What Are The Odds? Chasing A .700 Season In 60 Games

If you read The Athletic, you might have seen this article discussing some of the things that the small sample size involved in a 60 game season might make possible, things like a .400 hitter or Franmil Reyes leading the league in homers.  I want to look at another thing that a short season might make possible: a team with a .700 win rate.

Only 10 teams since 1901 have won 70 percent or more of their regular season games.  Four of those ten were between 1901 and 1910.  It’s only happened twice since the Braves moved to Atlanta: the 1998 Yankees and the 2001 Mariners.  But there’s about a 25 percent chance an almost 40 percent chance it happens this year, assuming this year’s 60 game season happens.

There are three factors which create this result.  First, of course, there are only 60 games.  Everyone knows that weird things can happen with small sample sizes.  For example, it’s why the playoffs are a crapshoot.  Simple calculations tell you that a team which has an underlying average win rate of 60 percent only has 4 chances in a thousand of exhibiting a 70 percent win rate in a 162 game season, but is ten times more likely to do so in a 60 game season. 

Second, the presence of COVID-19 as an additional risk factor into the cohesion of a team makes it very likely that the dispersion of underlying team qualities will increase, making 60 percent teams more likely, largely by making 30 and 40 percent teams more likely.  All it takes is one streak of infections in a single team to make their chances of putting a competitive team on the field for two weeks impossible.  At that point, their chances of winning fall and everyone else’s rise.  (Of course, if the infections hit a good team, that will make everything more equal, raising the chance that a poor team can somehow squeak into the playoffs.)

Third, though, and somewhat more subtle, is the fact that MLB is playing what are in essence a three league schedule with no interleague games.  Each league (East, Central and West) has two divisions (NL and AL) which play an unbalanced schedule: 10 games against each team in their own division and 4 against each team in the other division.  Depending on how the teams are distributed, this means that that a good team can play bad teams for a much higher percentage of its games. In particular, as we will see, one good team is particularly advantaged by the schedule.

So to look at the probabilities, I took the season simulator I wrote for the Crapshoot series and modified it to look at the dispersion of the best team’s win total in 162 game season, 60 game seasons, and, in a speculative scenario, a 60 game season in which there is a probability that a particular team could collapse owing to depleted ranks.

I seeded this simulation with the underlying power rankings derived from the 2019 season.  I separated home and road ratings for each team and adjusted the results for regression to the mean.  If that doesn’t mean anything to you, you can read the last series or ask me about it and I’ll try and answer, or request a copy of the Stata code I wrote and figure it out yourself.  But here are the ratings I derived (in the first two columns):

Just to remind people, if Team A at home plays Team B on the road, then the probability that Team A wins that game is Home Rating A/(Road Rating B + Home Rating A).  The ratings are normalized so that the highest rating (the Dodgers at home) gets 100.

The third column is the average of home and road rating, which I’m treating as the average strength of the team, just for exposition purposes.  The next two columns are the strength of schedule ratings; first based on the 2020 schedule and then based on the 2019 schedule.  There is a glaring change here: the Yankees’ expected win rate is considerably higher under the 60 game schedule, as is the Cardinals’ rate.  The Dodgers’ expected win rate falls slightly.  But assuming that the 2019 ratings are accurate for the 2020 mini-season, the Yankees need only outperform their expectation by 5 percentage points to put up a .700 record.

This carries through when we simulate both 2019 and the 2020 season.  I ran 10,000 simulations of each season.  In the 162 game season, there were 985 times when a team won 114 games or more; almost all of them were the Dodgers, Houston or the Yankees.  So we’re actually in a time in which mammoth season is already possible.  But in a 60 game season, there were almost 2400 over 3700 times a team won 42 games or more.  Obviously, the Yankees have the majority of these, but not all. Indeed, the Braves have about 100 seasons out of 10,000 when they win 42 or more. (That said, I still predict 60 wins for the Braves.)

This is not a prediction about this season, since teams are clearly different than they were last year.  But assuming the distribution of team quality is the same, or similar, we’ve got about a 2 ½ 4 times better chance of seeing a .700 season this year than we did last year.  Asterisk, anyone?

Note: I have edited this article as I noticed an error when it was it first posted, based on how seasons of 42 wins (exactly 70 percent of 60) were treated. There are lots of those seasons, which raise the probabilities dramatically.