The 24.8% Poison Apple: Hong Kong's Signature Distortion

A 24.8 Percent Cohort with One Specific Shape

Six hundred eighty-six Hong Kong-listed equities — 24.8 percent of the panel^[1] — carry the framework's Poison Apple archetype assignment. The share is the highest the framework has observed across the eight Asian markets it covers.

The cohort's mechanical signature is sharp. Its Transparency axis mean is 83.4, eleven points above the universe mean. Its Conflict-of-Interest Risk axis mean is 46.9, ten points below the universe mean. Its Balance of Power axis mean — 61.1 — sits a few points below the universe baseline. Mean composite 63.5.

A cohort whose disclosure axis runs eleven points above market average and whose conflict-of-interest axis runs ten points below it is not a cohort whose problem is hygiene. It is a cohort whose disclosure quality clears the HKEX listing-rule baseline reliably while the substantive transactions and control structures behind that disclosure are systematically more exposed than the universe baseline. Twenty-five percent of the Hong Kong universe sits in this configuration.

This is the signature distortion of HKEX governance. It is not the same problem as the 52.2 percent Chameleon dominance Note 1 examined^[2] or the five-origin variation Note 2 mapped^[3]. It is a third pattern, defined by a sharp axis asymmetry that the framework's three-axis structure surfaces and that composite-only scoring would conflate into a mid-tier band.

The Poison Apple Profile

The framework's Poison Apple archetype is reserved for issuers whose three-axis profile carries a specific shape: high disclosure, weak substance. Within the Hong Kong panel, the shape is well-defined.

The composite means of the Poison Apple and Chameleon cohorts are essentially identical — 63.5 versus 63.6. A composite-only ranking would place the two cohorts in the same band. The axis structure is fundamentally different. Poison Apple sits at +11.0 on T, −3.9 on B, and −9.7 on R relative to the universe; Chameleon sits at −2.6 on T, −1.4 on B, and +0.6 on R. The Chameleon cohort is, on average, slightly weaker than market on every axis but mostly close to it. The Poison Apple cohort is sharply elevated on disclosure and sharply depressed on conflict-of-interest, with the gap between its T and R means averaging 36.5 points.

This is the configuration the framework's three-axis architecture is designed to surface. A composite produced from any reasonable axis weighting averages over the asymmetry. The asymmetry itself is the finding.

The Distribution: Five Origins, Different Densities

The 686 Poison Apple issuers do not distribute evenly across the five Origin Types.

Two readings sit on top of one another. By count, Poison Apple is dominantly an HKL cohort — 534 of 686 firms (77.8 percent) sit in HK-Local. The universe-level Poison Apple share, like the universe-level Chameleon share, is largely an HKL pattern projected through HKL's 84 percent share of the panel.

By within-origin density, however, the picture inverts. HKL's within-origin Poison Apple rate of 23.0 percent sits below the universe average of 24.8 percent. HSH's within-origin rate of 62.5 percent is the panel maximum and the highest within-origin Poison Apple density the framework has observed across its eight Asian markets. RCH's 45 percent and PCH's 31.6 percent both sit above the universe baseline. FOR has no Poison Apple firms.

The HSH concentration is the cohort-level anchor of this Note. Twenty-five of forty H-share issuers — almost two in three — carry the Poison Apple assignment. Within Hong Kong, this is the cleanest within-origin signal the framework produces. The mechanical reason follows from the cohort's axis profile, examined directly in the next section.

Origin-Level Axis Profiles: T Converges, B Diverges

The four Origin Types with non-zero Poison Apple populations carry distinct axis profiles within the cohort.

Transparency converges across origins. The four PA subsets cluster within a 3.6-point band (82.7 to 86.3) on the T-axis, all well above the universe T mean of 72.4. The Poison Apple archetype enforces a high-disclosure baseline as part of its definition; issuers without strong disclosure register elsewhere in the archetype taxonomy.

Balance of Power diverges sharply. HSH PA's B mean of 41.1 sits 23.5 points below HKL PA's B mean of 64.6. The HSH PA subset is, by axis-mean, the most board-deficient cohort the framework reads in the Hong Kong universe — even further below the HSH origin's already-low B baseline (46.1) than the cohort average. The HSH Poison Apple cohort is the cohort within the cohort whose B-axis sits at the panel floor.

Conflict-of-Interest Risk runs in the opposite direction within the PA distribution. HKL PA's R mean of 45.8 is the lowest of the four PA subsets — substantially below the universe R baseline of 56.6 — while HSH PA's R mean of 58.0 sits at the universe baseline. The framework reads the HKL Poison Apple cohort as principally R-axis-weak and the HSH Poison Apple cohort as principally B-axis-weak. The same archetype label, in other words, captures structurally different distortions across origins.

