A couple of days ago, our friend Rob Neyer asked an interesting question over at Just A Bit Outside. The Kansas City Royals had started out the 2015 season with seven consecutive wins. I guess maybe that whole “Defending American League Champion” thing actually means something. Despite having come oh so close in the 2014 World Series, not many really fancied the Royals as contenders this year. Should the Royals’ seven-game perfect start change our expectations of them?
Rob quoted BP alumnus Joe Sheehan, who pointed out that last year there were 11 teams who had seven-game winning streaks and only four made the postseason. The Royals winning their first seven makes for a nice story at the beginning of the year, and certainly it’s a positive thing for the Royals, but it tells us no more than a seven-game winning streak in mid-June would tell us. Later on the JABO podcast, former pitcher (and JABO contributor) C.J. Nitkowski said that he could see a case for a streak to start the season being somewhat more valuable. It gives the players confidence going forward, he suggested.
Rob left it with the question of whether either one of them was right, “Mathematically speaking, I mean.”
Warning! Gory Mathematical Details Ahead!
There are 156 possible seven-game streaks that can happen over the course of a 162 game season. There’s Games 1-7, 2-8, 3-9, etc., all the way to game 156-162. For all teams from 1996-2014, I found their record in each of those seven game stretches. I also looked at a team’s overall record for the season. I looked to see how well a team’s record in Games 1-7 correlated with their eventual end-of-season record. Then, 2-8, 3-9, etc.
Now, of course, there’s going to be some inter-correlation there. The Royals, even if they lose every game for the rest of the year, are guaranteed a 7-155 record. But, since the span of games in each instance is seven wide, no individual bucket of games has an unfair advantage.
The correlations ranged from .455 at the high end (games 141-147) to .272 at the low end (games 38-44). So, seven games isn’t a fantastic predictor of anything, but some interesting patterns emerged. Take a look at this graph. Moving from left to right, we have rolling seven game blocks (games 1-7, 2-8, 3-9…) and on the y-axis, we have the correlation coefficient for that set of games.
We see that, in general, the graph starts out early in the season at a correlation of .40, but then the trend line points downward, bottoming out around games 64-70. But then something happens. As June turns into July and July into August, those seven game stretches are actually better correlated with the end-of-season, with a peak around games 107-113. There’s a tiny valley downward, and then a rebound around games 141-147.
So, are the first seven games a particularly significant predictor of what’s to come? Well, mathematically speaking, the first few games do seem to have some special magic compared to the rest of April and May. But if you really want to see what a team is made of, wait until you see what they do around Game 100… roughly the trading deadline. Maybe that’s a matter of teams showing their true colors in the dog days of summer. Maybe it’s the fact that around then, the good teams have added extra pieces and the bad teams have jettisoned those pieces. But there’s something about that point. Around Game 144 takes us into the last couple weeks of the season. Official pennant race season.
So, yeah, a hot seven-game start is good and it does mean a little more than a hot seven-game stretch in late May, inasmuch as a seven-game sample can tell you something interesting about a team. But if you really want to know a team’s true colors, look at what it does when the weather turns a bit hotter. Then again, maybe we’re reading too much into a correlation of .40, which means that there’s a lot of room for error. It is only seven games out of a very long season after all.
Thank you for reading
This is a free article. If you enjoyed it, consider subscribing to Baseball Prospectus. Subscriptions support ongoing public baseball research and analysis in an increasingly proprietary environment.Subscribe now