The Mexico ledger: in Cadillac's first year, the car — not the circuit — writes Pérez's result

Four out of five.

That is the first line of Sergio Pérez's Mexico City Grand Prix record — four points finishes in five appearances before Red Bull, the only miss a 2026 mechanical retirement. The ledger after that reads inverted: P3 in 2026, P3 in 2026, a wall on lap one in 2026, a Q1 exit and P17 in 2026.

Four seasons, four radically different outcomes. No circuit produces that bimodal a profile. Car competitiveness does, and so does the pressure a driver puts on himself. When Pérez returns to the Hermano Rodríguez pit lane on 29 October 2026 in a Cadillac (TWG Motorsport), the same ratio will decide the result — except that this time both variables are unknown at once. Twelve months without a race, and a car for which nobody holds a single lap of on-track data.

So the question this weekend is not whether Pérez can win. It is whether Cadillac can complete a clean, full-distance race at 2,200 metres — and whether the data from it will still be useful in 2027.

I build the dashboard twice before I trust the first number. Building it the second time here made one thing obvious: this entire preview contains no measured performance data. No lap times, no sector times, no GPS speed, no tyre data. What exists is a commercial and structural layer wearing a race-preview costume.

Mexico City sits roughly 2,200 metres above sea level. That altitude does three things simultaneously: it reduces aerodynamic drag and downforce, raises turbocharger rotational speed for a given boost target, and cuts the cooling mass-flow available to radiators, brake ducts and the power unit's electrical side. An established team carries a historical cooling model for this venue. Cadillac does not. The team becomes familiar with the place for the first time on Thursday media day. For a debutant, that is a higher-variance reliability environment than a sea-level circuit — a fact the source material never mentions.

The structure is straightforward. Cadillac is eleventh of eleven in its first season by definition rather than opinion. Cost Cap applies from year one, and the new-entrant aerodynamic testing allowance still needs reconciling against the published 2026 ATR schedule. The power unit supply arrangement is the largest missing chapter in the whole analysis: a debutant manufacturer running its own hybrid unit from day one would be extraordinary, and a customer supply in the opening seasons is the conventional paddock path. Until that is settled, no competitiveness judgement is complete.

Do not read garage location as performance. Cadillac's hospitality sits first at the paddock entrance, with the garage adjacent to Alpine — the 2026 last-placed constructor. That is an institutional seniority and fan-access marker, not a lap-time variable. Treating it as a competitive omen is like estimating qualifying pace from a queue outside the pit building.

Pérez's "Mexico curse" is a narrative artefact, not a mechanism. The record decomposes cleanly: four points finishes in five pre-Red Bull visits, the sole miss a 2026 mechanical failure; then third places in 2026 and 2026 in a title-capable car; then a lap-one retirement in 2026 while over-attacking; then a Q1 exit and P17 in 2026 in a car with no pace. The variance tracks car competitiveness and self-imposed pressure, not the circuit.

Second, the new-entrant base rate. The most informative sentence in the source is a negative one — first points are "more likely by the middle of the 2027 season." That is authorial opinion, not measured data, and against base rates it sits plausible to mildly optimistic. Eighteen months of learning for a from-scratch operation facing ten mature ones is reasonable, and it is consistent with the modern era in which Cost Cap and ATR concessions shorten the maturation curve relative to the 2026 precedent. The model says that much. It cannot say how Cadillac's wind tunnel will correlate with the track in year one, and for a new entrant that is the single largest technical unknown.

The dominant strategic variable for Cadillac in Mexico is not tyre offset — it is completing the race. For a debutant at altitude, a clean full-distance run is worth more in correlation data than any opportunistic tyre gamble. A team optimising for 2027 does not chase points in 2026; it banks data.

Third, the pit lane. A new team's crew, its double-stack procedures and its race-engineer communication are all first-season variables with no quantitative basis in this report. Home-race pressure, a stadium-section circuit and a novice crew is a textbook error-stacking environment. A botched stop at a home Grand Prix does disproportionate narrative damage relative to its sporting cost, and no points offset it.

