min read Regime Why we run three models instead of one Every regime model has a blind spot. Running a probabilistic model against two transparent trend filters turns disagreement itself into information. July 2026 · 4 min read Smart Money Smart money leaves a paper trail Big investors move quietly — but exchanges make them disclose. What bulk deals, insider filings and congressional trades can and cannot tell you. July 2026 · 5 min read Risk Credit usually knows first Bond investors have no upside to be optimistic about — which is exactly why credit spreads are one of the oldest early-warning systems for equity investors. July 2026 · 4 min read Risk Turbulence, absorption, and the anatomy of fragility Crashes are rarely about how far markets fall on a given day. They're about markets starting to move together — and there are ways to measure that before it hurts. July 2026 · 5 min read Method The honest backtest: why we publish our losses Most published backtests are marketing. The difference between in-sample flattery and walk-forward truth — and why a model you can watch fail is worth more. July 2026 · 5 min read
Research notes discuss ideas, not recommendations. Nothing here is investment advice — see the Disclaimer. New notes are added as the work earns them; get them via alerts or check back.