Interactive models
Reversal Lab
Read the fast market (leader A) and execute on the slow one (follower B). Nine exit detectors — EMA flips, z-score, Kalman velocity, BOCPD, CUSUM, OU optimal stopping — compared side by side, with a Monte-Carlo check over random paths.
Signal Orthogonalization
Strip a shared factor out of a signal — in radio (time & frequency domain) and in finance (α = β·factor + αorth). Watch variance and risk fall as ρ rises.
Problems Collection
A filterable set of quant-interview math problems — by difficulty, topic and firm — with hints, solutions and a solved-tracker that remembers your progress.
Sector Returns → PCA
Build 25 return streams from a market factor, sector factors and noise, then watch PCA read the structure back out — the market spike, the sector bumps, the noise floor.
Derivations & notes
Reversal Math — four rungs
The theory behind the Lab: quickest change-point detection as a delay-vs-false-alarm trade-off, from EMA crossovers up through Kalman trend filters, BOCPD, Shiryaev/CUSUM and OU optimal stopping.
Orthogonalization, two ways
Why projecting out a factor (S₂⊥ = S₂ − β·S₁) is the same move whether you're cleaning a radio channel or isolating idiosyncratic alpha — and the variance reduction you get for it.