Readership data temporarily unavailable. Check thesportspage.goatcounter.com for live stats.
The New York Mets’ 2026 paid attendance averages approximately 36,110 per game. That ranks them seventh in Major League Baseball — which, if you stop reading there, suggests a franchise in reasonable health relative to its investment.
Stop somewhere else instead. Stop at the secondary market, where Mets home-game tickets appeared in June and July 2026 at $4 — all fees included. Not $4 face price plus fifteen dollars in handling. Four dollars, everything included. That is less than the cost of the subway ride from most Queens addresses to Citi Field.
When attending a game costs more than the ticket itself, the attendance figure has stopped measuring anything about the team. What it measures now is the prior decisions of people who bought season packages in March, when a $352.2 million payroll and Juan Soto’s projected production made hope a rational asset class.
The arithmetic is worth doing slowly. The Mets hold the highest active-roster payroll in Major League Baseball at $352.2 million, with a Competitive Balance Tax payroll of $368.7 million that will trigger approximately $120 million in luxury tax payments — more than the entire player payroll of six MLB franchises. Juan Soto alone carries $61.875 million in 2026 compensation. The franchise built this number on the premise that a roster this expensive would make the October investment feel necessary.
Through July 4, the Mets are 36–52. They are last in the National League East, sixteen games behind Atlanta. Carlos Mendoza was fired at 34–47 in mid-June after a doubleheader sweep by the Cubs that featured six errors. The Marlins and the Nationals — teams the Mets were expected to pass before Memorial Day — are both ahead of them in the standings.
The ticket office compounded the problem. Rising Apple identified the franchise’s “biggest miscalculation of 2026” as the decision to withhold single-game tickets from public sale after June, betting on competitive pricing leverage in the second half. The theory was defensible in February: high payroll, high expectations, premium demand anticipated. By July, the team had collapsed and the held-back inventory had no buyers at any price that covered the cost of going. The tickets entered the secondary market at four dollars because sellers had already absorbed the ticket cost as a sunk expenditure, and any revenue was better than absorbing the carrying cost of a useless asset.
This is revealed preference in its most unambiguous form. When a fan sells a ticket for four dollars rather than attend, they are publishing a valuation the official box score will never capture: that watching this team, in person, on a Tuesday in July, produces negative net utility relative to the alternatives. The opportunity cost of four hours at Citi Field watching the Mets lose to the Nationals is greater than four dollars to a fan who already paid for the ticket in March.
The contrast with 2025 is structural, not incidental. Twelve months ago, the Mets set an all-time single-season attendance record at Citi Field: 3,182,057 fans over the course of the year. They had 19 sellouts. Nearly half their home games exceeded 40,000. They ranked fifth in MLB attendance after finishing seventeenth the year before. Total Citi Field revenue was $311.4 million. The franchise was selling hope at a premium, and the premium was justified by competitive play.
One year later: empty midweek sections, $4 all-in resale floors, and a ticket office sitting on inventory it cannot move. The swing from fifth to the secondary-market floor in twelve months is not a narrative about disappointment — it is a data series about the relationship between on-field performance and stated fan demand. The 2025 attendance record was not measuring Mets loyalty. It was measuring winning-adjacent enthusiasm, which is an entirely different thing, and which evaporates at a measurable rate when the wins stop.
The structural point: MLB announced attendance measures tickets distributed, not fans present. A ticket is counted as paid attendance the moment it is sold, regardless of whether anyone uses it. In competitive seasons, this gap is small. In seasons like 2026, it is a canyon. The 36,110 per-game figure is not wrong — it is just counting the wrong thing. The secondary market price is the corrected figure: what access to this game is worth to someone who already owns it and has updated their prior based on 88 games of new evidence.
The Mets will spend approximately $368.7 million in CBT payroll this year, plus $120 million to the Commissioner’s office for the privilege. That is $488.7 million allocated to a team whose fans are currently pricing away attendance at four dollars a seat. At some point, the gap between what the franchise declared it would spend and what the market says the product is worth becomes its own statistic.
It does not appear in any official box score. It appears on StubHub. It reads: $4.00. All fees included.
Six issues in six days. The week’s argument, repeated across four sports: patterns that held for twenty years were not laws — they were local customs, and the exceptions are more instructive than the rules ever were.
