Nine Mets Are at Their Career Floor. Simultaneously.
We pulled the full career record of every player on the Mets’ July 24 roster snapshot. Five are having the outright worst season of their careers (including two 11-plus-year veterans) and four more are at their second-worst. Then we checked the expected stats: for seven of the nine, the results are meaningfully worse than the underlying performance. This is why the model grades the roster 11th in baseball and refuses to grade the team like 43–60.
BBMI Research Desk
·
July 24, 2026
Career-floor census
The usual story about a 43–60 team is that it’s bad. Bad players, playing badly, getting the record bad rosters deserve. That’s the story the standings tell about the 2026 Mets, and it has one problem: the model doesn’t believe it. Our talent lens, which grades rosters record-neutral on established ability, has stubbornly ranked this roster 11th in baseball all month while the team sat 25th on the power-rankings board. Somebody is wrong: the lens, or the standings.
So we ran the census. We pulled the complete major-league career record of all 26 players on the Mets’ July 24 roster snapshot (every season, every plate appearance and inning) and asked one question per player: is 2026 the worst season of your career? Eighteen players have enough career track record to answer honestly (thresholds below; rookies don’t have a “worst season” yet, they just have a season). The answer is the kind of thing you double-check before publishing. We double-checked.
The census
| Player | Metric | 2026 | Prior career-worst | Career seasons | Status |
|---|
Hitters graded on OPS (qualifying prior seasons ≥200 PA; 2026 ≥100 PA) · pitchers on ERA (prior ≥40 IP; 2026 ≥20 IP) · stats through July 23.
Five outright career-worsts: Francisco Lindor (.697 OPS against a prior floor of .734, across eleven seasons), Marcus Semien (.601 against .669, across twelve), Luis Torrens, Cionel Pérez, and Kodai Senga, whose 8.85 ERA against a career built on the low 3.00s is less a decline than a genre change. Four more at second-worst: Bo Bichette, Brett Baty, Devin Williams, and Freddy Peralta. That is nine of the eighteen assessable players, half the measurable roster, at their personal floor or one notch above it, in the same season, on the same team. Jorge Polanco (.456 in 93 plate appearances) didn’t even qualify for the table, which may be the most merciful sentence in this article.
So which of them earned it?
A career-worst line answers what happened, not why. For that we added Statcast’s expected stats (xwOBA for the hitters, xERA for the pitchers), which grade the quality of contact and ask what each player’s results should have been on the same balls in play. The gap between actual and expected is the closest thing baseball has to a luck meter. Here is the meter, run over all nine floor-dwellers:
| Player | 2026 actual | Expected | Gap | Interpretation |
|---|
Hitters: wOBA vs. Savant xwOBA (gap in points; + = deserved better). Pitchers: ERA vs. Savant xERA (gap in runs; + = ran worse than deserved). Expected stats as of the July 25 pull; both columns computed on the same plate appearances.
The count: seven of the nine are producing results meaningfully worse than their expected numbers; only two look fully supported. Torrens (.278 actual, .281 expected) and Cionel Pérez (5.50 vs a 5.56 xERA) are having exactly the seasons their contact quality says they should: genuine down years, no asterisk available. Six others are running well behind their inputs: Semien’s +.040 gap is the largest on the team (his .268 wOBA sits on .308-quality contact, a mediocre season rather than a twelve-year floor), and Lindor (+.031), Bichette (+.029), Baty (+.025), Williams (a 4.54 ERA on 3.44 xERA contact), and Peralta (5.01 on 3.83) all follow the same pattern. And then there is Senga, who gets his own sentence because rounding him either way would be a lie: his +3.43 gap is the roster’s largest, but the 5.42 xERA underneath it would still be his career worst. The gap didn’t invent Senga’s bad season; it promoted a bad season into a catastrophe. One honest limit on all of this: expected stats measure contact quality, not everything: they can’t cleanly separate variance from positioning, sequencing, health, or age. What they establish is narrower and still striking: the results are worse than the underlying performance, seven times over.
The weird counterweight, and honesty demands it: the Mets’ three career-best seasons are all relievers whose ERAs are outperforming their xERAs: Raley’s 1.96 on a 3.51, Minter’s 1.71 on 3.49, Weaver’s 1.98 on 2.80. So the same roster has bad variance at one end and favorable variance at the other, sorted neatly by role. The 2026 Mets aren’t a bad team so much as a controlled demonstration of what happens when the coin lands wrong for the lineup and right for the eighth inning.
