The Doctors Said May. The Calendar Says May. The Players Say Otherwise.
The Week in Review: Issues #58–63
- #58May 25 — Marcus Freeman Won the Portal. Then He Started On 2027. — Notre Dame signed the #2 high school class in February, added four top-100 transfers in the spring portal window, and has the 2027 class’s early composite already outscoring the class that finished second in America. One-third of the 2026 roster will be first-year players. The newsletter called it the biggest roster-management bet in the program’s modern era, and the piece noted that the conversion rate from “great class” to “national title” is roughly 25–30% even for the best programs.
- #59May 26 — Bill James Says the Mets Should Be 22–24. The Bigger Story Is What That Means Over 162. — Through 46 games with 176 runs scored and 186 allowed, Pythagorean expected 22–24. They were 20–26: two wins of close-game misfortune. Projected full-season pace: 76 wins. The prescription for improvement: Lindor and Alvarez back before end of May.
- #60May 27 — Past the Pythagorean. BaseRuns Asks the Harder Question. — David Smyth’s formula goes one level deeper than Pythagorean, asking whether the runs scored were themselves the product of bad sequencing luck. For the Mets: not really. Their offensive components — singles, doubles, walks, home runs — are producing roughly what they should. The cavalry remained Lindor and Alvarez. Due before end of May.
- #61May 28 — Forty, Fifty-Seven, Thirty-Two. The Other City Has Forty-Nine. — The combined championship drought for New York’s three major franchises (Mets, Jets, Rangers): 129 years. Washington’s three (Nationals, Capitals, Commanders): 49 years. D.C. won two titles in eighteen months between 2018 and 2019. The math says New York fans carry more misery by every available metric. Whether misery felt is equivalent to misery measured is, as the piece conceded, a harder question.
- #62May 29 — The Tiebreaker Pretends to Add Data. Instead, It Adds Noise. — Part 2 of the Significance Series. A 7–6 head-to-head record in 13 games is the statistically most likely outcome between two equally matched teams. A two-sided binomial test cannot reject the null. An 11–2 record is required to reach conventional 95% significance. MLB activates the tiebreaker rule on a 7–6 record anyway.
- #63May 30 — Zero for Thirty-Eight. Then a Shower. Then the IL. — Seattle catcher Cal Raleigh went 0-for-38 between April 30 and May 12. His teammate Logan Gilbert convinced him to shower in full uniform to break the juju. On May 13 he went 2-for-4 with a walk. Two days later he was on the IL with an oblique strain. The piece argued regression to mean — not the shower — broke the slump, and that the oblique was probably the reason the slump started in the first place. Bayesian posterior on his true-talent batting average: approximately .215, near his career mark of .222.
Six issues: one college football recruiting analysis, two consecutive Mets analytical installments, one championship-drought retrospective, one significance-testing entry, and a catcher who showered in his uniform and then went on the IL anyway. The through-line is injury — from the oblique that probably caused Raleigh’s slump to the calf and knee that kept Lindor and Alvarez in the training room all week.
The Prediction Scorecard · Week 8 Resolution
Six open items entered this week: one carry-forward from Issue #45 (Devin Williams), one carry-forward from Issue #56 (Carolina’s Cup run), two timing predictions from Issues #59 and #60 (Lindor and Alvarez back before end of May), and two long-horizon predictions from Issues #58 and #63 (Notre Dame’s 2027 class ranking, Raleigh’s Bayesian posterior). Three resolve today. Three remain open.
