June 7, 2016
Do Pathetic, Embarrassing, Miserable Failures Breed Success?
As you may have heard, the Chicago Cubs started the season pretty well. As of the end of May, they were 35-15, playing exactly .700 ball. That projects to a 113-49 record over 162 games, which would be the most wins in a season since the Mariners won 116 in 2001, and the most in the National League since, well, since the Chicago Cubs won 116 in 1906.
But that wasn’t the only notable end-of-May record. The Twins and Braves were both 15-36, on pace for 48-114. The Reds, at 17-35, were on pace for 53-109. The 20-33 Padres projected to 61-101, raising the question of how Padres owner Ron Fowler would describe the Twins, Braves, or Reds. On the other hand, the 32-20 Red Sox were on pace to finish 100-62, the 33-21 Giants were on track for 99-63, the 32-21 Nationals on pace for 98-64, the 31-21 Rangers for 97-65, and the 30-21 Mariners for 95-67. So there were, at the end of May, four teams with a shot at 100 losses and six that could win 100.
As an aside, I am fully cognizant that “on pace for” is intellectually lazy and ignorant, unless it’s wielded cleverly by the likes of Jayson Stark or Cespedes Family Barbecue. (Especially the Cespedes Family Barbecue link. You should check it out. Go ahead, it won’t take long. I’ll still be here.) So no, I’m not implying that there actually will be four teams with 62 or fewer wins and six with 62 or fewer losses. I’m just setting the tone. Play along with me here.
One of things I’ve heard recently, in reference to the Cubs, is that they’re a great team, but they’re helped by the presence of some really lousy clubs. I heard one radio broadcaster opine that when there are really bad teams, it creates an opening for good teams to put up remarkable records. Or something like that. The general idea was, great records are more likely when there are teams with terrible records.
This makes sense on a couple levels. First, when there’s a high degree of parity, like in the American League last year, there are, by definition, neither truly outstanding nor truly execrable teams. Second, a great team like the Cubs can fatten its record by beating the stuffing out of the punching bags in Atlanta, Cincinnati, and San Diego.
There are a couple problems with this line of thinking, of course. The biggest of them is that since the advent of divisional play in 1969, teams don’t play a lot of games against any one team. Yes, the Cubs are 6-1 so far against the Reds (outscoring them by 40 runs), but they’ll play them only 12 more times this season. By contrast, the 110-44 1927 Murderers’ Row Yankees, arguably the most famous good team of time, played the 51-103 Red Sox and the 59-94 Browns 22 times each. (They went 39-5 against them!) Second, of course, there’s no guarantee that a good team will demolish a bad one every year. Minuscule sample size warnings and all, but the hapless Braves have a .500 or better record against only three clubs: 5-1 against the Marlins, 3-3 against the Phillies, and 1-1 against…the Cubs.
Still, this is worth researching, I figured. Do teams with really bad records make it more likely that there will be teams with really good records? And, to extend this syllogism (Really bad teams create an opportunity for really good teams, the xxxx season had one or more really bad teams, therefore that year had really good teams) into other terms I learned in Introduction to Logic in college, are the converse (Do teams with really good records make it more likely that there’ll be teams with really bad records?) and/or inverse (If there aren’t teams with really bad records, is it less likely that there’ll be teams with really good records?) true?
For starters, I had to define “really good” and “really bad.” I went with basic statistics. The average team, obviously, goes 81-81 (77-77 in the 154-game era). Two standard deviations encompass about 95 percent of observations, so a record that’s two standard deviations from the mean is in the top (or bottom) 2.5 percent of team records. That should theoretically occur three times in a four-year span in the current 30-team configuration. When there were 16 teams, it should occur about twice every five years. So I compiled standard deviations of wins for every 162- and 154-game season. The 154-game season began in 1904 and continued through 1960 in the American League and 1961 in the National League, excluding the World War I-shortened years of 1918 and 1919. The 162-game season’s prevailed since, except for the strike-shortened 1981, 1994, and 1995 seasons.
The standard deviation of victories in the 162-game era is 11.4, so the two standard deviation range is 58 to 104. I counted any team with 58 or fewer wins as really bad, and any team with 104 or more wins as really good. In the 154-game era, the standard deviation was 14.8 wins, so really bad teams won 47 or fewer games and really good teams won 107 or more. By these definitions, there were 57 really bad teams in the 107 seasons from 1904 to 2015 (excluding the five shortened seasons noted above), occurring in 45 seasons. There were considerably fewer really good teams, 20, falling in 16 seasons.
How often did really bad and really good teams coincide? Surprisingly rarely, it turns out. The 2013 Astros were really bad (51-111), but there were no really good teams that year; the best in the American League was the 97-65 Red Sox. The 1986 Mets were really good, winning 108 games, but the worst team that year was the 64-98 Pirates. In 1975, there was a really good team, the 108-54 Big Red Machine, and a really bad one, the 57-102 Tigers, but they didn’t play in the same league.