The two that shouldn’t happen together
Torrens and Senga have short careers; short careers hit new floors easily. The census gets strange with the veterans. Lindor has eleven qualifying seasons. Semien has twelve. A player’s worst season, by definition, happens once per career, and the longer the career, the harder a new floor is to set, because there are that many more old seasons to fall beneath. For two teammates with careers that long to both be setting their floor in the same summer is the roster-construction equivalent of both engines failing on the same flight. It happens. It is not the way to bet.
Five careers, one direction: season quality within each career
Each line is one player’s full MLB career · higher = better within that career · red dot = 2026
Every red dot is the lowest point on its line. Lindor’s and Semien’s lines are eleven and twelve seasons long. New floors are supposed to get harder to set, not easier.
The whole roster, one chart
The census cuts both ways, so here is all of it. For each assessable player, we placed 2026 within their own career: what share of their prior qualifying seasons were worse than this one? Zero percent means a career-worst. One hundred means a career-best, and the roster has three of those, because baseball refuses to be tidy: Luke Weaver (1.98 ERA), Brooks Raley, and Huascar Brazobán are all having the best seasons of their careers in the same bullpen where Pérez and Senga are having their worst. And then there’s Juan Soto, hitting .947, the 57th percentile of Juan Soto, which is the whole problem with grading Soto: his median is everyone else’s apex.
Where 2026 sits in every assessable Met’s career
Share of each player’s prior qualifying seasons that were worse than 2026 · 0% = career-worst, 100% = career-best
The pile-up on the left wall is the finding. A normal roster scatters across this chart; the 2026 Mets have nine players at or next to the floor and a franchise cornerstone parked at 0%.
What the model does with this
Here is the actuarial reading, and it is the entire reason the Mets sit on the watch list instead of the write-off pile. When one player has a career-worst season, the boring explanation (he got worse) competes with the interesting ones: injury, bad luck, a swing change that didn’t take. When half a roster does it at once, the boring explanation stops scaling. And now the expected stats have adjudicated it directly: six of the nine floor-dwellers are hitting or pitching better than their results, with the team’s franchise shortstop and its twelve-year second baseman carrying the two biggest luck penalties in the lineup. Some of this is age, and Torrens and Pérez remind us some floors are honestly earned. But the historical base rate for veterans producing career-worst results on better-than-that contact is regression toward the established record, which is exactly what the talent lens prices when it grades this roster 11th while the standings say 25th.
The uncomfortable timing is that deadline week is when this analysis stops being academic. A front office that reads the census as “bad players” sells the roster for parts and the gap resolves the grim way. A front office that reads it the way the model does (a good roster having everyone’s worst year at once) holds, retools around the floor-setters, and lets the base rates work. At 43–60, seventeen games under, nobody is claiming October. The claim is narrower and stranger: this is probably the best 25th-place roster in baseball, and next season’s projection will believe the lens, not the epitaph.
The bottom line
Five Mets are having the worst season of their careers and four more are one notch above it: nine of the eighteen players with enough track record to judge. For seven of the nine, the expected stats aren’t nearly as bad as the results, with Semien’s +.040 gap the largest on the team; only Torrens and Cionel Pérez look fully supported, and Senga’s collapse is real but result-inflated. The model’s position is unchanged and now has its evidence in two tables: the roster is 11th; the season is 25th; and most of the gap between them is variance, not decline.
Method & caveats: career records pulled from MLB’s official statistics API for all 26 active-roster players, stats through July 23, 2026; multi-team seasons combined from counting stats. Hitters measured on OPS (prior seasons qualify at ≥200 PA, 2026 at ≥100 PA); pitchers on ERA (≥40 IP prior, ≥20 IP in 2026); “career-worst” requires at least two qualifying prior seasons. Expected stats (xwOBA/xERA) are from Baseball Savant, pulled July 25 with the roster as of July 24; each gap compares Savant’s actual and expected on the same plate appearances, which is why Savant’s actuals can differ a hair from the census columns (different pull dates and sources; both stated, neither adjusted to match the other). Honest limits: 2026 is a partial season compared against full ones; OPS, ERA, and the expected stats are unadjusted for park and league environment; expected stats measure contact quality, not defense, sequencing, or health; and the census is a point-in-time snapshot of the July 24 roster: players traded or released before then aren’t counted, and transactions after it (Soto’s July 25 IL placement among them) don’t alter what the snapshot measured. None of these caveats moves the headline: they cut in both directions, and the left wall of Fig. 2 doesn’t care.