R-SIGMA · Opportunity Z-Score

How unusual is this market?

Every ZIP, base, city and college is scored by how many standard deviations its investment opportunity sits from the mean. A market at +3σ is a statistical extreme — a "crazy value" worth a hard look. Sigma tells you where to look. It does not tell you to buy.

Read this first. A high positive sigma means a market is statistically unusual — nothing more. That extreme can be a genuine bargain, a data artifact (a stale home-value estimate inflating yield), or a market correctly pricing in a risk our data does not capture. Sigma is a flashlight, not a verdict. Every outlier on the leaderboard below needs human due diligence before it means anything.

Distribution — National Gross-Yield Sigma

typical (|σ| ≤ 1) notable (1–2) rare (2–3) exceptional (> 3)

Leaderboard — Highest-Sigma Opportunities

# Market Surface DSCR Gross Yield Within-Surface σ National Yield σ

Ranked by national gross-yield sigma. The very top rows are exactly the cases the caveat warns about: an apparent 90%+ yield almost always means a stale or wrong home-value figure, not a real opportunity. Treat the top of this list as a data-quality queue first, an opportunity queue second.

Per-Surface Baselines

SurfacePrimary metric MedianRobust σ (1.4826·MAD) Plain meanPlain std n

Methodology

The opportunity z-score ("sigma") for each market:

σ = ( value − median ) / ( 1.4826 × MAD )

Why median + MAD, not mean + std. The whole point of this page is to find outliers. But a plain mean and standard deviation are themselves distorted by outliers: one absurd 95% yield row drags the mean up and inflates the std, which then shrinks every z-score and hides the genuine extremes. So the baseline uses robust statistics:

What's excluded. Items with no opportunity metric stay NULL — they receive no sigma and never enter the baseline. Non-finite values are dropped. The robust baseline is computed on the full non-null pool; we do not delete rows, because median/MAD already neutralise the tails.

Two sigmas per item. opportunity_sigma is computed within each surface against that surface's own metric (DSCR where available — milbase, collegemap, cities, seattle — else gross yield for zipdata). yield_sigma_national pools every market across all five surfaces on gross yield (annual rent / home value), the one metric comparable everywhere, and z-scores against that single national distribution. That is what lets us say a base is "+2.7σ nationally".

Sigma bands. typical |σ| ≤ 1 · notable 1–2 · rare 2–3 · exceptional > 3. A positive sigma on DSCR or yield is a better-than-average opportunity; a negative sigma is worse-than-average. The DSCR rate spine is the single 6.51% FRED MORTGAGE30US used across all surfaces.