Variant Perception

Variant Perception — Where We Disagree With the Market

The market is pricing Somero as a structurally-impaired, no-growth AIM micro-cap — applying a trough multiple to trough earnings, and treating a live governance revolt as background noise. The evidence says two different things. First, this is a double discount: a low-double-digit multiple on a down-year earnings number that is itself understated, attached to the only business in its peer set holding a 52% gross margin and 17% ROIC at the bottom of its cycle, with free cash flow that actually rose to $17.0m and net cash worth ~28% of the market value. Second, the 17 June shareholder revolt is a dated, off-model, asymmetric long catalyst that the price has not moved for. You do not have to win the unknowable cycle-timing debate to see the gap: the market has effectively pre-settled the "structural reset" case while the evidence leaves it open and skewed up.

Currency convention. Somero reports in US dollars; its shares trade in pence (p / GBX) on London's AIM. All $ figures are reported financials; all p figures are prices. FX ≈ 1.27, so 193p ≈ $2.45/share; market cap ≈ £107m ≈ $135m; net cash ~$30m, so EV ≈ $96m.

The variant scorecard

Variant Strength (0–100)

64

Consensus Clarity (0–100)

58

Evidence Strength (0–100)

75

Time to Resolution

Jul–Sep 2026

Source: analyst scoring — strength weighs materiality and evidence against consensus observability; resolution window set by the two dated events (mid-July governance update, ~September H1 results).

The score is deliberately not higher. Two things hold strength to 64: the central cycle-timing question is genuinely unresolved and resolves on forward prints, not on anything in the report; and consensus clarity is only 58 because the name carries effectively one sell-side analyst — the "market view" has to be read off the tape and the multiple rather than a thick estimate distribution. This is a real, monetizable gap, not a slam-dunk.


What the market actually believes

With single-analyst coverage, the most reliable consensus reads here are the tape (range-bound near 52-week lows, death cross, no reaction to the AGM) and the multiple, supplemented by the lone analyst's estimates. Each market view below is nailed to an observable signal.

No Results

Source: Research, Financials, Technicals and Catalysts tabs; price/multiple data as of 19 Jun 2026. "Consensus" on a single-analyst name is read primarily off the tape and the multiple.

The picture is internally consistent: a price stuck near 52-week lows on a falling 200-day, a low-double-digit multiple, a no-reaction AGM, and a cut dividend together describe a market that has filed Somero under "structurally challenged, leave it alone." That is the belief our evidence attacks.


The disagreement ledger

Two disagreements survive all five tests (consensus analyst view → contradicting evidence → material → resolvable on an observable signal → falsifiable). A third candidate — that the recurring aftermarket annuity is under-weighted — folds into Disagreement 1 as evidence rather than standing alone; the cycle-direction call itself is not claimed as edge: it is unknowable and forward.

No Results

Source: synthesized from the Financials, Moat, Forensic, People, Research and Catalysts tabs; ranked by expected value to a PM's underwriting.

Disagreement 1 — the double discount

What consensus would say: "Revenue has fallen three straight years to FY2015 dollars, management guides FY2026 lower and refuses to call a recovery, and they cut the dividend ~40%. This is a melting cyclical; ~$10m net income is the base and a low-double-digit multiple is fair."

Why our evidence disagrees: The market is pricing trough earnings and discounting the multiple for the same trough, while ignoring the quality of the cash underneath. Three evidence layers:

Loading...

Source: derived — 193p ≈ $2.45 at FX 1.27, divided by reported EPS ($0.18), adjusted EPS ($0.20, per Forensic), and mid-cycle normalized EPS (~$0.42, from ~$25m mid-cycle net income on ~54.6m shares, Financials/Long-term thesis tabs).

