Variant Perception

Where We Disagree With the Market

Consensus is collapsing a 5-to-10-year compounder thesis into a single H1 FY2026 margin print, and is anchoring its model on Unilever-era trajectory rather than the standalone tape that already exists. The €16.12 share price, the 10.4× EV/Adj EBITDA multiple, and the Q1 2026 published consensus (OSG 2.6% / OVG 0.2%) all point to a market that has priced MICC as a fragile carve-out whose only relevant data point is whether the September 2026 print clears 16.5% Adjusted EBITDA margin. The 30 April Q1 print delivered organic volume growth fourteen times the consensus assumption and the stock printed +14.3% on 4.6× ADV — yet the multiple has barely moved off the post-spin range, suggesting consensus models still extrapolate the 2013–2023 share-loss trajectory rather than re-baselining to the standalone evidence on the tape. The real underwriting variable is whether the €656M FY25 adjusting-items cushion compresses below ~€200M by FY27 — a question the H1 print cannot answer and that resolves in Q1 2028. We carry a Medium-confidence variant view: the consensus framing is too narrow on horizon and too cautious on AMEA mix-shift, but the bear's structural ammunition (governance, overhang, peer-trailing reported margin) is real enough that we do not yet underwrite an above-consensus position.

Variant Perception Scorecard

The 62 variant strength score reflects a real disagreement on horizon and segment economics, capped by three concrete limitations: (i) the web-research provider was offline this run, so the published Q1 consensus document is our only direct read on sell-side framing — no broker initiations, no published model bridges, no Visible Alpha; (ii) the bear case carries genuine structural ammunition that we cannot dismiss with a confidence above 60; and (iii) the cleanest resolution signal (FY27 adjusting-items disclosure) does not land until Q1 2028, so the variant view will live for ~21 months without a single decisive print. Consensus clarity is 55 because the only verifiable consensus signal is the Q1 2026 company-compiled document plus the 10.4× EV/Adj EBITDA implied multiple; sell-side rating distribution and price-target dispersion are not retrievable.

Consensus Map

The most observable consensus signal is the Q1 2026 company-compiled OSG/OVG consensus that management published before the trading update — and the gap between that document (2.6% / 0.2%) and the actual print (4.5% / 2.9%) is the single cleanest measurement we have of where sell-side modelling sat at the moment Q1 broke. The market's relatively muted multiple response since (+14.3% on the day, then range-bound) tells us that even after the beat, consensus has not re-baselined the trajectory — analysts appear to be treating Q1 as Easter timing rather than algorithm.

The Disagreement Ledger

Disagreement #1 — Wrong time horizon. Consensus would say the H1 FY2026 print on or about late July/early August is the single bridging event between carve-out plumbing and standalone economics, and it has built valuation around 16.5% as the threshold between Bull and Bear. Our evidence is that the H1 print can rationalise either side because management has already pre-warned H2 phasing — only the FY27 disclosure ~Q1 2028 collapses the structural-vs-transitional debate. If we are right, the market is doing 12-18 months of trading on a non-decisive datum and the durable rerate window opens later, but cleaner. The disconfirming signal is the FY27 adjusting-items disclosure printing above €350M or showing a new permanent bucket.

Disagreement #2 — Wrong segment focus. Consensus would say MICC is a global ice cream pure-play with a 24% emerging-markets revenue tail, valued in line with global staples that have similar EM exposure. Our evidence is that AMEA is contributing ~34% of group Adjusted EBITDA at a 22.9% margin and 10.9% OSG — those numbers are an emerging-market consumer compounder, not a tail. A SOTP that values AMEA at 17× and the rest of the group at 11× (vs the blended 10.4× the market is paying) lifts equity value by ~€3-4 per share before any algorithm delivery. The market would have to concede that consolidated reporting is structurally masking the highest-quality slice of the business. The cleanest disconfirming signal is AMEA OVG falling below 3% for two consecutive periods or AMEA margin slipping below 20%.

Disagreement #3 — Wrong quality of earnings (volume signal). Consensus would say Q1 2026 was an Easter-timing-aided print and the durable run-rate is 3-3.5% OSG with low single-digit volume. Our evidence is the magnitude of the consensus miss — published expectations of 0.2% OVG against actual 2.9% — combined with the broad-based regional pattern (AMEA +4.9%, Europe & ANZ +4.3%, Americas +2.6% all positive) suggests sell-side baseline volume estimates are still anchored to the 2013-2023 share-loss era, not to the post-spin reinvestment cycle. If we are right, the 40-60bps comparable margin algorithm is a floor not a target. The disconfirming signal is OVG retreating below 1.5% in any single quarter outside a known Middle East / supply-chain event.

