Variant Perception

Where We Disagree With the Market

Figures converted from EUR at historical FX rates — see data/company.json.fx_rates. CAD share prices and targets converted at CAD/USD ≈ 0.731 derived from EUR/USD 1.1702 and CAD/EUR ≈ 1.6. Ratios, margins, and multiples are unitless and unchanged.

The sharpest disagreement is on the denominator: every published target on this name anchors on a 16–18× P/FCF print on consolidated cash flow against basic shares, while the economically correct math — Free Cash Flow Available to Shareholders (FCFA2S) on a fully-diluted denominator — is ~48× and the cash that actually reaches public SVS holders is 45% smaller than the headline. The market is also mistreating the FY2025 ROIC collapse: sell-side has read 10.4% as the first observable signature of hurdle-rate compression, while the data fingerprint — $817m of capital deployed in the back half of the year with no time to earn — is overwhelmingly a denominator effect that should mean-revert in FY2026. And the Canadian VMS basket has been de-rated together (Leonard departure + AI fear + Asseco scope creep), pricing TOI as if its operational signals — +6% maintenance organic, peer-best deployment ratio, sibling LMN at +2% — were as soft as sibling LMN, when they are not. The variant view is neither a clean bull nor a clean bear: the operating engine is mispriced upward by the screen-quoted multiple and downward by the basket sell-off, and the resolution comes through ROIC reversion (or its absence), maintenance organic prints, and how sell-side denominators migrate over the next two reporting cycles.

Variant Perception Scorecard

The 62 variant strength reflects that the two sharpest disagreements (denominator and FY25 ROIC interpretation) push in opposite directions on the equity, which is itself an underwriting edge — the variant view is a paired correction, not a one-sided contrarian call. Consensus clarity is high because both sides of the published market view are observable: sell-side has cut three times in four months (RBC US$139 → US$117 → US$110; TD Cowen up to US$106; aggregate ~US$106 average), and a coordinated cluster of independent buy-side substacks (Summit Stocks, The Compounding Tortoise, Outsiders' Corner, Expanse Stocks) anchors the long-side narrative on the wrong denominator. Evidence strength is anchored by primary disclosure (FY2025 MD&A, NCI footnote, Q1 FY2026 results) rather than inference. Resolution in 6 months: Q2 FY2026 results on 31 July 2026 land the load-bearing ROIC observable; the Q3 FY2026 print (4 Nov 2026) laps the Asseco recast and is the first clean comparator.

Consensus Map

The published market view is internally inconsistent: the same sell-side that anchors a Buy on "lowest P/FCF since spin" is simultaneously cutting targets on the basket-driven ROIC narrative. That inconsistency is the structural opening — published consensus has not reconciled its own multiples with its own concerns. Our two highest-conviction disagreements sit on those two specific lines.

The Disagreement Ledger

Disagreement #1 — Wrong denominator. Consensus reads "16x P/FCF, lowest since spin" as a margin-of-safety entry; we read it as a screen artefact. The 1.56× gap between 83.3m basic shares and 129.8m fully-diluted (the Joday/Ijssel/CSI exchangeable Coop units, economically owners of 35.8% of consolidated NI) plus the 45% NCI leakage between $472m consolidated FCF and $257m FCFA2S together compress the true valuation by roughly a factor of three. The cleanest disconfirming signal is a sell-side note that publicly anchors on FCFA2S × diluted — once one major bank does this, peer notes follow, and the screen-quoted multiple loses its anchor without a fundamental catalyst. If a major sell-side house defends the basic-share, consolidated-FCF math explicitly in a follow-on note (as TOI's own controlling-shareholder structure has not changed since the spin), the variant view is weakened — we would have to argue why the market is only now getting the denominator wrong.

Disagreement #2 — Wrong reading of FY25 ROIC. Consensus reads 15.8% → 10.4% as evidence that European VMS multiples have re-rated and TOI is paying more for less. We read the timing of the dip ($817m deployed in H2 FY25 with no full-year earnings contribution) as a textbook denominator effect that will mechanically unwind. Consensus would say the company has never quantified its hurdle, so any ROIC math is inference — true, but the alternative explanation (clean denominator) is the simpler one and is internally consistent with management's Q1 FY26 behaviour (deleveraging breather, $20.6m bolt-ons only) which is what a disciplined hurdle-keeper does, not what a desperate deployer does. The cleanest signal is the 31 July Q2 FY26 ROIC print: above 12% and we are right; sub-11% with deployment still dormant and consensus is right. If the ROIC print comes in at 11–12% — exactly the grey zone — neither side wins and the debate extends to Q3.