The Poison Apple as an HKEX-wide pattern is the average of two distinct mechanisms. HKL PA — by far the largest subset — is the high-T, low-R pattern that drives the universe-level R asymmetry. HSH PA — small in count but extreme in axis profile — is the high-T, low-B pattern that gives the cohort its panel-maximum within-origin density.

PCH PA and RCH PA sit between the two, with B-axis means of 50.7 and 50.0 respectively and R-axis means of 49.3 and 55.2. The mainland-linked cohorts (PCH, HSH, RCH together) carry their PA distortion primarily through the board axis; the HK-Local cohort carries it primarily through the conflict-of-interest axis. A single archetype label averages both patterns, but the substantive governance signal differs by origin.

The R-Axis Mechanism: A Capped Distribution

The R-axis weakness in the universe-level Poison Apple cohort is not statistical noise. It has a measurable mechanical driver.

The framework's R-02 indicator measures the scale of related-party transactions relative to revenue, drawn from each issuer's connected-transaction disclosures under Listing Rules Chapter 14A^[4]. Of the 686 Poison Apple firms in the universe, R-02 is measurable for 519 (75.7 percent); the remainder either lack sufficient disclosure granularity or carry zero-revenue or non-trading periods that make the ratio undefined. The 519 covered subset is the cohort within which the R-axis mechanism can be observed directly.

Ninety percent of the covered Poison Apple subset carries an RPT-to-revenue ratio at the indicator's upper measurement bin (50 percent or higher). The cohort is not distributed across the RPT-ratio range; it concentrates at the top of it. Mean RPT-to-revenue ratio in the PA cohort is 0.905; median is 1.000. By comparison, the Chameleon cohort's covered subset (n=448 at the upper bin, mean 0.730) carries lower RPT exposure on average, and the combined cohort of Celestial, Hidden Gem, Time Bomb, and Kill Switch firms (mean 0.448, 91 firms at the upper bin) carries materially lower exposure still.

The Poison Apple cohort, in short, is the cohort whose connected-transaction exposure runs up against the measurement ceiling. The R-axis weakness is the framework's reading of that exposure. Chapter 14A disclosure surfaces the underlying transaction streams; it does not constrain them. The cohort that carries the heaviest transaction loads, relative to the revenue base of the listed vehicle, registers the cleanest R-axis signal the framework produces.

A Distinctive Concentration Across the Panel

The Hong Kong Poison Apple share of 24.8 percent is the panel maximum.

The Korean panel of 2,662 listed companies carries 14 Poison Apple issuers — a 1.4 percent share, against Hong Kong's 24.8 percent^[5]. The cohort-ratio gap of approximately 49 times is not a function of market size; the two markets are comparable in count. It is a function of how the same archetype taxonomy reads two different governance environments.

Both markets are characterized by controlling-shareholder concentration. Korea's controlling-shareholder pattern operates predominantly through the chaebol family-conglomerate structure, where the Korean Commercial Code mandates audit committee composition but leaves much of the substantive board architecture to disclosure practice. The Korean Poison Apple cohort's axis profile — Transparency mean approximately 76, Balance of Power approximately 46, Conflict-of-Interest Risk approximately 74 — describes a transparent-but-board-weak pattern. Korean PA firms sit on the high side of disclosure and the high side of conflict-of-interest controls, with board governance as the binding axis.

Hong Kong's Poison Apple cohort runs in the opposite direction. Transparency is high (83.4) and Balance of Power sits close to the universe baseline (61.1), while Conflict-of-Interest Risk is the deeply weak axis (46.9). The Hong Kong PA pattern is transparent-but-conflict-weak. The same archetype label, in other words, captures inversely structured distortions across the two markets — both are markets where formal disclosure compliance coexists with substantive controlling-shareholder reality, but the axis on which substantive control bites is different.

The cross-market comparison reinforces what Note 1 framed as a regional cluster. Markets where formal compliance and substantive control diverge produce Poison Apple populations of meaningful size. Hong Kong's panel-maximum share reflects, on this reading, the combination of an unusually well-developed listing-rule disclosure regime with a particularly heterogeneous mix of controlling-shareholder structures — HK family conglomerates, Cayman/BVI mainland-private founder blocks, and direct-listed mainland state-supervised parent vehicles, each of which generates ongoing related-party transaction streams that the disclosure regime surfaces but does not constrain.

