Corporate credit data — the SMarkets Grade

A credit grade computed only from the statements a company filed itself. Every threshold, every weight and every input line item is on this page, so you can recompute the grade by hand and disagree with it. No analyst judgement, no sector premium, no forecast. Version 2026-09-15.

6measures
4minimum to be graded
SM1–SM9scale, plus NR
0graded today, of 2,709

Read this before anything else: today the grade covers nothing

0 of 2,709 Korean listed companies currently receive a grade. Not because those companies are weak — because our tape does not yet carry the line items the grade needs. 2,507 companies have 2 of the 6 measures available, and the floor is 4. We publish that number rather than lowering the floor to make it look better.

These are the line items that are missing and what each one blocks. They come from the same regulator API we already use; it is a question of requesting more accounts per company, not of a new source or a new licence.

Line item not yet collectedWhat it blocks
Finance costsOperating income to finance costs (weight 30)
Short-term borrowings, current portion of long-term debt, long-term borrowings, bondsTotal borrowings to total assets (weight 20)
Current assets, current liabilitiesCurrent assets to current liabilities (weight 15), and the working-capital term of the Altman Z
Retained earningsThe retained-earnings term of the Altman Z (weight 10)

Two measures — liabilities to equity, and operating income to total assets — can already be computed for 2,507 companies from what we hold. They are not published as a grade, because a grade built on two of six measures is not a grade.

The scale

SM1 is the strongest and SM9 the weakest. A company that cannot be measured is NR, never a low grade. We use our own symbols rather than the letter grades the established agencies use, precisely so that nobody reads our number as one of theirs.

The 6 measures, with every threshold

Each measure is scored 1 to 9 against the thresholds below, then averaged using the weights. The thresholds read left to right from strongest to weakest: a value that clears the first threshold scores 1, the second scores 2, and a value past the last threshold scores 9.

MeasureWeightDirectionThresholds (score 1 → 8)
Operating income to finance costs30Higher is better12 · 8 · 5 · 3 · 2 · 1.5 · 1 · 0.5
Total borrowings to total assets20Lower is better0.05 · 0.1 · 0.18 · 0.25 · 0.35 · 0.45 · 0.55 · 0.7
Total liabilities to equity15Lower is better0.3 · 0.6 · 1 · 1.5 · 2 · 3 · 4 · 6
Current assets to current liabilities15Higher is better3 · 2.2 · 1.8 · 1.5 · 1.2 · 1 · 0.8 · 0.6
Operating income to total assets10Higher is better0.15 · 0.11 · 0.08 · 0.055 · 0.035 · 0.02 · 0.005 · -0.03
Altman Z (1968 formula)10Higher is better6 · 4.5 · 3.5 · 2.99 · 2.4 · 1.81 · 1.2 · 0.5

Finance costs, not interest expense. Where a company does not file a separate interest expense line, finance costs are used. Finance costs include items other than interest. We use the wider line because some filings do not break interest out, and we say so rather than substituting a guess.

How the average is taken — and why this part matters

Each measure is scored 1-9 against the published thresholds below, then averaged using the published weights. Weights of measures we could not compute are removed from the denominator rather than scored zero.

That last sentence is the whole difference. If a measure cannot be computed, its weight is removed from the denominator. Scoring it zero instead would mean a company with an unusually formatted filing quietly grades worse than an identical company whose filing we happened to parse. That is a defect dressed up as a rating.

The Altman Z is not ours

One of the 6 measures is the 1968 Altman discriminant function. We cite it in full and we compute it only where all five of its terms are present:

Altman, E. I. (1968), "Financial Ratios, Discriminant Analysis and the Prediction of Corporate Bankruptcy", The Journal of Finance 23(4), 589-609

What the agencies themselves did — the rating actions we track

Separately from our own grade, we archive the rating actions that exchanges disseminate. This is not a copy of an agency table: these are counts we computed from 21,956 archived disclosure rows, and we only count an action where the agency itself labelled it an upgrade or a downgrade. The 17,058 rows filed as “Other” are left uninterpreted rather than guessed at.

3,301rating moves counted
722downgrades
21.9%of moves were downgrades
130companies cut more than once

The share differs sharply by agency, and that spread is itself the finding — the same universe of borrowers, rated by different houses, produces downgrade rates that differ by a factor of three.

AgencyUpgradesDowngradesDowngrade share
CRISIL1,26725016.5%
ICRA5648713.4%
CARE40019733%
India2096423.4%
Infomerics847747.8%
Acuite544545.5%

Dates that the source filed impossibly — a year in the future, for instance — are flagged and left as filed (2 of them) rather than silently corrected.

What this is not

  • This is the SMarkets Grade. It is our own measure and is not the rating of any licensed credit rating agency.
  • Everything needed to reproduce it is published here: the inputs, the thresholds and the weights.
  • No analyst judgement, no sector premium, no management meeting, no forecast enters this grade.
  • A company with missing figures is NR, never a low grade.
  • Not investment advice.

Related

The statements this grade is built on: the financial-statement tape. Ratios from the same rows: the valuation tape. Filter companies yourself across Korea and the Gulf: the screener. See what else is in the catalogue.