RaoscaffResearch
Prediction Series · Lock · Issue P-339
Prediction Series · P-339

Guidance as a floor, not a range.

91% probability Heartland Group's FY2027 underlying NPAT reaches at least NZ$92m, against guidance of at least NZ$102m excluding TSB transaction costs. FY2026 underlying NPAT nearly doubled to NZ$90.4m — while reported NPAT was NZ$93.2m.

Type · Prediction Lock · floor-guidance anchored, banking Locked · 2026-09-04 · before the FY2027 result Resolves · ~2027-08-17 · Heartland Group Holdings FY2027 results (heartlandgroup.info) Scored · binary: reported FY2027 UNDERLYING NPAT >= NZ$92m yes/no
Heartland FY2027 underlying NPAT · our locked floor
NZ$92m
full-year underlying NPAT · vs at least NZ$102m guided

Heartland Group Holdings reported FY2026 underlying net profit after tax of NZ$90.4m, up from NZ$46.9m a year earlier, while reported NPAT came in at NZ$93.2m. Underlying return on equity improved 286 basis points to 7.1%, with reported ROE at 7.3%. The group's average underlying net interest margin widened 36 basis points to 3.98% on lower funding costs, and the impairment expense ratio decreased 55 basis points to 0.45% as Heartland Bank significantly strengthened asset quality. For FY2027 Heartland guided to underlying NPAT of at least NZ$102m, excluding any costs related to the proposed TSB transaction, alongside a target of return on equity of at least 7.5%. The board continues to target a total dividend payout ratio of at least 50% of NPAT in FY2027, excluding exceptional items. Source: Heartland Group Holdings FY2026 annual results.

— 1 · The Locked Call

Heartland's FY2027 underlying NPAT is at least NZ$92m — P = 0.91.

We lock a binary: Heartland Group Holdings' reported underlying NPAT for fiscal 2027 is NZ$92m or higher. Confidence 91%.

Guidance is at least NZ$102m. Our threshold sits NZ$10m, about 9.8%, below that — and NZ$1.6m above the NZ$90.4m banked in FY2026. The lock asks for a fractional improvement on a year in which underlying profit nearly doubled.

— 2 · A fifth kind of guidance

Not a range, not a point — a floor.

This series has now encountered five distinct guidance architectures, and each one changes how a lock should be built.

Most companies publish a RANGE: Auckland Airport's NZ$290-330m, Vector's NZ$540-560m, Adani Ports' INR 25,000-26,000 crore. Mercury published a single POINT, NZ$1,075m, conditioned on 4.1 terawatt hours of hydro. Sun Pharma published a PHRASE — high single digits — with no number at all. Wipro publishes a one-quarter-ahead ABSOLUTE DOLLAR range, the most precise form we have seen.

Heartland publishes a FLOOR: underlying NPAT of AT LEAST NZ$102m. There is no upper bound, and that asymmetry is informative. A company that says at least is telling you where it believes the downside sits, which is exactly the quantity a floor-style lock needs. It is more useful to this series than a midpoint, because a midpoint has to be discounted for the possibility that management centred it optimistically.

So the buffer here is 9.8% below a stated minimum rather than below the bottom of a range that management may have set generously.

— 3 · Three numbers the criterion has to separate

Underlying NZ$90.4m, reported NZ$93.2m, and a pending acquisition.

Heartland's FY2026 produced underlying NPAT of NZ$90.4m and reported NPAT of NZ$93.2m — a gap of NZ$2.8m, with reported the HIGHER of the two, which is the opposite of the usual direction and easy to get backwards. Underlying ROE was 7.1% against a reported 7.3%.

There is a third complication that matters more. The FY2027 guidance is explicitly stated EXCLUDING any costs related to the proposed TSB transaction. If that acquisition proceeds, transaction and integration costs will land somewhere, and reported and underlying will diverge by more than the NZ$2.8m seen in FY2026. A lock written loosely as Heartland FY2027 profit above NZ$92m could resolve HIT on the guidance basis and MISS on the reported basis in the same annual report.

The criterion below therefore names UNDERLYING NPAT on the guidance basis, excluding TSB transaction costs, and rules out reported NPAT in terms. This is the seventh distinct non-IFRS or adjusted-measure convention this series has had to name explicitly, after underlying versus reported EBITDA at SkyCity, normalised versus statutory at Skellerup, adjusted at Vector, EBIT ex-property at Fletcher, standalone versus consolidated across three Indian tranches, and JSE headline earnings at Standard Bank.

What supports the floor is the operating direction. Net interest margin widened 36 basis points to 3.98% on lower funding costs, and the impairment expense ratio fell 55 basis points to 0.45% as asset quality strengthened — margin up and credit losses down at the same time. The residual 9% is a New Zealand credit deterioration reversing that impairment improvement, or a funding-cost turn compressing the margin.

Locked on 2026-09-04 — scored against Heartland's reported FY2027 underlying NPAT.

RAOSCAFF locks P-339 on 2026-09-04, before the FY2027 result. Scored against underlying NPAT as reported by Heartland Group Holdings for the year to 30 June 2027.

Locked
2026-09-04 (commit timestamp on origin/main)
Resolves
~2027-08-17 — Heartland Group Holdings Limited FY2027 annual results
Source
Heartland Group Holdings Limited FY2027 annual results, underlying NPAT in New Zealand dollars (heartlandgroup.info investor centre)
Scored by
Binary: YES if reported FY2027 UNDERLYING NPAT is NZ$92m or greater; NO if below. UNDERLYING NPAT on the same basis as the at-least-NZ$102m guidance and the NZ$90.4m FY2026 comparative, EXCLUDING any costs related to the proposed TSB transaction as the guidance does — explicitly NOT REPORTED NPAT, which was NZ$93.2m against an underlying NZ$90.4m in FY2026. Also NOT underlying or reported ROE (7.1% and 7.3%; a separate target of at least 7.5% applies), NOT net interest margin (3.98%), NOT the dividend payout ratio, and NOT net receivables or lending volumes. Heartland's financial year ends 30 JUNE 2027.

Range, point, phrase, quarterly absolute, and now floor. Five guidance architectures, five ways to build a threshold.