The median US Altman Z-Score is 2.37

Altman's original cutoffs divide the Z-Score into three zones: below 1.81 is distress, 1.81 to 2.99 is grey, above 2.99 is safe. On 2026-09-07, across 2,773 US companies with filed statements on record and the model applied only where it applies, the median Z-Score was 2.37.

The median company is in the grey zone. The 25th percentile is 0.41, which is not near the distress boundary but far below it.

From GET /api/public/distribution/altman_z, build dated 2026-09-07. One deduplicated company per row, rounded to two decimals, halves away from zero. Banks, insurers, brokers, equity REITs, mortgage REITs and specialty finance are excluded from the cohort, 865 companies, on the grounds that the model was not fitted for them.

More than a quarter of a market is not a distress signal

Listed US companies do enter bankruptcy, and the annual rate at which they do is a small fraction of the market. A zone holding more than a quarter of the companies the model applies to cannot be predicting that event with any useful specificity. Whatever the distress zone is identifying, it is a much larger population than the one that fails.

This is not a criticism of Altman. It is what happens when a discriminant function fitted on a matched sample of failed and surviving manufacturers in the 1960s is run across every listed company in 2026. The base rate in the fitting sample was fifty percent by construction. The base rate in the market is nothing like that, and a classifier calibrated on the first is uninformative when applied to the second without recalibration. The same base-rate problem sits under the Beneish M-Score's one published cutoff.

The practical reading is the ordering, not the label. A company at 0.4 is worse placed than a company at 5.0 on the five things the model measures. That is a real statement. "It is in the distress zone" implies a probability of failure that this distribution does not support.

Where the median sits, by sector

The median company is in the safe zone in four sectors, the grey zone in two and the distress zone in three. Two sectors, Financial Services and Real Estate, publish nothing here at all: after the exclusions, fewer than thirty companies remain in each, and thirty is the smallest cohort this product will compute a percentile from.

Healthcare and utilities land in the distress zone for opposite reasons. Healthcare is full of companies with no revenue and an operating loss, which empties two of the five terms. Utilities carry a rate base so large that every term with assets in the denominator is small. Neither result is about solvency. A second published ratio misreads the same two cohorts for the same two reasons, against an edge taken from a rating table rather than fitted on a 1960s sample. Healthcare's median interest coverage of -7.14 is an operating loss divided by an interest bill, and the utilities cohort's middle half sits below the investment-grade edge while borrowing against a regulated revenue stream.

The term that moves without a filing

Four of the five terms come from the latest filed fiscal year. The fourth, market value of equity over total liabilities, does not.

That single term is weighted 0.6 and is unbounded above, and it is the reason a company's Z-Score falls when its share price falls with nothing having changed on any statement. It is also why the distribution has such a long right tail: the 95th percentile is 18.79 and the upper winsorisation edge is 67.26, which is a company with a large market capitalisation and almost no liabilities rather than one that is fifteen times safer than the median.

A Z-Score is a filed number with one market number bolted into it, and the freshness of the two halves is different: four terms hold still until the next report, and the fifth moves with every session. That is worth knowing whenever a Z-Score is compared against a threshold to the second decimal. The percentiles here are one build, dated 2026-09-07, and the market term inside them is that day's.

What the five terms are

Z = 1.2 (working capital / assets)
  + 1.4 (retained earnings / assets)
  + 3.3 (EBIT / assets)
  + 0.6 (market cap / total liabilities)
  + 1.0 (revenue / assets)

The third term carries the heaviest weight, so operating profitability relative to the asset base does most of the work. The second term, retained earnings over assets, is quietly an age proxy: a company that has been profitable for thirty years has accumulated retained earnings that a company listed last year cannot have, whatever its condition. Both of those facts explain more of the sector pattern above than any judgement about the sectors does.

What this leaves out

These are cross-sectional percentiles from one build, dated 2026-09-07, on US listings only, at the verified tier, one company per row. They describe where a score sat, not whether any company is likely to fail and not whether any company is worth owning.

The zone labels are Altman's original cutoffs for the original model on public manufacturers. Later variants exist with different coefficients and different boundaries for non-manufacturers and for private companies. This product computes the original, so the boundaries quoted here belong to it.

The exact number of companies in each zone is not stated. This product's public screener reports at most fifty matching rows to an anonymous caller, so a zone count cannot be reproduced by a reader from the public endpoints, and a figure nobody can check is not published here. The percentiles above are what the public distribution endpoint states, and they bound the zone shares without asserting them.

Percentiles are winsorised at the 1st and 99th before they are computed, clamped to the observed edge rather than dropped, so the cohort size is the cohort that produced them. That clamp sat at -62.71 and 67.26. Two companies were excluded for holding a Z of exactly zero, on the rule that an exact zero here is a missing value rather than a filed one.