There’s been speculation here at several points about whether the Braves 2018 edition is streaky.  The main reason we might want an answer to that question is the supposition that a streaky team will have trouble when they get into the playoffs unless they get on a long hot streak, because a single cold streak probably eliminates you.

But as a practicing statistician for 35 years, one thing I’ve learned (and it’s not just me… there’s an academic literature here) human beings totally suck at discerning randomness because we’re so good at discovering patterns.  We’re so good at it that we find loads of patterns that aren’t really there.  So why do people think the Braves are streaky?  They look at examples: four wins in a row from June 29th-July 2nd followed by four straight losses from July 3th-6th.   (Indeed, that particular 8 game stretch led to a lot of the speculation, as did July 24th-August 3rd: 4 losses followed by 5 wins.)

But every team has four game win streaks and four game loss streaks.  How do you know that the Braves streaks are unusually high or low?

I’m here to tell you the Braves were no more streaky than you expect an 88-68 team to be.  I have created a very conservative metric of streakiness (conservative in the sense that one ought to expect teams to be somewhat streakier than this metric) and the Braves 2018 season sits right about where you’d expect them to be.

Here’s what I did.  I took 88 wins and 68 losses and scrambled them randomly.  I then looked at the streaks that developed from the scrambled season.  I then did this 10,000 times and looked at the average streak table across all these seasons.  Here’s what I got, compared to what has actually transpired:

Streak Actual Expected
<-5 0 0.5
-5 1 0.8
-4 4 1.8
-3 5 4.2
-2 7 9.6
-1 23 22.0
1 17 17.1
2 11 9.6
3 5 5.4
4 2 3.0
5 3 1.7
6 1 1.0
>7 0 1.1

An average 88-68 team has 22 one loss streaks and 17.1 one win streaks.  The typical team has an expectation of just over 1 seven game or longer win streak and will lose five or more in a row about every other year.

Look how close the actual values are to the expected values.  The 2018 Braves team looks completely normal for a team that is 88-68.  They do have about 2 more four game losing streaks than you’d expect;  but that is made up for by no 5 game losing streaks.  Similarly, they have an extra 5 game winning streak, but they are short the 7 game winning streak a team this good usually has.  We can run fancy statistical tests which demonstrate that there is nothing particular streaky about the Braves’ 2018 season.  The reason this metric is conservative is that we would expect extra streaks since games are played in series.  If you play a better team than you three or four games in a row, or worse than you three or four games in a row, your chances of getting swept or sweeping ought to be higher than the scramble method I used.

If you want to see a streaky team, go back to the famous 1982 team:

Streak Actual Expected
-11 1 0.0055
-10 0 0.01
-9 0 0.03
-8 0 0.07
-7 0 0.17
-6 0 0.38
-5 1 0.88
-4 5 2
-3 2 4.5
-2 4 10.1
-1 23 22.5
1 18 18.5
2 4 10.1
3 7 5.5
4 3 3
5 1 1.6
6 2 0.89
7 0 0.47
8 0 0.25
9 0 0.14
10 0 0.07
11 0 0.03
12 0 0.02
13 1 0.0082

 

This team combined a 13 game winning streak (something that happens once every 122 years for an 89-73 team) an 11 game losing streak (once every 182 years) as well as three extra 4 game losing streaks and extra three- and six-game losing streaks.  To make up for these extra streaks, they had dramatically fewer two-game streaks, both winning and losing, than expected.  Note by the way, this team had a three game losing streak that bounced them from the playoffs, although there was a rainout that erased a Braves lead in the first game, which convinced me, up until 1995, that God was not a Braves fan.

So the Braves may do well in the playoffs, or not, but it won’t be because they are an unusually streaky team.  They aren’t.