2020 was a strange season of baseball. I was glad we had a season at all, but the short schedule was enough to take me out of it to a large extent. I swear it had nothing to do with the fact that the Nationals played terribly all season. Actually… it was entirely due to that. When the playoffs came around, I got back into it. With the exception of having more teams and more games, the playoffs felt like real baseball again. But when I saw that the Dodgers and Rays, the teams with the best records in the regular season, had made the World Series, I was reminded of the anomaly of baseball in the time of Covid.
I began to wonder, had the two teams with the best records made it because it was a short season? How often had the best two teams made it to the Series? The second question was easily answered. In the wild card era (since 2012), it had only happened once when the Red Sox and Cardinals met in the 2013 Championship. Those teams had identical 97-65 records and played a competitive six game series. But otherwise, the World Series has been consistently a mix-up of teams from farther down the pegging. Had 2020’s matchup occurred because the season was shorter? Was a sample of sixty games a superior predictor of playoff success than a full 162?
That question was a bit harder to answer. To repeat, my hypothesis was because the last sixty games was the whole season, that the way the team played in those games would be better reflected in the way they performed in the playoffs. Perhaps this would be because rosters would be closer to opening day rosters. Injuries would be less likely to create a dramatic difference. Maybe there is a “momentum” to a team. I know that is a concept that has been statistically discredited, but as a fan it’s hard to shake the belief in a team getting on a roll or firing on all cylinders. Anyone who watched the 2014 Royals barely make the wild card, get behind in the one game playoff, then completely turn around and look invincible would be persuaded.
But that is anecdote, so I turned to the numbers. My plan was to consider that records for every playoff team since the 2012 season opened up the field to two wild cards. I thought that considering anything before that era would muddy the waters. So, I totaled up the overall record and the record in each team’s last sixty games. Then for fun, I threw in their record in the last thirty games. If my hypothesis held for sixty, perhaps it would hold even stronger for thirty.
It took time, but I went on Baseball Reference and scanned through each team’s season to get my numbers. Time consuming, but simple. Now I had a chart of the 64 teams who made the playoffs from 2012-2019 and their records in three samples. But in order to get a correlation, I needed to quantize success somehow. My solution was to assign one point for each postseason win. I skipped the wild card game because I thought granting a bonus point to every team that won that game would skew the results. This meant a maximum 11 points to a team that had won the World Series and a minimum of zero for a team knocked out in a Division Series sweep. I thought it was fair to credit a losing team with a few points because in my mind a team that drives the series to a final game is better than a team that gets swept.
As I filled in the table, I got excited seeing cases that confirmed my hypothesis, the 2012 Giants, the 2019 Nationals for example, both wild card teams that had been much better in the second half of the season. But that was confirmation bias, and I could feel it as I saw counterexamples that made me wince like the 2018 Red Sox, a team that slumped in August and September only to roar to a resounding win in the postseason. We need numbers to tell the truth, gut instinct is no substitute. There was a great deal of noise in the tables, more than an intuitive glance could filter.
When I put the win percentages in a column, and the win points in another, I ran a correlation. The results were disappointing to my hypothesis. I had very little invested in proving it right and I still felt a pang. I can only imagine how a researcher who spends months or years on a project feels. It is no wonder that there is so little published work that proves a hypothesis wrong. It’s easy for a lazy reader to dismiss a disproved hypothesis as foolish with 20/20 hindsight. Once the results are in, it’s easy to say they were always obvious. In this case though, I think it’s good to have numbers to disprove an intuitive guess. It turns out that the overall record has a .305 correlation to playoff wins. The sixty-game record drops down to a .111 correlation and the thirty-game record is nearly neutral at .073. It turns out the 162-game schedule is a much better predictor of playoff success than how hot a team is going in. Who knew?
Well, I suppose baseball knew.
The Attractive Nuisance 