Guided lab · synthetic data
How much should one view move the market prior?
Change the view-error variance, then step through prior → view → posterior → allocation. The lab keeps every other input fixed so you can see what confidence alone changes.
Scenario Canonical: A beats B by 4% Neutral: view equals prior spread Failure: non-positive Ω
View variance Ω: 0.0025
Back Step Play Pause Reset
Black-Litterman four-stage flow Four nodes reveal the equilibrium prior, uncertain relative view, posterior mean and final weights as the learner advances.
1 · Prior A 4.000% B 1.000% π = δΣwₘₖₜ
2 · View A − B = 4% Ω = 0.0025 uncertain evidence
3 · Posterior A 4.727% B 0.818% blend, then audit
4 · Weights A 59.09% · B 40.91%
Stage 1 of 4 · Start from market-implied equilibrium returns.
Live audit
Posterior A4.727%
Posterior B0.818%
Weight A59.09%
Weight B40.91%
μᴮᴸ = π + τΣPᵀ(PτΣPᵀ + Ω)⁻¹(q − Pπ)
Canonical result loaded. Press Step to reveal why each number exists.
Keyboard: Right/Space steps, Left goes back, R resets, P plays or pauses. Add ?reduced-motion=1 to disable autoplay.