The 1975 Cincinnati Reds were known as the Big Red Machine, because they wore Red uniforms and ground up opponents like a machine. The 1975 Reds, however, started the season 9 and 10. After sweeping the Dodgers in their first series they lost two straight to the Padres, 5 to 2 and 3 to 2. Salvaging the last game of that series, the Machine headed north to LA, where the Dodgers got even plus one: 52, 31, 76, 54. Returning to Cincinnati the Reds beat the Astros three times but lost the fourth, 76, and then lost one to the Giants, 54. After a couple of wins they lost to the Braves, 54, and then to the Astros, 64. They were 9 and 10.
The 2010 Pirates opened the season not 910 but 1012; scattered among some wins they lost to the Dodgers 102, to the Diamondbacks 91 and 156, to the Giants 93 and 60. They beat the Reds three straight then (43, 53, 54), but then took it on the chin five straight times: 8 to 1, 8 to nothing and 20 to nothing to the Brewers, 43 and 52 to the Astros. They then lost two more 103 and 173, but then won three in a row, reaching 1012.
It is early in the season, and one of the things we all do early in the season is to search relentlessly for signs that our team is better or worse than it appears to be. I am sure you figured out the pattern there; the Big Red Machine started out 910, but they lost those first ten games by a total of 16 runs, 34 to 50. The 2010 Pirates started out 9 and 12, but their first ten losses were by a combined score of 94 to 18, or 76 runs; their first twelve losses were by a total score of 121 to 24, or slightly more than eight runs a game.
Of course we all know that the 1975 Reds (10854) turned out to be quite a bit better than the 2010 Pirates (57105), and it is easy to say now that the earlyseason data reflects the difference; the Reds were losing, but just losing, whereas the Pirates were losing with some style. But is this really a predictive difference? Let us suppose that your hometown nine has about ten losses now, but that they’re not really losing; they’re just a run or two short sometimes. Should you feel better about that? Or does it really matter?
I looked at all teams since 1970, strike years included, and figured:
a) Their runs scored in their first ten wins,
b) Their runs allowed in the first ten wins,
c) Their runs scored in their first ten losses, and
d) Their runs allowed in their first ten losses.
In terms of the margin of defeat, the 2010 Pirates are on one end of the scale, minus 76 runs, and the 1975 Reds are on the other end, 16 runs, actually tied with the 1992 Red Sox, who started out 1110 but (like the 1975 Reds) were just barely losing. In the first ten wins, the scale runs from +66 runs (the 1975 Yankees and the 1980 Brewers) to +13 runs (the 1986 Padres).
Looking first at the wins, we can sort these into levels: +50 runs or more in the first ten wins, +40 to 49, +30 to 39, and +29 or less. There are 78 teams in the data which won their first ten games (ten wins) by a margin of more than 50 runs, outscoring their opponents by an average of 80 to 26 in those games. Those 78 teams finished the season with an average wonlost record of 8376, a .523 winning percentage:



Average Runs Allowed

Average Final Wins

Average Final Losses




Average Runs


Margin

Teams

WPct.

+50 or more

78

80

26

83.2

75.9

.523

+40 to 49

201

71

27

80.8

77.4

.511

+30 to 39

515

63

29

79.5

79.4

.5004

+29 or less

410

53

29

78.1

813

.490

Clearly, the margin of earlyseason victory does appear to be an indicator of the strength of the team. The teams which won by larger margins in their first ten wins went on to have better seasons. Let’s look at the losses:



Average Runs Allowed

Average Final Wins

Average Final Losses




Average Runs


Margin

Teams

WPct.

50 or more

55

27

81

73.8

83.5

.469

40 to 49

235

27

70

77.4

80.8

.489

30 to 39

525

28

62

79.7

79.3

.501

29 or less

389

29

53

81.2

78.4

.511

For some reason all of the deviation in the spread is in the scores of the winning teams, rather than the losing teams, don’t know what to make of that, and for some reason it is more common for teams to be offensively explosive early in the year (78 teams) than for them to be getting blown out regularly (55 teams); don’t know whether that is a real effect or just random data. Anyway, the general conclusion is the same: it clearly does mean something if you’re losing badly, as opposed to just losing. Let’s combine the two into one—margin of victory in your first ten wins, minus margin of defeat in your first ten losses:




Average Runs Allowed

Average Final Wins

Average Final Losses





Average Runs


Combined Margin


Teams

WPct.

+20 or more


65

106

81

84.3

75.5

.527

+10 to 19


165

97

83

81.7

76.7

.516

+2 to 9


318

93

88

80.6

78.4

.507

+1 to 1


120

89

89

79.1

79.3

.4995

2 to 9


290

85

90

79.1

79.9

.498

10 to 19


180

85

99

76.8

83.0

.481

20 or more


66

80

105

72.9

84.2

.464

Some of you are going to be saying, "Well, OK, this is obvious, because we knew anyway that bad teams tend to lose by larger margins than good teams, and that good teams tend to win by larger margins than bad teams, so anyone would assume that this had to be true of the first 10 wins to the same extent that it would be true of any other set of 10 wins or 10 losses."
But I honestly didn’t know how this study would turn out, for two reasons. First, a relationship which is statistically obvious going from A to B is, in many cases, not obvious at all if you look from B to A. In other words, the fact that good teams win by larger margins does not necessarily mean that teams that win by larger margins will tend to be better teams, to an extent that is readily apparent in the data. And second, we’re making an inference from very small data groups, which is always an unreliable process. So. .. I really didn’t know how the study would turn out. But it turns out. . .if you’re not losing by much, take comfort.