| From | Prediction | Outcome as of Sunday Morning | Grade |
|---|---|---|---|
| #45 May 12 (carry) |
Devin Williams’s true talent settles in the 2.80–3.50 ERA range (xERA-driven Bayesian projection). The April ERA of 9.95–10.29 was a small-sample artifact that would dilute as the May line accumulated. | Season ERA now 4.32, down from 10.29 in April. May ERA: 0.00 across 10+ straight scoreless appearances. Five saves on the month. The April denominator still weights the cumulative number, but xERA remains 3.07 and the May line has confirmed the underlying projection in full. ERA continues migrating toward the prior, exactly as described. | HIT — upgrading from last week’s PARTIALLY HIT. The model called this correctly from the first day of May. The ERA is still catching up to the pitcher. The pitcher is exactly who Issue #45 said he was. |
| #59 May 26 | Francisco Lindor and Francisco Alvarez both expected to return to the Mets lineup before the end of May. Their return would add roughly half a run per game to the offense and lift the Pythagorean projection from 76 wins toward the low-to-mid 80s. | Today is May 31. Lindor remains on the IL with a left calf strain; the Mets have not given a return date, though June is now the working assumption. Alvarez remains on the IL with a right knee issue; he may begin a minor-league rehab assignment “as early as next week.” The Mets are 25–33, scoring 223 runs and allowing 250 through 58 games — a Pythagorean pace of approximately 72 wins, worse than the 76 cited in the piece. | MISS — on the timing. “Before the end of May” did not hold for either player. This is a medical-calendar miss, not a talent-evaluation miss. The Bayesian projections for what both players will do when healthy are unchanged. The timing was wrong, and the scorecard reflects that. |
| #60 May 27 | Same underlying timing prediction as #59: “When Lindor returns — expected before the end of May” the BaseRuns A-term rises. “When Alvarez returns” the D-term rises. Both players’ components, when restored, would add roughly half a run per game. | Inherited miss from Issue #59. Neither player has returned. The BaseRuns framework is analytically correct as written; the timing was borrowed from an optimistic medical projection that did not hold. | MISS — same root cause as Issue #59’s timing prediction. The analytical framework stands intact. The calendar call does not. |
| #56 May 23 (carry) |
Implicit forward question from the Carolina piece: does a 9–1 playoff start (and underlying 8–0 entry to the ECF) translate to a Cup, given the historical rate of 3-of-4 prior teams doing so? | Carolina swept the Montreal Canadiens 4–1 in the Eastern Conference Final and is now in the Stanley Cup Final vs. Vegas Golden Knights. Game 1 is Tuesday, June 2, in Raleigh. Vegas swept Colorado in the Western Conference Final. The Hurricanes have home-ice advantage. Carolina’s full playoff record: 12–2. | PENDING — upgraded. Carolina needed to reach the Final for the historical comp to apply. They’re there. The 3-of-4 Cup conversion rate for teams that started 8–0 or better now has a live subject. Sunday Edition No. 009 will grade. |
| #63 May 30 | Cal Raleigh’s Bayesian posterior, discounting the injury-contaminated 2026 sample, settles near .215 — essentially his career true-talent level. When healthy, his production returns to that baseline. | Raleigh took 30 swings at moderate intensity off a batting tee in Arizona on Friday. Expected to hit again this weekend and fly to Seattle for a check-in with the club. No official return timeline. His 2026 slash line stands at .161/.243/.317 through the oblique injury. | PENDING — no return data yet. The piece’s structural argument (regression to mean broke the slump; oblique probably caused it) was validated by the sequence of events before publication. The talent prediction awaits his return. |
| #58 May 25 | Notre Dame’s 2027 recruiting class will be “one of the three best in America, and quite plausibly the best” when it fills out to its full complement of signees in December. | The 2027 class is currently ranked 7th nationally with 18 commits. Early-cycle rankings are compressed by volume: programs above Notre Dame have more commitments, not necessarily stronger ones. The blue-chip ratio remains strong. Signing Day is December 2026. | PENDING — far too early to grade. The class will expand significantly through fall and December. Ranking will shift as the back end of the class fills in. |
What We Got Right This Week
Devin Williams: The Bayesian Model Gets Its Full HIT
Seven weeks ago, Issue #45 made a specific claim: the Devin Williams who posted a 9.95 ERA in April was not the real Devin Williams. The career xERA of 3.07 represented his true talent level, and the April blowups were a small-sample artifact that the Bayesian prior — built on three-plus years of dominant relief work — should outweigh. The newsletter upgraded this to PARTIALLY HIT last Sunday. Today it is a full HIT.
Across all of May, Williams has not allowed an earned run. Ten-plus straight scoreless outings. Five saves. His season ERA has migrated from 10.29 to 4.32 and will continue migrating as the April denominator shrinks into irrelevance. The underlying pitcher is, exactly as the model said, the closer who finished eighth in Cy Young voting in 2023 and posted a sub-2.50 ERA in each of his last three full seasons. The model was right. The April ERA was noise. The newsletter called it when the ERA was still 9.95.
Issue #63’s Structural Argument Was Validated by Events
The Raleigh piece made two predictions: (1) regression to mean, not the shower, broke the slump; and (2) the oblique strain was probably the reason the slump started. Both held up before the piece was even published. The slump had already ended — by the laws of probability, on May 13. The oblique had already been confirmed by the IL placement on May 15. The Bayesian framing in Issue #63 had pointed toward an injury explanation for why Raleigh’s 2026 numbers looked nothing like his career numbers. The sequence of events validated that inference.
