Decoding RaymondsIndex: What "Validation" Actually Means for a Leading Risk Score

This week gave us a clean picture of how markets discover risk the slow way. Korea's exchanges classified 284 listings as a potential risk group — 74 on KOSPI and 210 on KOSDAQ — as the KOSDAQ index sat below 1,000 points for more than a month. Alongside it: 137 KOSDAQ names trading under ₩1,000, and a rule change that pulls full-capital-erosion companies into delisting review from the August half-year reports. Roughly 50 KOSDAQ names are already at delisting risk on the market-cap threshold alone.

Read that list carefully and you notice something: every criterion on it is backward-looking. A share price under ₩1,000. A market cap beneath the line. Negative shareholder equity. These are outcomes — the last visible symptoms of decisions that were made quarters earlier. By the time a company qualifies for the "risk group," the capital efficiency has already decayed, the raised cash has already gone idle, the reinvestment has already been skipped. A lagging screen is honest, but it is a receipt. It records what already happened; it does not warn.

The claim a leading score has to earn. A "leading" indicator promises something much harder: that it can separate the companies that will fail from the ones that won't, before the financial statements make it obvious. That is a testable promise, and the test has a name — concordance.

Concordance asks a simple question. Take every company, score it today, and rank them from most to least risk. Then wait, and observe who actually ran into distress. If your ranking is any good, the firms that failed should have been sitting near the top of your risk ordering before the fact. Concordance is the share of comparisons where the eventually-distressed company was ranked riskier than the surviving one. A coin flip scores 50%. A useful model scores meaningfully higher.

Korea parallel. This is exactly how RaymondsIndex is evaluated rather than merely asserted. Across a universe of 3,109 Korean listed names, the relational risk score reached 85.9% concordance with realized distress outcomes, and the separation between failed and surviving firms carried an effect size above Cohen's d > 0.8 — conventionally a "large" effect. The number that matters here is not any single company's score; it is whether the ordering held up against what the market later confirmed the slow way.

The academic frame. None of this is new as a discipline. Altman's original Z-score work (Altman, 1968, Journal of Finance) showed that a weighted combination of accounting ratios could discriminate bankrupt from solvent firms with far better than chance accuracy — the founding idea that distress is predictable, not just diagnosable. Shumway (2001, Journal of Business) sharpened it, showing that hazard models using time-varying signals forecast bankruptcy more accurately than static single-period scores, precisely because risk is a moving relationship, not a snapshot. Concordance (closely related to Harrell's c-index) is the standard yardstick for exactly this kind of ranking problem. RaymondsIndex extends the tradition from accounting ratios toward the relational structure around a firm — ownership, capital, and governance behavior — on the same evidentiary standard: rank first, then let realized outcomes grade the ranking.

What it means for an individual investor. The 284-name list will keep growing as the August reports land, and each new entry will arrive with the same problem — it is already priced in by the time you can read it. The practical question is not "which names are on the public risk list today?" It is "whose relationships are drifting toward that list a year before the screen catches them?" That is the gap a validated leading score is built to close. So here is the honest one to sit with: if the only risk signals you act on are the ones that made the newspaper, whose losses are you actually underwriting?

(General market commentary, not investment advice.)

#RaymondsRisk #RelationalRisk #CorporateGovernance #LeadingIndicators #ModelValidation #KoreaEquities