There’s only one playoff race this year. That might not sound right to you. The Yankees and Orioles are deadlocked in the AL East. The Guardians are holding the Twins and Royals at bay in the Central. The Padres are looking menacingly northwards toward LA. But those aren’t playoff races, because everyone involved is making the playoffs either way. The only race where the winner is in and the loser is out is the one for the last NL Wild Card spot, and it’s taking place between divisional rivals: the Mets and the Braves.
It’s weird seeing so little actual drama in September. Those two teams are the only ones with playoff odds between 20% and 80% right now. That’s rare for this time of year. There were seven such teams last year, six in 2022, and three in the top-heavy 2021 season. Even if we go back to the 10-team playoff era, the years from 2015-2019 averaged four teams in that 20-80% range with a month left in the season. Despite no truly dominant teams, the playoff races are abnormally set in stone this year.
That only makes the last race more exciting, though. In one corner, we have the Braves, who came into the year as the best team in baseball. They were so good, in fact, that we used them as a model when delving into some new depth chart data this spring. What might the Braves look like if they lost Spencer Strider and Ronald Acuña Jr. to injury? Our model thought they’d be a .551 team.
In a strong case of reality mirroring forecasting, they are missing both of those players, and a few key contributors have missed chunks of time to boot. Perhaps not coincidentally, they’ve played to a 76-64 record, a .543 winning percentage. It’s a classic story: The Braves are a great team playing without their very top players and doing fairly well anyway. The division is out of reach, but not many teams win their division in a year where the reigning MVP and a top five Cy Young finisher both miss most of the season.
The other side of the race is a band of merry overachievers who just happen to play on a historically heartbreaking team. The Mets didn’t come into the season as playoff favorites – our odds gave them a 30% chance of qualifying for October, broadly in line with other public projections. This was supposed to be a down year for them, the bridge between last year’s Justin Verlander/Max Scherzer bonanza and a future filled with Uncle Steve’s marquee additions and ticker tape parades.
That might still be the way things go. We give the Braves a substantial edge in our odds – 67.9% as compared to 45.9% for New York. But that’s not the way the race is going at this exact moment. The two teams are tied for the last Wild Card spot with 22 games left to play. Dead heat in the standings; lopsided odds to make the playoffs. That sounds pretty strange.
The first thing I like to do when I see an unintuitive result like this one is to compare other predictions to the outlier. If we’re calling this an unbalanced race while every other statistical model is saying it’s a tossup, there’s more digging to be done. If we’re broadly in line with everyone else, that’s just as revealing.
Baseball Prospectus maintains their own playoff odds. They’re constructed meaningfully differently than ours, and we don’t share methodology or anything, so aside from both being based on the concept of projecting future performance, there isn’t a ton of overlap. They agree with our take on the race – they have the Braves with a 69.1% chance of making the playoffs and the Mets with a 39.4% chance. Baseball Reference isn’t as sure – they have the Braves at 54.9% and the Mets at 48.1%. Finally, gambling markets have it 66% to 48%, pretty close to both us and BP.
I was already inclined to take our model at its word, and the evidence only leads me further in that direction. Baseball Reference’s model is explicitly backward-looking; that’s a feature rather than a bug, because it makes it much easier to calculate and more intelligible, but at least for me, that means that our model has an inherent advantage in predicting the future. In past years, I’ve tested the difference between these two methods, and found that incorporating projections leads to better outcomes.
Now that that’s settled, there’s another obvious question: Why? That’s a huge gap in odds considering each team’s record. You’d expect a tied race to have broadly similar odds this late in the year. This isn’t a season-spanning marathon; it’s a sprint so short that it feels like a pure toss-up. They have the same record now, and they don’t have many games left. Why shouldn’t they be even?
Each team has 22 games remaining. There’s no difference in the quantity, in other words. The quality, on the other hand? The Mets have seven games against division leader Philadelphia, three against Milwaukee, and three against the Braves. Those are all tough ones. They do have a total of nine games against the Reds, Blue Jays, and Nationals, but they have more hard games left than easy ones.
Meanwhile, the Braves have a slightly easier road. They have their own nine games against the Reds, Blue Jays, and Nationals. They also have three against the Marlins. They play the Dodgers four times, which is always tough, and close the season with three against the Mets and three against the Royals. We think that they’re going to play opponents with an aggregate .491 true talent the rest of the way, while we peg New York’s opposition at .509 true talent. That’s a fairly large gap – over a full season, that’s something like four wins in expected difference. It’s more like a half a game over the remainder of the season.
Another feather in Atlanta’s cap? They play at home for the three games the two teams will play against each other. That series will determine the playoff tiebreaker – they’re knotted at 5-5 so far this season, so whoever wins this three-game series will also win the season-long one. Having home field advantage in those games really matters. Home teams win around 54% of the time in baseball over the long run. A 54% chance of winning a single game translates to a 56% chance of winning a three-game series, so even if the Mets and Braves were evenly matched, the tiebreaker edge would go to Atlanta.
