/cdn.vox-cdn.com/uploads/chorus_image/image/67551383/1228756634.jpg.0.jpg)
Expanded playoffs are likely here to stay, and we’ve already seen how opening up the contention window changes the tenor of the regular season. Ostensibly, there was meaningful baseball played until the very final pitch of the regular season even though the playoff picture was more or less set in the American League weeks ago. In the National League, a handful of teams were still battling for the final few playoff spots, and every other contender (except the Dodgers) was jockeying for seeding position.
Opening October to 16 teams doesn’t remove division chases from the regular season, but it does move them. Instead of the best teams fighting for their lives in the final weekend, we saw several teams fecklessly flailing for the last life raft. After the Giants dropped three out of four against the Padres, the Brewers backed into October on a day they lost 5-1 to end a season where they never had a winning record.
Under this system, the do-or-die game 162’s will largely be played by mediocre teams, while fans of the good-to-great teams will be rooting for seeding. Seeding is obviously important. A team would much rather face a weaker opponent. In the NBA, the eight seed has only won 5 out of 74 first round series against the number one seed. Baseball, though, is much more random than basketball, and even wretched teams can upset excellent teams in a three-game series. For baseball fans, is seeding important enough to get worked up about?
Using code written by Jim Albert, I simulated 1,000 60-game seasons using the Bradley-Terry model which means that talent levels were randomly assigned to the 30 MLB teams. A team with a talent level of zero would be estimated to win roughly half its games. A team with a talent level of 0.2 would win 55 percent of its games (.550 winning percentage), and a team with a talent level of -0.2 would win 45 percent of its games (.450 winning percentage).
For each season, I assigned seeds one through eight with the first six going to division winners and the final two going to the Wild Card winners. In actuality, seeds one through three should go the teams with the best record in each division, seeds four through six should go to the second division winners, and seeds seven and eight go to the Wild Card winners. That’s why the Padres, the second-best team in the NL, are the fourth seed rather than the second. How I decided seeding may not be correct, but it is more fair and it gives teams more to play for. Also, it took me all weekend just to figure out how to assign seeding, and I wasn’t about to start over.
On average, playoff teams were better than .500 teams though in real-life and in simulations, it’s common for teams with losing records to reach the postseason or even win the World Series.
As we would expect, talent level gradually declines over the first six seeds, but then bounces back up slightly once we reach that first Wild Card team. It’s somewhat surprising, but I also don’t think it’s controversial to say the Reds are better than the Marlins. Note that a seed of zero means the team did not reach the postseason.
:no_upscale()/cdn.vox-cdn.com/uploads/chorus_asset/file/21918837/Talent_Level_by_Seed.png)
Using the average talent level of each seed, I made each matchup play each other 10,000 times to get an estimate of how likely it was that each seed would win an individual game against their first-round opponent. From there, I calculated the probability of each seed winning the division. What I found was that each spot an average team climbs in the seeding raises their odds of advancing past the first round by about three percentage points.
Seeding in All Simulations
Matchup | High Seed Wins | Low Seed Wins | Talent of High Seed | Talent of Low Seed | High Seed Wins Series% | Low Seed Wins Series% |
---|---|---|---|---|---|---|
Matchup | High Seed Wins | Low Seed Wins | Talent of High Seed | Talent of Low Seed | High Seed Wins Series% | Low Seed Wins Series% |
1, 8 | 5803 | 4197 | 0.377 | 0.016 | 62.4 | 37.6 |
2, 7 | 5584 | 4452 | 0.273 | 0.085 | 59 | 41 |
3, 6 | 5444 | 4556 | 0.212 | 0.029 | 56.5 | 43.5 |
4, 5 | 5198 | 4802 | 0.154 | 0.108 | 52.9 | 47.1 |
Of course, the determining factor in how likely a team is to advance isn’t their seed, but their talent. Even in my wrong-but-fair method of determining seeding, talent levels weren’t so cleanly distributed. There were great teams in the eight seed and horrible teams in the first seed simply because there’s a lot of variance in a 60-game season. That the 2020 Yankees could lose the East to the Rays isn’t all that shocking, but if you predicted they would have a worse record than Cleveland or the White Sox you’d be pelted by tomatoes.
Even in a 162-game season, teams can dramatically over or underperform their talent level. The 2018 Dodgers, for instance, nearly lost the NL West because they fell short of their Pythagorean record by 10 games and the Rockies outpaced theirs by six.
Picking a simulated season at random, the effects of seeding are much murkier.
Seeding in Unique Simulation
Matchup | High Seed Wins | Low Seed Wins | Talent of High Seed | Talent of Low Seed | High Seed Wins Series% | Low Seed Wins Series% |
---|---|---|---|---|---|---|
Matchup | High Seed Wins | Low Seed Wins | Talent of High Seed | Talent of Low Seed | High Seed Wins Series% | Low Seed Wins Series% |
1, 8 | 6322 | 3678 | 0.416 | -0.165 | 69.4 | 30.6 |
2, 7 | 5078 | 4922 | 0.107 | 0.088 | 51.2 | 48.8 |
3, 6 | 5814 | 4186 | 0.239 | -0.09 | 62 | 38 |
4, 5 | 5275 | 4725 | 0.232 | 0.131 | 54.1 | 45.9 |
The second seed in this league was a good-not-great team that actually did themselves a disservice by not losing more games. The seventh seed was more talented than the sixth seed, so if the second seed won the third seed instead, they would have had a better chance of advancing past the first round. It also should not be controversial to say that any system which could reward losing is a bad system.
That’s exactly what this current playoff format does. Sportsline released their odds for each team to advance and in each league, the number one seed doesn’t have the best chance to advance. In the AL, that distinction belongs to the fifth-seeded Yankees. The Rays actually have the lowest chance to advance despite having the best regular season record and run differential in the league.
Hopefully, we never see another 60-game season. In a 162-game season, talent and seeding figure to more closely align and the outliers won’t be so outlandish. Even then, the 16-team playoff format means the great teams are playing for seeding, and higher seeding is generally, but not always, good. While baseball fans tune into mediocre teams flopping lifelessly into October, the great teams are fighting for something that is, at best, a small benefit and at worst, a detriment. Expanding the playoffs is supposed to inject revenue excitement, but the battle for seeding just isn’t compelling.
Kenny Kelly is the managing editor of Beyond the Box Score.
Loading comments...