The Mets’ Next 17 Games, Mapped.
The Mets sit at 10-21 and have, between Saturday and the eighteenth of May, six series and seventeen games to play. The Sunday Edition has framed where they sit. Saturday’s issue framed what 10-21 has historically meant. This piece does the operational version of the same exercise: takes the schedule the Mets actually face, applies a probability model the analytics community has used for a generation, and produces a number for each series.
The number is the log5 probability — named for Bill James’s 1981 formulation, which lets you combine two team-level win rates into a head-to-head expectation. The formula assumes the league average is .500, which it is by construction, and ignores everything except each team’s underlying quality. It does not know about Soto’s return. It does not know about Lindor’s availability on a given day. It does not know which side of Devin Williams will show up. What it does know is what each team has done so far and how that maps to a single, comparable rate.
The Setup, In One Equation
The log5 Formula
For each team we estimate true-talent win rate from run differential per game (RD/G), using the long-known approximation WPct ≈ .500 + (RD/G)/10. This is more stable than the raw record over thirty games — the Mets’ .323 record probably understates them; their RD/G of −1.13 says their honest level is closer to .387. The model uses the honest number.
The Six Series
| Dates | Opponent | Opp’s True % | P(Mets win/game) | Expected Wins |
|---|---|---|---|---|
| May 2-3 | @ LA Angels | .419 | .467 | 0.93 of 2 |
| May 4-6 | @ Colorado | .478 | .408 | 1.22 of 3 |
| May 8-10 | @ Arizona | .627 | .273 | 0.82 of 3 |
| May 12-14 | vs Detroit | .528 | .361 | 1.08 of 3 |
| May 15-17 | vs NY Yankees | .652 | .252 | 0.76 of 3 |
| May 18-20 | @ Washington | .355 | .534 | 1.60 of 3 |
Sum the expected wins across all six series and the model returns 6.41 wins out of 17 games. That is a .377 pace — better than the team’s current .323, but only slightly. Better, fundamentally, because the schedule includes the Nationals (worse than the Mets by every available measure) and the Tigers (around .500). Worse than even-money on the days they face Arizona or the Yankees, who would have been favored against any team, never mind the Mets in their current state.
The single series where the Mets are projected favorites is the one against Washington at the end of the window. That is the entire margin in the schedule.
Three Scenarios, In Cumulative Record
| Scenario | Pace | Result, 17 games | Total record | Implication |
|---|---|---|---|---|
| Playoff push | .600 | 10–7 | 20–28 | Best plausible outcome. Still 11.5 GB. |
| Mid-tier (model) | .377 | 6–11 | 16–32 | Treading water at a sub-.400 pace. Wild card pace requires .615 the rest of the way. |
| Collapse continues | .323 | 5–12 | 15–33 | Effectively eliminated by the All-Star break. |
The conditional-probability piece on Saturday made the historical case: roughly eleven percent of teams that started below .400 in the last decade made the playoffs. The forward-projection piece — this one — makes the operational case: the next seventeen games are not friendly. They include the Diamondbacks at Chase Field and the Yankees in the subway series. The model says the Mets, if everything breaks honestly to true talent, walk out of May 18 at sixteen wins and thirty-two losses. That is not the end of the season. It is, however, the end of the cushion.
“The model gives them six wins. The schedule gives them one favorable series. Believing requires the schedule to lie.”
— The Professor, on operational pessimismWhat the Model Does Not Know
Three honest caveats. First, the run-differential approach treats every team as a constant; it does not know that Soto returned the second week of April or that Williams might be replaced in a leverage role. Second, the league constant is an approximation; the actual league mean is whatever it is on any given day, and small drift can move the numbers by one or two percentage points per game. Third, log5 assumes independence between matchups, which baseball does not respect — an exhausted bullpen carries forward; a hot bat clusters across consecutive series.
That said, the magnitude here is robust. The Mets are a .377 team facing a schedule whose average opponent quality is roughly .510. The model says they will win about six. They might win seven. They will not win ten.
The Sunday Edition will rerun this piece on May 17 against actual outcomes. If the Mets are at thirteen, fourteen, fifteen wins by then, the door has closed audibly. If they are at sixteen or seventeen, the model will have undershot and the conversation about June and July restarts.