Disagreement #3 — Basket discount, idiosyncratic operational quality. The Canadian VMS family has de-rated as a basket on the Leonard + AI + Asseco overhang. Inside the basket, sibling LMN prints maintenance organic +2% and deploys 0.4% of market cap. TOI prints +6% and deploys 6.5%. Consensus discounts both equally because the narrative drivers (Leonard, AI, basket positioning) are family-wide. Our variant view is that operating data should eventually dominate over basket positioning — the 4-percentage-point maintenance-organic gap and the 16× deployment-ratio gap are not narrative; they are quarterly disclosure. The cleanest signal is the relative move of TOI vs LMN inside the next two prints. If LMN improves and TOI rolls — basket wins, our variant view is wrong. If TOI extends operational lead while basket compresses uniformly, this disagreement compounds with #2.

Evidence That Changes the Odds

The audit logic for a PM: rows 1, 2, and 3 are the three load-bearing evidence items. Row 1 (NCI leakage and FCFA2S definition) is observable in the FY2025 MD&A; row 2 (ROIC mechanics) is observable in the balance sheet plus the acquisition timing; row 3 (organic maintenance) is observable in the quarterly revenue tables of TOI, CSU, and LMN. If any of those three reverse, the corresponding variant view weakens. None of the three are inference-only; all three are primary-disclosure.

How This Gets Resolved

What Would Make Us Wrong

The denominator argument is the easiest of the three to overstate. The 35.8% NCI leakage and 1.56× basic-vs-diluted gap have been public since the 2021 spin. If the market has not yet migrated to FCFA2S × diluted, the most charitable explanation is that the market has consciously decided the controlled-co structure does not warrant the discount — Joday and Ijssel are not selling exchangeable Coop units, no dilutive event has occurred for five years, and CSI's parent governance is famously aligned with public shareholders. If that interpretation is right, the "definitional re-rating" we describe never arrives, and the variant view #1 is wrong by being too procedurally cynical. The fragility we have to honor: a stable, structurally-aligned NCI may simply be priced as if it were not there. We would update our view if no major sell-side firm migrates to FCFA2S diluted within 12 months despite the FY25 forensic disclosure, or if a holder-rotation event compresses the spread between basic and diluted P/FCF math without operational catalysts.

The FY26 ROIC reversion argument is empirically falsifiable inside two quarters. If Q2 FY26 ROIC prints sub-12% and Q3 FY26 prints sub-12% again, with H1 FY26 bolt-on M&A below $117m, the "denominator-effect-mean-reverts" thesis has no remaining defence. We would have to concede that either (a) the FY25 acquisition cohort is structurally lower-margin than the historical book and the consolidated invested-capital base is permanently higher relative to operating income, or (b) the hurdle has compressed and management is correctly walking away from deals at the cost of deployment. Both readings — margin-mix and hurdle compression — are bear cases. The variant view fails if Q2 and Q3 FY26 both deliver this combination, and the right institutional response is to migrate the position size toward the bear's downside.

The basket-divergence argument depends on TOI continuing to print ≥5% maintenance organic while LMN stays at ≤2%. If LMN improves into the +4% band on its own (a Q4 FY25 0% print already inverts the trend), the operational gap that justifies idiosyncratic TOI alpha closes from 4 points to under 1. The basket then converges and our claim that TOI's quality justifies an undeserved discount loses force. The cleanest disconfirming event: a LMN MD&A in Q3 FY26 showing maintenance organic recovery to +4-5% combined with TOI rolling below +5%. That is observable, not inferred, and lands inside our 6-month window.

Beyond these three, the broader fragility is that consensus could be right on direction and we could be right on math. A sell-side analyst who is bullish on TOI for the wrong denominator reason and a bear who is bearish for the wrong ROIC reason can still both be cleaner directional bets than a variant view that requires the reader to hold two contradictory corrections simultaneously. Variant perception that is technically more accurate than consensus but does not translate into a cleaner directional position is research that produces more underwriting work, not more conviction.

The first thing to watch is the 31 July 2026 Q2 FY2026 ROIC print — a number above 12% with bolt-on M&A picking up from Q1's $20.6m re-validates disagreements #2 and #3 simultaneously and is the single observable that moves the most weight in our framework; a number stuck below 11% with deployment still dormant is the cleanest single piece of evidence the bear case will get inside this calendar year.