Six real techniques that resolve three structural flaws in classical portfolio theory — estimation noise, non-normal tail risk, and a market that never sits still — plus the honest limits of each, and a plain guide to telling real quantitative work apart from the ads that borrow its language.
Not a crystal ball. A better seatbelt.
Every technique below stands on someone real — Wigner's nuclear physics, Bachelier's forgotten 1900 thesis, Itô's wartime solitude, Mundell and Fleming's open-economy model, a Nobel laureate or two, even a 2200-year-old Greek geometer. None of it was invented for this app. We just finally said their names out loud, instead of quietly using their math and moving on.
Random Matrix Theory separates real correlation signal from pure statistical noise in the covariance matrix.
02Shrinks a noisy sample covariance matrix toward a stable target, with the blend ratio solved analytically.
03Optimizes for the worst case inside an uncertainty set, instead of trusting one point estimate.
04Infers which hidden market regime — calm or stressed — is most likely active right now.
05Optimizes directly against expected loss in the worst-case tail, not just overall variance.
06Measures whether assets crash together in a crisis, something ordinary correlation can't see.
An interactive Mundell-Fleming instrument — drag the sliders on fiscal policy, monetary policy, and capital mobility, and watch a small open economy respond. Includes GFC 2008, COVID 2020, and the 2022 inflation surge as historical replays.
DEEP DIVEFinance used this math before physics did, not after — from a forgotten 1900 thesis to the exact correction term running inside MFL-PORTFOLIO-SCIENCE's own forecast cone today.
DEEP DIVEFischer Black (of Black-Scholes) and Robert Litterman's 1990 Goldman Sachs fix for the error-maximizer problem — and Attilio Meucci's 2008 generalization of it.
DEEP DIVEVolatility and correlation that actually move with the market, not a static snapshot — Robert Engle won a Nobel Prize for the math behind it.
DEEP DIVEEconophysics — physicists' own network tools turned on the stock market — plus an honest note on exactly what MFL-PORTFOLIO-SCIENCE built from it.
LIVE NOWNot settled history like the rest of this page — a real, ongoing experiment. This is the theory quietly driving the Portfolio Research Console's "Learned" weights, now visible in a real Active Inference Lab tab.
LIVE NOWA real, published climate-to-inflation channel — NOAA's ocean-temperature index and the IMF research tying it to Australian inflation — and how it already feeds QUANT2's regime model and Quant-Lite's Shock Monitor, with honest limits.
RESEARCHBifurcation, the cusp catastrophe, self-organised criticality and critical slowing down: which parts we can measure, real storm-rehearsal numbers by risk tier, and a regime idea that failed its own test.
READ THISThe specific red flags that separate a real quantitative system from an "AI trading platform" ad — written by people who build the real thing.
General education only. Nothing here is personal financial advice or takes into account your objectives, financial situation or needs. Past performance is not a guarantee of future results.