Streakingby Jeff Sackmann
June 27, 2007
There's nothing inherently good or bad about a team being streaky. Ten wins and 10 losses count the same in the standings whether they come in bunches of 10 in a row and 10 in a row or neatly alternated over a 20-game stretch. That isn't to say that the former wouldn't cause a few more ulcers among the front office and the faithful, but team streakiness isn't usually the sort of thing that gets much attention from analysts.
And that's probably how it should be. That doesn't stop the topic from being interesting, though. (Especially when, as you'll see over my next few articles, it's overlaid with other team characteristics.) Like most things that happen on a baseball field, we are inclined to explain streakiness to external factors, such as a team's youth, reliance on an inconsistent slugger, or an uneven bullpen. Again like many things on a baseball field, much of what we try to explain is really just luck; send a team out there for 162 games, and even the steadiest 81-win squad isn't going to win all the even-numbered games and lose all the odd ones.
For now, I'm not going to do much to explain the phenomenon, I just want to get a grip on what we're working with. If we're going to discuss team streakiness, we need some way to measure it.
In their simplest form, streaks are just a matter of stringing together wins or losses. A team that bunches its wins (and by extension, their losses) is considered streaky, while one that doesn't conspicuously do so is not. Quantifying that should be pretty simple.
Take, for instance, last year's Chicago White Sox. They were 90-72, 18 games above .500. If you pick your way through their game log for the season (or, better yet, write a script to do the work for you), you'll find that on 82 occasions, they won a game they day after a win, or lost a game a day after a win. In other words, about half the time they were in the middle of a streak.
It's not quite that simple, though. If a team is 18 games better than .500, of course they should be a little bit streaky. There's no way to win 90 games in a 162-game season without stringing some wins together here and there. Some of the streakiness we measured above is really just an indication that the team was good. (If we were looking at a 72-90 team, we'd see the same phenomenon, only as a reflection of the team's ineptitude.)
Thus, setting aside the 18 times that the White Sox had to have a win following a win, they were in the middle of a streak on 64 occasions (82 minus 18) out of a possible 144 (162 minus 18). As it turns out, the resulting rate of 44.4% is almost exactly the median among MLB teams in the expansion era. In not-quite-layman's terms, that means that, setting aside the streaks that can be attributed to the team's likelihood of winning, there was a four-in-nine chance that on any given day the Sox would be amidst a streak.
The first obligation of anyone devising a new measuring tool is to share the resulting best and worst. Here are the streakiest teams of the expansion era:
YR TM W L STRK 2002 TOR 78 84 59.6% 1996 COL 83 79 58.9% 1969 HOU 81 81 58.6% 1996 CIN 81 81 58.6% 1981 DET 60 49 58.2% 1994 OAK 51 63 57.8% 1999 BAL 78 84 57.7% 1998 BAL 79 83 57.6% 1966 CIN 76 84 57.2% 1961 CHA 86 77 57.1% 1966 ATL 86 77 57.1% 1986 CLE 85 78 57.1% 1975 SFN 80 81 56.9% 1991 MON 71 90 56.3% 1991 SEA 83 79 56.3%And the least streaky:
YR TM W L STRK 2005 SLN 100 62 20.2% 1995 CLE 100 44 23.9% 1991 CLE 57 105 25.4% 1962 NYN 40 121 26.3% 1985 PIT 57 104 27.2% 1965 MIN 102 60 27.5% 1977 MIL 67 95 27.6% 1966 CHN 59 103 28.0% 1993 ATL 104 58 28.4% 1988 BAL 54 107 28.7% 1967 KC1 62 99 29.0% 1968 CAL 67 95 29.1% 1970 CIN 102 60 29.2% 1976 CIN 102 60 29.2% 1989 OAK 99 63 29.4%One thing pops out looking at those lists. It appears that I might not have correctly handled team quality in determining streakiness. Most of the very consistent teams are 100-game winners or losers (or close), while most of the streakiest are close to .500. Why is that?
There are, I think, two factors at play. First, there is some overcorrection. When you're dealing with a team that is 50 (or, in the case of the 1962 Mets, 82) games above or below .500, streakiness just can't be separated all that effectively from team quality.
Second, extremely good or bad play may look like streakiness, but it doesn't leave a lot of room for streaks that don't fall into the season's pattern. If those '62 Mets had been more even in distributing their wins and losses, they would've won at least one of their first nine games, but probably no more than two. Extremely poor play looks a lot like streakiness, while in cases like these, it's probably isn't.
That said, there are plenty of cases where very good teams were very streaky:
YR TM W L STRK 1977 PIT 96 66 53.0% 1978 LAN 95 67 50.7% 1973 LAN 95 67 49.3% 1985 NYA 97 64 48.4% 1990 PIT 95 67 47.8% 1993 TOR 95 67 47.8% 2006 MIN 96 66 47.0% 1992 ATL 98 64 46.9% 2005 CHA 99 63 46.8% 2006 DET 95 67 46.3%In addition to those teams (chosen from among the 142 95-game winners in the last 45 years), you may have noticed the 90-game winner and 90-game loser in the "streakiest" list above. It's rather difficult for a mediocre team to be consistent, but a very good or bad team can be quite inconsistent, at least once we take their demonstrated skill level into consideration.
Oddly enough, all of the teams in the last list were streakier than the 2002 Oakland A's, the club that strung together 20 consecutive victories. Tough to believe, though it begins to make sense when you recognize that the A's never lost more than four in a row and did that only three times. With losses somewhat evenly distributed throughout the season (except for that monster run), it stands to reason that the A's wouldn't go down as one of the streakiest teams in baseball history.
Jeff Sackmann is the creator of MinorLeagueSplits.com. With Kent Bonham, he founded CollegeSplits.com. Jeff and Kent blog about college baseball and the draft, and you can follow them on Twitter for bite-sized snacks of minor league and college stats. Jeff also has an email address.