The Boston Red Sox were 71-49 on the morning of August 12, eight games ahead of the Yankees in the AL East, and the public baseball analytics community had spent the previous three weeks quietly updating its preseason projections. The Red Sox had been picked for 84 wins. They were on pace for 96. The Yankees, by the same projection set, had been picked for 95. They were on pace for 86. The Mets had been picked for 93. They were on pace for 79. Three weeks into the dog days of the season, the three most-watched teams in the league had moved 10 wins in either direction from where the preseason models had placed them. The August standings were producing the kind of mid-season surprise that the projection community had been warning about for three months.
The structural reason this happened is the same reason it happens every August: baseball is the sport where the regular-season sample is large enough to identify real talent gaps, but the daily noise within that sample is wide enough to produce two-month stretches that look completely different from the projected baseline. The Red Sox were probably not actually a 96-win team. The Yankees were probably not actually an 86-win team. Both were closer to their preseason median than the August standings implied. By the end of the year, the regression toward the projection would close most of the gap. By August 12, the standings were already producing playoff-implications conversations that the math did not yet support.
What follows is what the August 12 standings actually told us, where the public projection models were ahead of the daily standings narrative, and which playoff teams the underlying numbers were most confident about with seven weeks of season remaining.
What the August 12 standings actually looked like
The American League: Red Sox 71-49 leading East, Yankees 63-57, Blue Jays 64-56 (closer to wild card race than to division). Cleveland Guardians 65-55 leading Central by 3. Mariners 64-56 leading West by 1 over Astros. The AL was producing exactly the parity the preseason projections had suggested it would, with the Red Sox-Yankees inversion as the headline storyline.
The National League: Brewers 71-49 leading Central by 6 over Cubs. Phillies 71-49 leading East by 4 over Mets. Dodgers 68-52 leading West by 2 over Padres. The NL was tighter, with three division races still open and the wild card field crowded enough to include teams that the projection models had as bubble candidates.
The projection systems had been calling some of this. FanGraphs had updated their playoff probabilities on the morning of August 12 to give the Red Sox a 78% chance of making the postseason — up from 47% on July 1, but still below the 92% their actual record implied. The gap of 14 percentage points between the model and the standings is the regression band. The model thought the Red Sox would lose more games than their hot streak suggested, partly because the underlying pitching numbers were below their record-implied performance.
Where the underlying numbers disagreed with the standings
| Team | Record (Aug 12) | Run differential | BaseRuns expected record | Model gap |
|---|---|---|---|---|
| Red Sox | 71-49 (.591) | +92 | 67-53 (.558) | -4 wins overperforming |
| Yankees | 63-57 (.525) | +38 | 67-53 (.558) | +4 wins underperforming |
| Mets | 56-64 (.467) | -22 | 57-63 (.475) | +1 win (close to model) |
| Phillies | 71-49 (.591) | +98 | 69-51 (.575) | -2 wins (close) |
| Brewers | 71-49 (.591) | +76 | 65-55 (.542) | -6 wins overperforming |
The Red Sox and Brewers were both overperforming their underlying numbers by 4-6 wins. The Yankees were underperforming by 4 wins. The Phillies and Mets were close to their model expectations. The standings, when read against the model, were telling us: Boston and Milwaukee are probably real playoff teams but not 100-win teams; New York is probably a real playoff team but not the bottom-of-the-pack also-ran the standings suggest; Philly and the Mets are probably exactly where the model has them. Our small samples piece covers the broader pattern of how mid-season standings overstate variance in either direction.
Where this gets weird
The clean “Red Sox are overperforming, Yankees are underperforming, regress accordingly” reading misses three things.
The first is that BaseRuns and projection systems are themselves imperfect. The 4-win overperformance for Boston is not a guarantee they regress; it is the model’s best estimate of where their true talent sits. If the Red Sox have actually improved structurally — better bullpen, better defensive positioning, a real platoon advantage that the projection systems don’t fully capture — the regression will be smaller than the model expects. The 4-win gap might shrink to 1 or 2 wins, not all the way back to baseline.
The second is that the August trade deadline (which closed July 31 this year) had produced different roster constructions than the projection systems had as inputs. Several contending teams had added pitching depth, bullpen arms, or platoon bats that the projection models would not fully reflect for another 30-45 days. The Red Sox specifically had acquired two relievers at the deadline whose contribution to their hot streak was structurally real but had not yet been integrated into the model’s win-rate projection.
The third is that the wild card race in both leagues was tight enough that single-week stretches in late August could move teams in and out of the field. The Reds, Diamondbacks, Cardinals, and Pirates were all within 4 games of the third NL wild card slot. Any of them could plausibly finish in or out. The model’s playoff-probability outputs for those teams were all in the 25-55% band, which is the range where September performance decides everything.
The seven weeks ahead, by projection
- Red Sox vs Yankees series in early September is the AL East decider. Boston needs to take the series to maintain the structural advantage; New York needs to sweep to close the gap.
- The Phillies-Mets race will go to the final week. Both teams are projected within a single win of each other. The September head-to-head matchups (six games scheduled) will decide the East.
- Brewers running away from Cubs. Milwaukee’s lead is wider than the model expected. The Cubs would need a 15-5 stretch in September to catch them; the projections give that scenario about 18% probability.
- AL West stays tight to October 1. Mariners and Astros are within one game in both record and projection. Neither team has the structural advantage to pull away.
The callback
That August 12 morning when the Red Sox were 71-49 and the Yankees were 63-57 and the public baseball coverage was treating the inversion as the season’s central storyline was the cleanest expression of how the regular season’s daily noise distorts mid-season narratives. The Red Sox were a real playoff team but probably not a 96-win team. The Yankees were a real playoff team but their record understated them by four wins. The Mets were exactly the team the model had been describing all season. By October 1, most of the variance would close. The teams the underlying numbers were highest on — Phillies, Brewers, Dodgers, Mariners — would be the playoff favorites. The teams that ran hot in August would have come back to earth. The teams that ran cold would have caught up. The August standings are the loudest version of the regular-season noise. The October bracket is the quiet version that survives it. The playoff scaling piece covers an adjacent version of the same problem in basketball. The model has been right about most of these teams. The standings just have not caught up yet.
Projection data via FanGraphs; BaseRuns context via Baseball Reference; standings via MLB.com.