The week’s six issues were retrospective — historical arguments, no new forward-looking quantitative claims. All active predictions are carry-forwards, updated against data current as of July 5.
| Issue | Prediction | Outcome (July 5, 2026) | Grade |
|---|---|---|---|
| #82 | Mets sell signal: at .422 win rate they have crossed below the threshold that historically justifies holding contender-built rosters at the deadline | Mets now 36–52 (.409) through July 4. Mendoza fired at 34–47. Signal confirmed and deepening. Trade deadline August 3. Every additional loss is the model being right in slow motion. | HIT |
| #84 | Payroll efficiency collapse: at $352M, marginal win cost would exceed $9M per win if the team finished below .500 | 36–52 at $352.2M = $9.78M per win, before luxury tax. Add ~$120M in CBT payments: effective spend per win approaches $13M. The model’s math is running on a real team in real time. | HIT |
| #90 | Swanson “visitor not resident”: true talent ~.183; early .234 boosted by favorable Citi Field matchups; regression predicted as schedule normalizes | Swanson regressed to .175 by mid-June — overshooting the predicted floor. Exploded July 1 (3 HR, 8 RBI; first Cubs player with 26 RBI in 10 games since 1939). Current: .203 / xBA .206. Direction was right; the underlying talent estimate held. July 1 is an expected variance spike, not a new player. | PARTIAL |
| #95 | Mets OPS clustering (descriptive snapshot, June 11 data — no forward projection made) | Mendoza fired; lineup reshuffled. Structural diagnosis — Soto isolated, rest of roster clustered at .671 — remains accurate. The piece was right about the problem. The solution has not arrived. | N/A |
| #97 | Clark–Robertson distance match (partial-season caveat explicit in article) | Clark 2026: 21.2 / 8.2 / 4.0, Fever 9–7 (3rd East). Season ongoing; the full-season comparison requires full-season data. Direction is holding, prior is intact. | PENDING |
| #98 | Host nation semifinal rate: 59% prior across 22 World Cups, labeled “a prior, not a promise” | USA in Round of 16; plays Belgium July 6. Final is July 19 at MetLife. Prior still live; no grade possible until the knockout rounds resolve. | PENDING |
| #83 | Sorsby enforcement: Big 12 litigation and NFL supplemental draft as dual institutional checks on eligibility violations | Big 12 lawsuit active. NFL cancelled the supplemental draft entirely — one of the two cited enforcement mechanisms is gone. A two-variable model with one variable removed is a different model. Revision issue queued for Monday. | PARTIAL |
The Mets sell signal (#82) is the clearest entry on the board. The piece ran when the Mets were at 35–48. They are 36–52 through July 4 — one win in the past ten days while Atlanta beat them 14–3 on the Fourth. The deadline is August 3. The sell signal has been live for three weeks. Every additional loss makes the rational decision at the deadline harder to argue against.
The payroll efficiency call (#84) is running exactly as modeled: $9.78 million per win before the $120 million luxury tax levy. Adding the CBT payment pushes the effective per-win cost toward $13 million. No amount of Juan Soto can close that gap from a roster that scores like the 2026 Mets.
The Swanson regression (#90) was directionally correct. He fell to .175 before the July 1 explosion — a true variance spike, not a talent change (xBA .206 confirms the underlying estimate was accurate). He joined historic company on July 1. He is still a .200 hitter in a good month.
The Sorsby peer enforcement prediction (#83) is a partial miss that requires a correction. The NFL’s elimination of the supplemental draft removes one of the two enforcement mechanisms the original model cited. A two-variable framework with one variable gone is not the same argument. The Big 12 lawsuit is still active, but it carries less weight when it acts alone. A revision issue posts Monday.
The Mets clustering piece (#95) was accurate as of June 11. It cannot be credited for anything after that: Mendoza was fired, the lineup has been reconfigured, and the specific player configuration is in flux. The structural diagnosis — a severe talent gap between Soto and the rest of the roster — remains correct. The piece described the problem. It is not responsible for whether the problem gets solved.
No other forecasting errors to report. An all-retrospective week produces no new misses. The running ledger: two hits, two partials, two pending, one N/A. The model is performing at approximately the expected error rate.
