mpriors

pricing

Free teaches the concept.
Pro is the workflow.

The free tier must be good enough to teach the category, or nothing upstream of it works. So Observer isn't a demo — it's the whole concept, uncrippled.

for learning the category

Observer

students, educators, the curious

$0
  • Two baskets
  • The full decomposition — group + solo on every member, every session
  • The complete chart family: comparisons, ghost priors, σ-units
  • Delayed market data
  • The methodology, all of it — nothing is paywalled about how the numbers are made
Start free

ships when Pro earns it

Desk

for people who run it like a job

$99/mo
  • Everything in Practitioner
  • The book view — Neff across everything you watch
  • Discovery scans with the false-discovery math on the label
  • API access and export
Later — not before it's ready

Priced against data subscriptions, not against apps: Practitioner costs less than most standalone live-data feeds, and the analytics come with it. Desk ships only after Practitioner retention proves the core — we'd rather earn the tier than announce it.

· honest answers

Things you'd ask us anyway.

why is the free tier so complete?

Because the category is new. Nobody wakes up wanting "leave-one-out decomposition" — they discover it by watching their own basket split a move into group and solo. Observer exists to make that discovery free.

will mpriors tell me what to trade?

No — and that's structural, not a missing feature. Mpriors describes: how stretched a divergence is, how similar stretches historically behaved, when a relationship's assumptions changed. What to do about it stays yours. See the method →

what happens when a relationship breaks?

You get an assumption-watch notice naming what moved — β break, cohesion decay, a record stretch — the moment the measurement stops meaning what it meant. Mean reversion's failures are rare and violent; watching the assumptions is the honest defense.

can I check the math?

Please do. Every number reconciles to change = group + solo exactly, every statistic names its test and window, and the methodology is published in full — including the estimator's known biases. Our best users are the ones who check.

Start with two baskets, free.

No card, no trial clock. Build a basket of five names and watch the decomposition unlock.