This is the first weekly Cantillon Deep Dive: one stock, run through our Six Gates research framework in public, from macro regime down to an executable plan. ISRG earned the debut by producing the most violent three-act sequence of the year. Wednesday night: a clean Q2 beat, sold off hard in after-hours trading to a low of 350. It tried to come back twice, first to 370, then a weaker second attempt in premarket that only reached 367, a lower high. Friday: the recovery failed, the guidance got repriced, and the stock closed at 345.42, down 14.15%, below every acceptance zone on a two-year lookback. The invalidation this framework set Friday pre-market fired within eight hours. No entry ever triggered. That is not a failed call. That is the entire point of having gates.
Most research frameworks tell you what to own. None of them tell you when institutions are actually positioned in it. The Six Gates run both questions in sequence: Gate 0 asks whether the macro regime permits the trade at all. Gate 1 tests business quality. Gate 2 prices three worlds: bear, base, bull. Gate 3 maps catalysts and tags each with the regime it needs to work. Gate 4 checks institutional flow through the IVT terminal. Gate 5 forces the thesis to survive a hostile committee, then defines entry conditions, size, and the invalidation that voids everything, before any capital moves.
Each gate has a kill condition. A stock that fails a gate stops there. The full methodology, including the exact research prompts and sizing rules, ships as The Six Gates PDF, included with membership. This week you get to see it work under the worst conditions a framework can face: a thesis that died within hours of being written.
This is the biggest change of the week, and it is bigger than ISRG. On Friday morning the scan read Inflationary Pump with a +1 bonus: equities on, risk on, bonds off. By Friday's close, all three layers were red. Full Risk-Off, bonus -2, composite collapsed from +3 to +1, transition deteriorating. Longs are now blocked at the first gate: a negative bonus in your direction is the framework's hardest stop.
Note the mismatch worth watching: the regime engine reads Full Risk-Off while the VIX still prints 17, low fear. Flow has turned before fear has. That combination usually resolves with fear catching up, not with flow calming down.
The headline was a beat. Q2 2026, reported July 16: EPS $2.80 against $2.41 expected. Revenue $2.89B, up 18.4% year over year. Procedures grew 16%, Ion procedures 36%, 468 da Vinci systems placed against 395 a year ago, installed base near 13,000. The razor-and-blade model is intact: every system placed becomes a recurring instrument stream.
Friday repriced the fine print, not the quarter. Management flagged US procedure growth moderating to roughly 12%, softness in benign categories as patients defer elective procedures, insurance coverage losses tied to expiring ACA credits, and margin compression from rising costs and tariffs. None of that breaks the business. All of it questions the growth rate the old multiple was built on. The demand-durability red flag moves from minor to elevated, and Gate 1 gets a mandatory re-run at the October filing.
This is the first Cantillon Deep Dive, so everything is open: the broken valuation map, the regime-tagged catalyst calendar, the full IVT flow readings, the snapback sizing math, and the one level that flips the story back. From No. 002 onward, Gates 2 through 5 are for Elite members, and future issues will carry live entries.
Friday pre-market, this gate had three worlds on file: bear 348 to 350, base 380 to 410, bull toward the 460s shelf. The after-hours session had tested the bear zone and bounced twice, the second attempt only reaching 367, a lower high, which read as sellers defending the zone rather than buyers reclaiming it. Friday's close at 345.42 overran the entire bear zone and every acceptance area on the two-year chart. The market did not defend the bear world. It moved into it and kept going.
The framework's rule when price exits the map: you do not project old zones downward, you wait for the auction to build new ones. There is no volume history at these prices to lean on. What we can say: the guidance cut gives the bear world real fundamentals now, roughly 12% US procedure growth against a multiple that assumed 15% or better, so the October print, not a chart level, is what re-anchors valuation. The old zones stay on file as overhead references.
| Reference | Level | Role After the Break |
|---|---|---|
| Friday close / new print | 345.42 | Price discovery zone. No acceptance yet. Not a support level, a location. |
| Old bear zone | 348 – 350 | First overhead supply. Reclaiming it is the failed-breakdown trigger (see Gate 5). |
| Old base range | 380 – 410 | Second overhead shelf. Trapped Friday-morning buyers live here. Heavy resistance on any recovery. |
| Catalyst | Timing | Dir | Regime Required | Status vs Full Risk-Off |
|---|---|---|---|---|
| Q3 2026 earnings | Mid-October (dated) | Two-sided now | Works in any regime | Live, but carries the burden of proof on the 12% guide |
| da Vinci 5 placement ramp | Quarterly proof | Bull | Works in any regime | Live, slow-burn |
| Ion procedure growth | Ongoing | Bull | Works in any regime | Live, slow-burn |
| ACA credit / coverage resolution | Undated, political | Bull | Macro-dependent | Dead weight until dated |
| Multiple re-rating | Undated | Bull | Needs loose liquidity + risk appetite | Blocked twice over in Full Risk-Off |
The gate survives on a technicality: October is dated and inside six months. But notice what changed. On Friday morning the October print was a bullish catalyst. After the guidance cut, it is two-sided: it either proves the 12% moderation was the floor, or it confirms deceleration. Nothing dated defends the price between now and then. That is three months of price discovery with no scheduled help.