Calling the July 4 loss the Mets’ “rock bottom.” Thirty-six and fifty-two is bad. But the team has 54 games remaining, the $352 million roster will play those games regardless of what the standings say, and the trade deadline on August 3 is the actual decision point. Rock bottom implies there is nowhere left to go. There is. The question is whether the front office makes the rational call or the emotionally expensive one.
The Swanson “renaissance” narrative after July 1. Three home runs in a single game is real. So is a .206 expected batting average. xBA does not care about individual memorable performances — it measures the underlying contact and launch angle profile that generates outcomes over a full season. Updating the prior too aggressively after one data point is exactly what this newsletter exists to warn against.
The $4 Citi Field ticket as financial signal. No outlet treated this as the market data point it is. A $4 all-in resale price is not fan-griping. It is the secondary market’s real-time valuation of the Mets’ entertainment product, adjusted for sunk cost and updated by 88 games of evidence. The official 36,110 per-game figure is counting a different thing. The four-dollar ticket is the corrected number, and it tells a more accurate story about this franchise’s current relationship with its fanbase than any announced attendance ever will.
The 2025-to-2026 attendance collapse. Twelve months ago: Citi Field attendance record (3.18 million), 19 sellouts, $311.4 million in stadium revenue, fifth in MLB. Today: $4 tickets and empty midweek sections. That swing in twelve months is the kind of data series that belongs in a revenue-model case study, not buried in secondary-market anecdotes. The fanbase did not disappear. It updated its prior.
Monday opens with the Sorsby correction. The NFL’s decision to eliminate the supplemental draft removes one of the two enforcement mechanisms the peer enforcement model relied on. That is not a footnote — it requires its own issue to map the new landscape with one pillar missing. The Big 12 lawsuit carries a different weight when it acts alone.
After that, the queue runs through the Analog Year series (2021 Mets comp, Astros analog, cross-sport persistence) and the Half-Life series continuation. By the end of July, the trade deadline piece will write itself. The sell signal from Issue #82 has been live for three weeks and the window is narrowing.
The World Cup final is July 19 at MetLife Stadium. The 59% host-nation semifinal prior is still live and will be tracked game by game through the knockout rounds.
And then there is Notre Dame. The dynasty question — whether the Irish have built something structural or accumulated something that will regress — has been building in the queue since the EO series mapped the regulatory environment that protects their position (see Issue #16). The full argument publishes Monday in a dedicated issue. But we will say it plainly here, because this week’s centerpiece is about the gap between declared numbers and honest ones, and honesty requires putting calls on the record rather than hedging until the answer is already obvious:
The Sports Page is on record: Notre Dame wins the 2026 College Football Playoff national championship.
The first meaningful audit is November 7 against Miami — the first elite crossover opponent on the schedule that will tell us whether the Irish’s roster construction holds against an SEC-quality front. If Notre Dame holds that game, the prior strengthens significantly. If they do not, this prediction will appear in the accountability section of Sunday Edition No. 24, and it will deserve to be there.
That is how predictions work here. You say the thing. You track the thing. You own the result, either way.
This issue is running on a day that has nothing to do with sports and everything to do with why The Sports Page exists.
Ninety-nine issues ago, someone looked at a box score and asked a question the box score did not answer. Not “who won” — that was always available. Not even “how” — the game log covered that. The question was more specific: what does this number actually mean? What is it hiding? What does the distribution look like if you go back twenty years? Where does this fit in the larger argument that baseball and football and basketball are making about how variance and talent and money and luck interact in ways that almost nobody describes honestly?
That question is the entire newsletter. The ninety-nine issues are ninety-nine attempts to answer it, or to answer the question that follows the first answer, or to document the cases where it turns out the question was the wrong question entirely. The work is cumulative. It compounds. The body of evidence that exists today makes the next issue more rigorous than the first one, because there is now something to argue against.
To the one who noticed the numbers first — this issue is yours. The streak holds. The stack of questions is taller than it was at Issue #1, which is the correct direction for a project built on intellectual honesty about uncertainty. The next 401 will be argued against a foundation that did not exist when the first one ran. That makes them considerably harder to get wrong without being held accountable — which is exactly as it should be.
Happy birthday. The numbers were always worth noticing.