The Time Bomb Mirror: One Issuer, 0.04 Percent

The framework's Time Bomb archetype is the inverse of Poison Apple. Time Bomb is reserved for issuers whose three-axis profile carries weakness on every axis simultaneously. In the Hong Kong panel of 2,768, exactly one issuer carries this assignment. The cohort is 0.04 percent of the universe.

The single Time Bomb firm sits in the HKL origin. Its axis profile (T 43.3, B 45.0, R 46.7) places all three axes below the framework's threshold for axis-level adequacy. By comparison, other markets in the framework's panel typically carry Time Bomb populations in the 0.5 to 1.0 percent range. Hong Kong's near-zero count is structurally informative.

The HKEX listing-rule architecture sets a disclosure floor that lifts the Transparency axis well above the failure threshold for almost every issuer that maintains a current listing. An issuer whose disclosure quality drops below that threshold typically does not maintain its current-listing status — the firm receives a suspension notice under Listing Rule 6.01^[6], undergoes Kill Switch criteria under the framework's separate override taxonomy, or both. The space the Time Bomb archetype occupies in other markets — issuers persistently weak on every axis but remaining listed and trading — is structurally compressed in Hong Kong by the listing-rule architecture itself.

The Poison Apple population is, in this reading, the architectural complement to the absent Time Bomb population. Issuers whose underlying governance might present as uniform-axis weakness in another listing regime present in HKEX as the high-T-asymmetric pattern — disclosure compliance preserved on the listing register, with the substantive weakness routed to the B or R axis instead. The 686 Poison Apple firms are what the disclosure floor preserves on the listing register; the single Time Bomb firm is what it does not. Both findings reinforce the same structural reading. The HKEX listing-rule architecture surfaces governance weakness reliably on the Transparency axis. It does not, by itself, constrain the substantive weakness it surfaces.

Where This Sits in the Series

Three findings stand from the Poison Apple distribution.

First, the cohort is large and structurally distinctive. Twenty-five percent of the Hong Kong universe carries an axis profile in which disclosure runs eleven points above market average and conflict-of-interest runs ten points below. Composite-only rating systems classify these issuers in a mid-tier band; the framework's three-axis architecture surfaces the asymmetry as a specific governance signature.

Second, the cohort runs through two distinct sub-mechanisms by origin. HKL Poison Apple — the bulk of the universe-level cohort — carries the high-T-low-R pattern, with R-axis weakness driven by Chapter 14A connected-transaction exposure that registers at the framework's measurement cap in 90 percent of the covered subset. HSH Poison Apple — small in count but the panel-maximum within-origin density — carries the high-T-low-B pattern, with B-axis weakness driven by direct mainland incorporation and state-supervised parent-shareholder structures. The two sub-mechanisms produce the same archetype assignment but reflect different governance environments.

Third, the absence of a meaningful Time Bomb population in Hong Kong is the architectural complement to the Poison Apple presence. Issuers whose underlying governance would otherwise present as uniform-axis weakness in another listing regime register in HKEX as the high-T-asymmetric pattern — disclosure compliance preserved on the listing register, with the substantive weakness surfacing on the B or R axis. The framework's reading is that the listing-rule architecture displaces the visibility of poor governance from the disclosure axis to the substantive axes.

The remainder of the series develops the regulatory architecture this Note has invoked. Note 4 turns to the Independent Director regime under the post-2023 nine-year tenure cap, examining how the formal board-architecture requirements intersect with the B-axis weakness identified in the mainland-linked Poison Apple subsets. Note 5 takes up Chapter 14A connected transactions directly, examining the percentage-ratio thresholds and aggregation rules that govern when Majority-of-Minority approval engages, and how the resulting transaction architecture relates to the R-axis weakness driving the HKL Poison Apple pattern. Note 6 establishes a baseline for the climate disclosure mandate that took effect within the panel's coverage period.

The Poison Apple cohort identified in this Note is, with the R-weak Chameleon spine identified in Note 1 and the five-origin map of Note 2, the third structural finding the series puts on the table. Together, the three notes describe what the framework reads when it scores the HKEX equity universe against its three-axis architecture. The remaining notes turn from description to mechanism.

The Apex G-Score framework currently covers 2,768 Hong Kong listed companies under v2.0 calibration. The R-02 indicator (Connected Transactions / Revenue ratio) is parsed from issuer continuing-connected-transaction disclosures under HKEX Listing Rules Chapter 14A; universe coverage of R-02 is approximately 60.8 percent (n=1,683 of 2,768) under the v1.5 parser, and PA-cohort coverage is 75.7 percent (n=519 of 686). The Korean Poison Apple cross-market reference draws on Apex Governance Substack content dated 29 April 2026; verification of Korean panel figures is recommended at publication.