| IVT Reading | 4H | Daily | Read |
|---|---|---|---|
| Bias / Swing | BEARISH · 263-bar bear | BEARISH · 371-bar bear | Structure never turned. Friday confirmed it with force. |
| VWAP Dev | -19.41% · FAIL | -28.55% · FAIL | The deepest extension of the entire decline. Nothing tradeable lives out here, in either direction. |
| Conf Score | 3 (3F) | 1 (2F) | Confluence collapsed with the price. Thursday's building bid is gone. |
| SD2L Extreme | ACTIVE +2F | ACTIVE +2F · set 3 | Third extreme in five sessions. Each snapback from these has been weaker than the last. |
| WR (500) | 75% · 27b avg (8/8) | 67% · 43b avg (3/4) | The stats stay strong, but they were built in a regime that no longer exists. |
| Breadth | +3/8 · Trend -2 | +5/8 | Market breadth still constructive. ISRG is being singled out, not dragged down. |
Wednesday's SD2L fired into the after-hours flush and paid its historical 75%, a 20-point snapback off the 350 low. A second attempt only reached 367, a lower high, weaker than the first. Friday's SD2L is a different animal: it fired into a guidance-driven repricing, in a Full Risk-Off regime, at -28.55% deviation, after the prior extreme's entire gain was confiscated in one session. Same signal, worse world. The framework reads that honestly: shorting -28.55% deviation is chasing, and buying a knife with a negative macro bonus is a quarter-unit proposition at absolute maximum. For most readers the correct position is none.
Verdict: WATCH. ISRG failed Gate 0 for longs when the regime flipped, and the valuation map is void until a base prints. Under the framework's verdict language, a Gate 0 failure is re-tested on the next regime change, not argued with. The business still passes Gate 1. This is a good company entering price discovery in a hostile regime, which is precisely the setup where patience is a position.
| Plan Element | Level / Rule |
|---|---|
| Standing aside is the base case | Gate 0 blocks longs at bonus -2. No trend short either: -28.55% dev means the short entry is long gone. |
| Snapback, for those who take it | SD2L set 3, both timeframes. Maximum 0.25R (0.5R SD2L standard, halved for the negative bonus). T1 only, no runner. Requires a confirmed reclaim bar, not a new low. A fresh low below the entry bar kills it immediately. |
| The failed-breakdown trigger | A daily close back above 350 puts ISRG on the front of the watchlist: breakdowns that fail within a week trap the most aggressive sellers of the move. |
| The re-entry conditions (full trade) | All three, in order: regime bonus back to 0 or better · a printed base with acceptance, not a wick · Dev PASS with an S2+ tier. Until then ISRG is research, not a position. |
| Gate 1 re-run | Mandatory at the Q3 print, mid-October. The 12% guide is now the number that decides the bear world's depth. |
| # | Gate | Result | Key Number |
|---|---|---|---|
| 0 | Regime Permission | FAIL (long) | Full Risk-Off · Bonus -2 · all layers red · VIX still 17 |
| 1 | Business Quality | PASS* | EPS $2.80 beat · US procedures guided to ~12% · re-run in October |
| 2 | Valuation | VOID | Closed 345.42, below the full two-year map · rebuild after a base |
| 3 | Catalysts | MARGINAL | October print now two-sided · nothing dated before it |
| 4 | Flow Confirmation | NO ENTRY | Dev -19.41% / -28.55% FAIL · SD2L set 3 · 0.25R max snapback |
| 5 | Verdict + Size | WATCH | Reclaim 350 = failed breakdown · full trade needs regime + base + tier |
Here is what this week actually demonstrated. The framework wrote its own obituary conditions Friday pre-market: an invalidation level, defined before any entry, on the record. The market triggered it within eight hours, the entry never fired, and zero capital was lost learning the thesis was wrong. Everyone who bought Friday morning's bounce learned the same lesson and paid 14% in tuition. The gates are not there to be right. They are there to make being wrong cost nothing.
Dev % distance from the AVWAP anchor; entries require price inside the band (PASS) · Conf institutional confidence score, (xF) = confirming timeframes · SD2L two-standard-deviation oversold snapback signal · S1–S4 signal tiers, higher = more independent confirming factors · WR(500) historical win rate over the last 500 bars · R one standard risk unit · Composite Cantillon regime layer sum, max 9 · Bonus regime tailwind/headwind applied to conviction.