First, the denominator is wrong on quality: gross margin held in a 52–58% band for a decade as revenue halved — what fell was the operating margin, on fixed SG&A, not the gross margin, which is the moat's signature. Second, the denominator is wrong on level: reported EPS is further depressed by a one-off $0.86m valuation-allowance tax charge (effective rate jumped to ~33% from ~22%), so the cash economics already beat the headline. Third, the cash is real and abundant at the trough: FCF rose to $17.0m (66% above net income, CFO/NI 1.75x), capex ran under 1% of sales, and the balance sheet holds $33.2m net cash — ~28% of the market cap — so EV is only ~$96m against through-cycle EBITDA of ~$30m, i.e. ~3x mid-cycle.

What the market must concede if we are right: that you cannot value a 52%-GM, 40%-through-cycle-ROIC, net-cash franchise off a single down-year's reported earnings — and that even granting a flat-revenue world, the trough cash flow comfortably funds the wait. The bear's 145p target requires the conjunction of a structural reset and ignoring the cash floor; the evidence makes that conjunction unlikely.

Cleanest disconfirming signal: gross margin breaking below ~50% on a non-volume (mix/price) basis — the single most direct moat tell. That would mean Ligchine pricing or Hammerhead dilution is eroding the pricing power that is the entire quality case, turning "trough" into a genuine reset.

Disagreement 2 — the off-model governance option

What consensus would say: "Activists make noise at AGMs all the time; the board is a Delaware corporation that can re-seat directors on plurality and respond advisory-only. The stock didn't move on the vote, so there's nothing here."

Why our evidence disagrees: This is the inverse of a typical overhang. The positioning is concentrated long and accumulating into the fight — Kelly to 12.4% (from 3.4% a year earlier), VN Capital crossing a threshold on 14 April 2026 — and the board itself conceded the dissent reflects "governance arrangements, legal constitution and capital allocation," committing to a dated mid-July response. The grievances are precisely the self-help levers bulls have wanted for years (redomicile, US listing, majority voting, a capital-return reset). With one analyst, no model carries any probability for that path, and the flat price proves the market is pricing neither tail of a discrete binary roughly three to four weeks out — backstopped by a net-cash floor that caps the downside.

What the market must concede if we are right: that a Delaware-on-AIM constitution is a fixable structural discount, not a permanent governance-quality haircut, and that a thin ~33m float means even a modest structural concession re-rates the shares before the September numbers land.

Cleanest disconfirming signal: the mid-July update arrives advisory-only — resolutions reaffirmed on Delaware plurality, activists placated with tokens, no redomicile or capital-return policy. That confirms the board can ignore a 68%-against vote and the option is worth little.


Classifying the variant views

No Results

Source: classified against the eight high-quality variant buckets; banned weak forms ("high quality but undervalued", "market too pessimistic", "execution risk remains") are explicitly excluded.

Neither view is "the stock is cheap." Disagreement 1 is a measurable claim about the quality and level of the earnings the market is capitalizing; Disagreement 2 is a probability claim about a dated event the only covering model cannot represent. Both are falsifiable on named signals.


The evidence layer a PM can audit

The items below are the ones that actually move the probability of the variant view — each with its consensus read, the variant read, and, crucially, its fragility.

No Results

Source: Financials, Forensic, Moat, Business, People, Competition and Research tabs. Fragility column is the honest counter to each item.


How this gets resolved — observable signals only

Every signal below is observable in a filing, an RNS, a half-year income statement, or a holdings notification. None is "better execution" or "time will tell."

No Results

Source: forward calendar verified against the 17 Jun AGM RNS (July update committed), the historical reporting pattern (H1 interims early-mid September), and live buyback/holdings filings.


Red team — what would make us wrong

This is written to kill the view, not protect it.


The one signal to watch first

If you put a single line on the watchlist today, make it the mid-July 2026 board update — it is the nearest dated event (~3–4 weeks out) and entirely off-model; on a ~33m-share float a structural concession (redomicile, US listing, or a published capital-return policy) is the condition that would re-rate the structural discount before the September demand print lands. But the event that decides the larger thesis is the ~September H1 result: a second consecutive up-half in North American Boomed + Ride-on screed revenue with gross margin holding at or above 52% is what turns "trough on trough" from a variant view into a consensus one — and a non-volume break below 50% is what proves us wrong.