That just leaves the biggest argument in favor of the Braves: They’re better. There’s nothing that suggests that from the way the teams have played in 2024, to be clear. The two are extremely close in run differential – both are scoring and allowing runs at the pace you’d expect from borderline playoff teams. Both have the underlying numbers to support it. BaseRuns thinks that the Braves should be 77-62 and the Mets 78-62. That’s a statistical dead heat. The Chris Sale-led Braves have allowed far fewer runs, but Francisco Lindor and the Mets offense have scored more.
That counts for a lot. Current performance informs all of our projection systems. To give an example, Mark Vientos came into the season with a .319 wOBA projection. That’s a slightly above-average offensive line, good for a 103 wRC+. So far this year, he’s been much better than that. He has a 145 wRC+ on the season, and we’re now projecting him for a 116 mark (.331 wOBA) the rest of the way. That’s better than his career numbers, even counting this year’s outburst, because projection systems weigh recent performance more heavily than data from years ago.
A lot of Mets fit into that category. Lindor and J.D. Martinez have batted a lot and beaten their projections by a mile; we expect both to be exceptional down the stretch. Even the Mets’ disappointing hitters aren’t doing particularly poorly this year. Pete Alonso and Brandon Nimmo were expected to be offensive linchpins, and while they haven’t quite hit their preseason expectations, both of our models continue to expect good things from them the rest of the way.
Atlanta, on the other hand, has weathered a season of offensive disappointments. Marcell Ozuna has been outstanding, but he’s the only Braves hitter with even a 120 wRC+. For comparison, six Atlanta regulars hit that mark last year. This isn’t just a matter of missing Acuña, though that obviously hasn’t helped. Matt Olson is in the middle of the worst year of his career. Sean Murphy’s shocking transformation into an offensive pumpkin couldn’t have come at a worse time. Austin Riley and Ozzie Albies had down seasons and are both on the IL. Michael Harris II didn’t quite crack that list (115 wRC+), but he’s hitting a desultory .251/.293/.375 this year.
So yes, that’s all a bummer for Atlanta. But while a down 2024 has certainly lowered both our projection systems’ opinions of these players and my own expectations for them, it’s not like 2023 didn’t happen. If you count 2024 twice as much as 2023, Matt Olson has a 125 blended wRC+. Pete Alonso checks in at 122. Harris has a 94 wRC+… and counterpart Harrison Bader is at 81. Murphy, who is having by far the worst season of his career after five straight solid offensive seasons, averages out to a 101 wRC+. Francisco Alvarez, the young catcher everyone (including me) has been drooling over for two years, is at 92.
In other words, there might be some disappointing offensive performances in Atlanta, but they’re disappointing largely because the players compiling those lines have already shown us that they’re more talented than that. That’s kind of the point of projection systems – they stop us from overreacting to data just because of pesky dividing lines like “years” and “seasons.” It’s important to consider all the information that’s available about players when you’re trying to figure out how they’ll do next year. Do you think Bobby Witt Jr. is going to put up a 172 wRC+ next year? Probably not! Do you think Corbin Carroll is going to improve on his 106 wRC+? Almost certainly. Do you think Alex Verdugo is as bad as his 84 wRC+? Maybe don’t answer that one.
You can look at the season-to-date stats all you want. You can talk about how the Mets have clutch hitters and guys peaking at the right time and exciting youngsters. But do you really think that they’re going to keep outscoring the Braves by half a run a game? That sounds unlikely to me, at least as a central expectation. Our projections have the two teams even in run scoring against neutral opposition for the rest of the year. Sure, Atlanta is depleted, and the team’s best healthy hitters (outside of Ozuna) aren’t lighting the world on fire. But even without their best hitters, the Braves have plenty of offensive weapons. And hey, if you’re intent on slicing things into tiny samples where only the most recent data matters: the Braves have out-hit the Mets in the past month. Only caring about recent samples cuts both ways, after all.
If you accept that the two offenses are roughly the same, the reason Atlanta is favored comes into focus. The Braves, despite Strider’s injury, have an excellent pitching staff. Sale is the slam dunk Cy Young winner. Reynaldo López has been a revelation. Max Fried is great as always. Spencer Schwellenbach looks like yet another Braves minor league success. The bullpen is the third-best in baseball, with an embarrassment of late-inning arms.
Meanwhile, the Mets are dealing with injury issues of their own. While the Braves are missing several talented hitters, the Mets are feeling the pinch on the mound. Kodai Senga, their ace, has only made one start all season and won’t be back before October at the earliest. Christian Scott, their top pitching prospect and a welcome source of upside, hasn’t pitched since mid-July. Paul Blackburn was one of their biggest deadline acquisitions, and yep, you guessed it, he’s hurt.