Issue #53’s Plate-Discipline Result Held in the Week’s Standings
The piece from three weeks ago (published as Issue #53) established that the team-level correlation between pitches per plate appearance and winning percentage is r ≈ 0.07 — indistinguishable from zero. The current standings continue to provide no reason to revise that conclusion. Patient teams and free-swinging teams are sorting by other factors. The coefficient of determination is not cooperating with the narrative, as usual.
What We Got Wrong This Week
The “Before End of May” Timing Calls Were Wrong
Issues #59 and #60, published Tuesday and Wednesday, both stated that Francisco Lindor and Francisco Alvarez were expected back before the end of May. The newsletter did not invent this timeline; it reflected the prevailing medical projection at the time of publication, referenced by the Mets’ own manager and multiple credentialed reporters. But “expected” is not “guaranteed,” and the scorecard does not grade on intentions. It grades on outcomes.
Lindor’s calf strain has extended beyond its initial projection. Alvarez’s knee issue, originally described as a six-to-eight week return, is progressing but has not reached game action. Both remain on the IL today. The Mets’ Pythagorean pace — approximately 72 wins based on 223 runs scored and 250 allowed through 58 games — has deteriorated slightly since Issues #59 and #60 were published. The offensive improvement those pieces projected has not materialized, because the offensive improvement required the players to be available.
The analytical frameworks in those pieces are intact. Pythagorean expectation is not wrong. The BaseRuns model is not wrong. The talent projections for Lindor and Alvarez remain valid. The timing prediction was wrong. Injuries are not excuses. In this case, they are the data. The scorecard marks them as misses and the newsletter moves on.
“The analytical frameworks are intact. The timing was wrong. There is a difference. The scorecard marks both.”
— The Sports Page, Sunday Edition No. 008Over-Reactions and Under-Reactions
Over-Reactions
Issues #59 and #60 treated the Mets’ injury timeline as settled fact rather than a distribution with a tail. The phrase “expected before the end of May” was used without a hedge. The analytical conclusion about what the returns would mean for the offense was well-supported. The confidence in the specific date was not. Medical timelines for soft-tissue injuries are point estimates with wide confidence intervals; the newsletter treated the mean as a certainty. Going forward: injury timelines will carry an explicit uncertainty acknowledgment before they enter a projected-win calculation.
Under-Reactions
The newsletter under-reacted to Carolina’s trajectory after the Issue #56 embarrassment. Last Sunday’s correction focused on the publication-timing error; the underlying historical comp — 3 of 4 teams with 8+ unbeaten playoff starts won the Cup — got a footnote rather than its own section. Meanwhile, the Hurricanes swept Montreal in five games and are now in the Final with home-ice advantage and a 12–2 playoff record. The historical analysis deserved a cleaner standing in the scorecard. It will get one next Sunday when the Cup is decided.
What Readers Read · May 24–May 30
Readership data pending — the analytics pipeline installed recently, and last week’s counts were not captured. This section will populate starting with the first full week of data.
The Road Ahead
The queue entering June carries nine pieces, with three that have become significantly more urgent since they were drafted. The Mariners piece — built around Seattle’s 54% pre-season scenario probability — is newly relevant with Cal Raleigh still on the IL and the Mariners at 30–29, improbably leading the AL West. How a team projected for 85 wins is actually performing at that pace without its primary catcher is the question that piece was made to answer.
The superstition piece (a formal statistical treatment of why slump-busting rituals always appear to work) ties directly to Issue #63’s Raleigh-in-the-shower narrative. Both share the same Bayesian regression-to-mean framework. Issue #63 named it as a forthcoming methods piece. It publishes next.
The Mets sell-or-hold question (Issue #60 in the queue) was drafted when the Mets were a plausible buyer at 20–26. They are now 25–33, on pace for 72 wins, and playing without their two best position players. The trade-deadline calculus at 25–33 looks meaningfully different from the one the piece was written for. The newsletter will review the piece before publication to ensure the framework matches the current team. If it no longer fits, it gets revised or held.
In hockey: the Stanley Cup Final opens Tuesday in Raleigh. Game 1, Vegas at Carolina, 8 p.m. The historical comp from Issue #56 — 3 of 4 prior teams with 8+ unbeaten playoff starts went on to win the Cup — now applies to a 12–2 Carolina team with home-ice advantage. That comp gets its full grade in Sunday Edition No. 009. Vegas is the counter-argument: they also went 8–0 in their first two rounds before sweeping Colorado. This is two elite teams in peak form. The newsletter will be watching.
“Sixty-four issues in. The Bayesian model is working. The medical calendar was optimistic. Carolina is in the Final. The math updates when the data does.”
— The Sports Page, on the long game