Overall, there were only seven seasons that featured both a really bad team and really good one in the same league. I’ve listed each team’s record as well as what I’m calling its “normalized” record. I calculated a normalized record to remove the effect of the really bad team on the really good team’s record, and vice-versa. Take, for example, the 1970 Baltimore Orioles. They were really good, going 108-54, playing in the same league (though not the same division) as the really bad Chicago White Sox, against whom the O’s were 9-3. Excluding their games with the White Sox, they were 99-51. That’s a .660 winning percentage. Apply that to their 12 White Sox games, and they could be expected to go 8-4 (OK, 7.92-2.08). Add an 8-4 record to 99-51 and you get 107-55, still well within our definition of really good.
Here are the seven instances of plus- and minus-two standard deviation teams in one league:
· 2004 National League: The best and worst teams were the 105-57 Cardinals and 51-111 Diamondbacks, respectively. However, St. Louis played only six games against Arizona, going 5-1. Normalized records: St. Louis 104-58, Arizona 52-110.
· 1998 National League: There were two expansion teams in 1998, the Diamondbacks and the Devil Rays, but both avoided 100 losses. The worst team was the 54-108 let’s-blow-up-the-World-Championship-team Marlins, in the same division as the 106-56 Braves. But the Braves went only 7-5 against Florida. Normalized records: Atlanta 107-55, Florida 53-109.
· 1988 American League: The best team in the two-division AL was the 104-58 A’s and the worst was the 54-107 Orioles. This was back in the days of the sort-of balanced schedule (13 games against teams in the same division, 12 against teams in the other division). Oakland was 8-4 against Baltimore. That’s good, but the A’s had a better record against five of the remaining 12 teams in the league. Normalized records: Oakland 104-58, Baltimore 54-107.
· 1970 American League: The Orioles were 108-54, the White Sox 56-106. Baltimore was 9-3 against Chicago. Normalized records: Baltimore 107-55, Chicago 57-105.
· 1963 American League: Prior to 1969, there were no divisions. The Yankees were the league’s best team, 104-57, and the Senators, a 1961 expansion team, were 56-106. New York had a 14-4 record against Washington, the most wins it had against any American League team. Normalized records: New York 101-60, Washington 59-103.
· 1932 American League: The Yankees won the league by 13 games with a 107-47 record. The Red Sox were 64 games behind at 43-111. The Yankees took their season series against Boston, 17-5 (all those ESPN Sunday Night Baseball games that year must’ve been duds), but had the same record against the seventh-place White Sox and the fifth-place Tigers. Normalized records: New York 105-49, Boston 44-110.
· 1909 National League: The 110-42 Pirates finished 65.5 games ahead of the 45-108 Braves, and Boston was Pittsburgh’s easiest opponent, as the Bucs went an absurd 20-1 against the Braves. Normalized records: Pittsburgh 104-48, Boston 51-102.
The syllogism doesn’t really seem to hold, does it? There’ve been 57 really bad teams, defined as finishing with two standard deviations fewer wins than .500. They coincided with only seven really good teams. And by looking at normalized records, there’s a good argument that four of those teams (the four most recent ones) would’ve been really good without the ineptitude of the really bad team. So the 57 really bad teams coincided with only seven really good teams, four of which would’ve probably been really good even if the really bad teams hadn’t been really bad. That’s not convincing.
How about the converse: Do really good teams create an environment for really bad teams? There’s a little more evidence for this, given the relative paucity of really good teams. There’ve been only 20 two standard deviation outliers to the positive side in baseball history, and of them, seven occurred the same year as a really bad team. However, using the normalized records above, it appears that five of those sevenreally bad teams would still have a really bad record had they played the really good teams as they played the rest of their schedule (all except the 1963 Senators and the 1909 Braves).
And the inverse: If there isn’t a really bad team, is harder for there to be a really good team? Well, there’ve been 13 teams (2001 Mariners, 1998 Yankees, 1993 Braves, 1986 Mets, 1984 Tigers, 1975 Reds, 1969 Orioles, 1961 Yankees, 1954 Indians, 1931 Athletics, 1927 Yankees, and, yes, the 1906-07 Cubs) whose win totals were over two standard deviations above the mean without the benefit of a really bad team to beat up. In fact, of the eight teams with best winning percentages in history, only one, the 1909 Pirates, played in the same season as a really bad team.
Those eight teams—the 1906 Cubs, 1909 Pirates, 1954 Indians, 2001 Mariners, 1927 Yankees, 1907 Cubs/1931 Athletics, and 1998 Yankees, in descending winning percentage order—are the only teams to finish a season with a winning percentage greater than .703. As play began on June 6, the Cubs were winning at a .709 clip, a pace exceeded over a full season only five times. (And then they won Monday.) They could wind up being one of the winningest teams in baseball history. But if they do, I think this research shows, it’s because they’re really good, not because the Braves, Reds and Padres are really bad.