Disagreement #4 — Wrong implementation assumption (overhang). Consensus would say the Unilever overhang is an asymmetric supply problem that caps the rerate even on operational delivery. Our evidence is softer: the two unexplained April-May volume spikes look more like absorbed-at-market clearings than discounted blocks, Unilever voted proportionally at the AGM rather than blocking, and the bond book demonstrated genuine demand for the credit. If we are right, the overhang is a known finite event that absorbs supply rather than capping multiples — but our confidence here is Low because the structural argument (0.214% ADV/mcap, six-month-old listing) is genuine and the variant could be wrong on first principles. The cleanest disconfirming signal is a series of -3% or wider discounted block trades.

Evidence That Changes the Odds

The single most disconfirming piece of evidence for our variant is item #5 — the 95%-of-target bonus paid on a 48% net-income decline. Even if we are right that AMEA is mispriced and the Q1 volume signal is real, a remuneration framework that rewards adjusted outcomes regardless of headline results is exactly the structural ammunition a credible bear would use to argue the €656M cushion is permanent by design. We carry the variant view in spite of this, not because of it; if FY26 bonus pays above 100% on a financial miss, the variant-strength score should drop by at least 15 points.

How This Gets Resolved

Five of these seven signals resolve within 12 months — but only signal #1 (FY27 adjusting items) directly collapses the structural-versus-transitional debate that underpins the most important variant view. Signal #2 (AMEA quarterly cadence) is the most decision-useful continuous signal because every reporting period either compounds or weakens the EM-compounder hidden-value thesis. Signal #3 (OSG vs published consensus) is the cleanest test of whether sell-side models have re-baselined; a Q2 print where OVG holds 2%+ but consensus has not revised upward would tell us our variant is correct but slow to be priced — a frustrating but real outcome for a multi-year holder.

What Would Make Us Wrong

The variant view depends on three load-bearing assumptions that can each fail independently. First, AMEA's 22.9% Adj EBITDA margin and 10.9% OSG signature are assumed to be the steady-state run-rate, not a peak that fades as base effects compound. AMEA carries Türkiye on IAS 29 hyperinflation accounting, has a fresh India acquisition that the company itself describes as loss-making, and competes with well-capitalised local incumbents (Yili, Amul, Mengniu, Lotte) in each of its largest single-country pools. If AMEA volume growth retreats to 3% with the segment Adj EBITDA margin compressing to 20%, the mechanical mix-shift arithmetic that powers the variant collapses and the consolidated multiple consensus is paying becomes the right multiple — not because the market was wrong, but because AMEA is no longer the EM-compounder it looks like on FY25 numbers.

Second, the Q1 2026 volume beat may be Easter-timing plus a clean comparable inside a category that simply lapped a weak prior year. CEO ter Kulve conceded the Easter contribution was "marginal" — but management has every incentive to characterise a beat as durable. If H1 OVG retreats to sub-1.5% with management pointing back to "Q1 was about timing", the consensus model we are arguing against turns out to be approximately right and our variant on the volume signal evaporates. The structural variant on AMEA mix would still survive that outcome, but the near-term path to re-rating would extend further out and the variant strength score should compress.

Third, the €656M adjusting-items cushion may compress mechanically to the FY26 guide (€425-450M) and then stop compressing. Management's bonus framework is designed around adjusted outcomes; the 95%-of-target FY25 payout on a -48% GAAP net income year is the single piece of evidence that says incentive design and "transitional plumbing" are pointing in the same direction. If FY27 prints €350M+ in adjusting items, the bear's structural-cushion thesis is right and our entire variant collapses to a re-statement of the bull case that the market already knows about. The honest read is that we cannot underwrite the variant with high confidence until at least one of (a) FY26 H1 prints with a comparable +50bps walk on adjusting items inside the guided range, or (b) FY27 disclosure lands below €250M. Until then, the variant is a directional view, not a position-sized one.

A fourth risk worth naming: we may be confusing absence of evidence with evidence of absence on consensus. Without retrievable sell-side initiations or a Visible Alpha read, the only consensus signal we have is the Q1 company-compiled document and the implied multiple. Sell-side may have already re-baselined volume estimates after Q1 in models we cannot see; the multiple may already reflect the AMEA mix-shift in a way the headline EV/EBITDA does not show; consensus may be more bullish than we infer. If so, we are not disagreeing with the market — we are agreeing with it more verbosely.

The first thing to watch is the H1 FY2026 reported Adjusted EBITDA margin walk in the company's late-July / early-August disclosure — specifically the comparable-vs-reported reconciliation that decomposes margin change into productivity, cocoa, FX, and India dilution. That single table will tell us whether the algorithm is on track, whether the cushion compression is mechanical, and whether the Q1 volume signal is durable, in a way no other near-term signal can.