Writing about baseball this early in the season is difficult. It is a sport that involves a lot of random variability, which is hard enough to make sense of in 162 game samples, let alone 10 game samples. All kinds of things can happen in 10 games, yet every year after the first couple of weeks of the season there is a rush to crown the latest breakout team, player, coach, or training method. This season the focus of our attention has been the Kansas City Royals and their 7-0 start. Articles abound with reminders of all the things this team learned from their postseason run last year, how the offseason additions of key veterans has brought valuable
presents presence, and even how this is the hot start that is meaningful. Are they a pretty good team? Sure. FanGraphs has them projected for 82 wins. We may want to settle down on the positive reaction to these 7 wins.
In his newsletter (subscribe!) this week, Joe Sheehan pointed out that over the last couple of seasons about half the league has had a win streak of at least 7 games, with many of the teams ending up quite unimpressive. Looking to last season 11 teams had a win streak of at least 7 games and only 4 made the playoffs. A 7-game win streak does not tell us much about a team that we didn't already know. Teams, good and bad, can have great weeks. Russell Carleton picked up on this thread in his article on Thursday showing that a team's performance in the first 7 games of the season is not the most predictive of the team's final record. The first 7 games are a better predictor than any 7-game slice in late April and May but even as such account for only 16% of the variance in end of season record.
The Royals, and other hot-starting teams, were a topic of discussion for Rob Neyer and C.J. Nitkowski on Tuesday's Just A Bit Outside (JABO) podcast. Neyer and Nitkowski had a great conversation about the start some teams have jumped out to this season. In discussing Sheehan's work that I mentioned above, one comment/question from Rob Neyer piqued my attention:
" To win 90+ games you need to put together a few streaks over the course of a season. Right? It's just almost impossible to do it if you don't put those streaks together."
This is an interesting question. Can you win 90+ games without having a reasonably long win streak? I am sure you can, but what are the odds of it happening? I decided to check it out.
Using the Baseball Reference Play Index Streak finder I gathered all streaks of 5, 6, 7, 8, 9, and 10 games from the last five seasons (2010-2014). Rather arbitrarily I am deciding that at least a 5-game win streak is reasonably long. For now I am just looking at whether a team had a win streak of the given length. They may have had multiple instances of the requisite streaks, but in the work here I am just coding it as having happened or not. The number of streaks is another question worth investigating, but I will leave it for now. With the streak data in hand I then looked at the end of season win totals for the teams that had these streaks. I categorized teams into two bins: (1) 85 or more wins, and (2) 84 or fewer wins. Neyer's quote above has him asking about 90+ win seasons, but later in the conversation he mentioned 85 wins as a benchmark, so I went with the lower total as the boundary for this investigation.
The data used include the 150 team seasons from the last five years. In that time there have been 66 teams with 85+ wins in a season. Here are the data for the 5-game win streaks:
Well done, Mr. Neyer. In each of the 66 seasons in which a team won 85+ games, the team had at least one 5-game win streak. This does not mean that a 5-game win streak will guarantee an 85+ win season, as you can see in the table there were 69 teams that had a 5-game win streak but did not break 85 wins on the season. Nevertheless, the odds of an 85+ win season are greater than an 84 or less win season if the team has a 5-game win streak during the season.
I started this article by talking about the Royals' 7-game win streak. So it is probably worth showing the data for 7-game win streaks. Here they are:
To some extent this runs counter to what Neyer had suggested must be the case in terms of high win total teams needing win streaks during the season. Although, that depends on an interpretation of how long the streak(s) needs to be. While it has been true for 5-game streaks, it is has not been so for 7-game streaks. Twenty-four of the teams that had at least 85 wins on the season did not manage a single 7-game win streak during the season. So it appears that it is not impossible to win a lot of games without a solid win streak involved. Also consider that almost the same number of teams (29) that did have a 7-game win streak did not break the 84-win plateau. With all that said, the odds of an 85+ win season are much greater than an 84- win season if a 7-game win streak happens. That should keep Royals fans confident in the outlook of their team's season.
It may seem conspicuous that I skipped over the 6-game streak data. Well, here they are:
Again we have data that runs somewhat counter to Neyer's suggestion. Seven teams won 85+ games in a season without so much as one 6-game win streak. However, as has already been demonstrated they did win 5 in a row at one point (also true of those 24 non-7-game streak teams), so at this point we are quibbling over one or two games in a sequence. Regardless, it is interesting to note.
Rather than carry on showing the data from all of the other streak lengths I mentioned looking into, I will just say that that, unsurprisingly, as the length of the streak increases so too do the odds of winning 85 or more games that season. However it is not perfect. About half of the 85+ win teams (36) had at least one 8-game win streak, but that means 30 teams won 85+ games without an 8-gamer. So long streaks are not necessarily needed to win 85 or more games. Some teams have long streaks and fail to reach 85 wins. Even a 10-game win streak during the season does not always produce an 85+ win season. Just ask the 2013 Blue Jays or 2010 Rockies.
None of what I have presented here was done so in an effort to make Neyer's comment look silly. He raised an interesting idea on a podcast that I wanted to investigate, and as it turns out he was for the most part correct. Teams that win a lot of games in a season have tended to go through at least one decently long win streak (and in many cases multiple win streaks). But doing so is not a guarantee of season success. As noted here, and many other places, it is very difficult to identify the quality of a team over the course of any small sample of games. If the Royals keep up their torrid win pace (it should be noted that they lost their next two games after the 7-win streak) over the rest of the season, then we will need to consider if something was overlooked in the projections. For now, we should just consider the start they have had as banked wins and treat them like any set of games in the middle of June. There is still a lot of baseball to be played.
. . .