Fourth, expected value. For a backmarker, points arrive only through Safety Car, VSC, red flag or midfield attrition. The asymmetric payoff makes an early opportunistic stop more rational for Cadillac than for any established team: limited downside, open upside. Whether the team intends to play that way is not in the source, so it is inference, not reporting.

And the fifth part is the one that grates: the source never names Pérez's Cadillac teammate. Without a same-car reference frame, no serious driver judgement is possible. Is he still fast? Can he lead development? Both answers come from the other garage, and the other garage is absent. That is a hole in the source rather than in the analysis, but the effect is the same: the only sporting question that matters goes unanswered.

Now the disagreement. "Expected return, guaranteed disappointment?" is a straw man. Disappointment is not guaranteed by team status; it is manufactured by bolting a homecoming narrative onto a car the author himself says will not score until mid-2027. The expectation gap is an editorial artefact.

This is where I have to guard against my own habit. Model worship is a live risk: a debutant's simulation will look tidy precisely because there is no historical comparison to contradict it. I publish the error bars alongside the finding and state what the model cannot see — weather, politics, a promoter's cash position, and a first altitude visit by a team whose cooling model does not yet exist.

The same caution applies to garage position. Opening at the paddock entrance signals institutional preference; converting that into a performance signal is the equivalent of treating a transfer rumour as a signed contract. Two separate ledgers, kept separate.

The most under-weighted variable in this preview is the twelve-month layoff. A full season without racing removes competitive sharpness, and more importantly removes a whole generation of car-behaviour feedback. Pérez's last competitive reference points are the 2026 ground-effect cars, not the 2026 regulation cars. For a new team that matters, because what it needs is unbiased feedback.

That produces a tension nobody is writing about. A veteran arriving at a new team pushes toward familiar directions; a new team needs disciplined, neutral correlation work. Whether Pérez's feedback accelerates or distorts Cadillac's development is the genuinely open question of the season, and it will be settled by process, not lap time.

Age and career curve compress the upside. By 2026 he is in his mid-thirties with 250-plus starts. The realistic value proposition is calibration and commercial pull, not a performance step-change. That is a legitimate reason to hire him — and it also means the sporting framing around the signing flatters it.

The commercial layer is the strongest and least written. A home driver at a sold-out venue is a sponsorship asset no rookie replicates; Pérez is the anchor of Cadillac's Latin American programme. The article reports the excitement without connecting it to the commercial mechanics. That is the old Force India error: performance kept in one book, money in another, then both collapsed into a single ledger.

The empty paddock taught the tyre model what noise had hidden. Mexico this year has plenty of noise — homecoming, new team, sold tickets. The model has to be kept clear of it.

A race weekend is a ledger, and every lap writes a correction. My three-column book opens like this.

Pace: is the Cadillac visibly slower through the speed traps than its immediate neighbours? If so, that is deliberate cooling and brake-duct conservatism — a debutant signature, not a concept failure. Call it drag level, not decline.

Strategy: how long is the first stint? A long, slow opening stint points to data capture; an early stop around a Safety Car window points to a points gamble. That is the clearest public window into whether the team is optimising for 2026 or 2027.

Execution: the pit stop. One clean stop under home pressure is worth no less than the data from the full race.

And one flat sentence in favour of consensus: yes, Cadillac is not points-competitive this season. That is clear and needs no three-paragraph proof.

I do not chase the cliff; I chart the grip that disappears before it. The cliff here is not 2026's. It is 2027's. If Cadillac leaves Mexico with data that correlates to its wind tunnel, the season is a success whatever the result sheet says. If it does not, next year's points projection will be made in the same darkness as this weekend's.

For Pérez the question is simpler, and therefore harder: can he take that car where it is not yet, without assuming he knows where it is?