The Mets replacements have been solid when called upon. David Peterson is having a career year, with a 2.75 ERA and 4.00 FIP. We think he’ll be somewhere in the 4.00 ERA range the rest of the way, which is quite impressive. Tylor Megill is on the opposite side of the FIP-ERA gap, with a 4.95 ERA that belies his 3.79 FIP. But those guys are back-of-rotation fillers with third starter upside, and they’re replacing the team’s best pitcher.
On the season, Mets starters are somewhere between 15th (RA9) and 23rd (FIP) in the league in WAR. That’s partially because of some bad performances from since-jettisoned options. Adrian Houser was awful before getting DFA’ed. Blackburn was ineffective before getting hurt. But some of it comes from the current rotation. Jose Quintana is showing signs of a late-career swoon. Luis Severino is a workhorse, but his run prevention numbers are right around league average. Sean Manaea, the nominal ace, is roughly 30th in WAR league-wide.
None of this feels particularly out of step with these pitchers’ previous careers. We’re projecting the Mets for one of the worst rotations in baseball the rest of the way, 25th in WAR, just like they’ve been pretty bad for most of the year. Meanwhile, Atlanta’s group came into the year already looking like the better unit, and they’ve done nothing but improve their projections since then. It didn’t take our models long to believe in Chris Sale’s resurgence; when the guy with a career 3.00 ERA starts pitching to a 3.00 ERA, you can do the math. López’s projections haven’t budged much, so if anything, we’re docking Atlanta’s rotation some points. Heck, we didn’t even project Schwellenbach for a single inning in the preseason, and he looks sensational.
Throw in the bullpen gap – the Mets’ unit is average, nothing like Atlanta’s dominant group – and we’re talking about a 0.37 run-per-game gap in projected run prevention. Meanwhile, the two offenses have identical projections. You should account for park effects here when you’re doing cause and effect – Citi Field suppresses offense, which makes the Mets’ hitters look worse and their pitchers look better – but the overall point is clear. The big thing we see being different between these two teams is that we think this year’s iteration of the Braves just has better pitchers.
How much better in probabilistic terms? Luckily, we can answer that question. I took the Braves’ projected full-season runs scored and allowed totals. That’s a +79 run differential and an 89.6-win Pythagorean projection. Add 0.37 runs per game to what they allow, and they’d drop to an 82.9-win projection. Over 22 games instead of 162, that works out to 0.9 wins.
I estimate that half a win is worth roughly a 5% odds differential. I took that from looking at the one-day change in Atlanta’s playoff odds in our coin flip model when they lost yesterday while the Mets were off. If half a win is five percentage points, that means 0.9 wins comes out to nine percentage points. We know from up above that the strength of schedule discrepancy comes out to about half a win the rest of the year; that’s another five percentage points, 14 points total.
If New York and Atlanta each had the same playoff odds (they add up to more than 100% since they could nip an NL West team for a slot), they’d have 57% odds. We have the Braves 11 percentage points higher than that. That’s basically the same as the 14 I’m estimating for strength of schedule and run prevention differential. In other words, the back of the envelope math works here.
Why does our playoff odds model make the Braves a strong favorite over the Mets despite similar records? It’s because it considers the makeup of both teams as well as their remaining schedules and reaches an informed conclusion. You don’t have to agree with all of its inputs. If you think that the Mets pitching staff is better than our numbers, feel free to mentally tighten that gap. If you think Atlanta’s offense will start sputtering, hey, you’re the boss.
But if you think our assumptions about team strength are reasonable, the result makes sense. The model didn’t give the Braves a 22-point gap in playoff likelihood on accident. It did so because we project them to allow fewer runs than the Mets against neutral opposition the rest of the way. They’re playing against easier opposition, though, not neutral. Add those two up, and you can entirely account for the difference in our odds. We’re projecting the Braves as 68%-46% favorites in this race because we project them to win more games – 1.2 more, at last model update.
That feels like a satisfying answer, at least to me. The model isn’t pulling numbers out of thin air. Its outputs are right in line with what you’d expect if you solved everything from first principles. Like I said, the projections aren’t gospel, just our best estimates. But the model is putting out exactly what we’d expect given our inputs. We think the Braves are just better, and clearly the projections feeding other playoff odds estimates do too. We quantify that belief with projections, put it into the model, and get our answers: Even though the Braves and Mets are tied in the standings, the Braves are meaningfully more likely to still be playing come October. That doesn’t predict the future – but I do think it explains the situation on the ground right now quite well. It’s an uphill battle for New York, but a win tomorrow paired with an Atlanta loss would pretty much leave the teams on even footing.
