By Roger Weber
In baseball, a good portion of outcomes are more or less random.
According to statistical rules or randomness and probability, an average team playing a 162-game season might win as few as
68 or as many as 94 games just because of chance. It's most likely, of course, that they'll win somewhere in the vicinity
of 81 games, but nevertheless chance can heavily affect a team's season outcome, especially in a sport where so few games
separate good teams from bad teams. After all, the difference between the best and worst teams is usually only around 20 percentage
It would be an historic and arguably wasteful use of time to look
in depth at all teams ever to play Major League Baseball, even though that might be necessary to examine the actual talent
of every team ever to play. But to determine the best and worst teams there are easier ways to ensure luck doesn't skew the
results too much and show a good team as bad or a bad team as good.
The goal is to figure out which teams were especially good and
which were especially bad. So it makes sense that we want to include teams that without question fall outside the range that
a simply average team could attain. The way this is achieved is to go several standard deviations away from the mean, essentially
including only totals that we know are good teams, ones that finish outside of a confidence interval centered at an impressive
record. The probability that an average team will fall outside of this range (55-107 wins in a 162-game season) is .00006.
Excluding Teams due
to unusual Circumstances
1) Teams from 1943-45. World War II rocked baseball. Many players
went into combat and all were affected in a major way. The easiest and most fair way of dealing with this era is to eliminate
2) Teams pre-1900. Although there were certainly great teams,
it is not reasonable to compare them to modern teams.
3) Teams pre-1920. These teams actually are included in the
study, although a separate category is made excluding them. Like the pre-1900 teams they are difficult to compare to modern
Grappling with time
I have experimented with era adjustments a good deal. I have pondered
long and hard as to whether to adjust for competition levels or not and have decided against it. It is not a fair judgment
to penalize certain teams for playing in an era we cannot actually honestly compare to another – there are statistical
ways of attempting but none has ever truly satisfied me. To say that one field of competition was better than another is to
state a valid point. Modern baseball probably employs a higher level of talent than in the distant past. Still, that doesn't
do us too much good in this study. Regardless of the actual talent of the players, a team that finishes well above its competition
is a great team. It is not fair to penalize a team for not having players that were not available at the time.
In fact it is true that several major league teams today might
beat the great teams of the early 1900s, but the fact that they have more talent is only a function of the time in which they
play. To me, a team's greatness is defined more by how it dominates its competition and plays in the situation it is in than
by how it might play in another era. John McGraw, if he managed the 1905 Giants today, certainly would not have played with
all the players on his team at the time. He would have at his disposal players of all races. It is not fair to punish his
team in 1905 for not including players that were not allowed to play.
In many ways this comparison is like the comparison between the
best teams. Most obviously it is important to recognize that to be a truly terrible team losses are the most important ingredient.
A bad team must lose many, many games. The 2003 Detroit Tigers, for example, were terrible. But they were good enough to win
a few games at the end of the season to avoid setting the all time losses record. Wins and losses are indisputably the number
one measure of a team's futility.
Second, we expect a truly terrible team to have lost by tremendous
margins. It can be argued that the worst team might win below its ratio of runs scored and runs allowed by losing even winnable
games. But then we must think what we are measuring in the study. I would think that it makes more sense to measure which
team actually "was" the worst team in the sense of being the worst allotment of players with the least ability to even play
the game instead of measuring which team "played" the worst regardless of ability. So in that case, it is more important for
a team to have lost games by a tremendous number of runs. But it does not matter whether that team is offense-powered or defense-powered.
It should ideally have bad hitting, bad pitching and bad fielding. So not only should it lose by a great many runs, but its
ratio of runs scored to runs allowed should be horribly low.
We also want to measure complete and total futility. To be a truly
terrible team it must express uniform and absolute terribleness. It should prove to fans that it was far below even its lowest
competition. Like a best team, it should separate itself from the pack and be undisputedly in the cellar of the standings.
Therefore any opponent against which it has a winning record is in some way, or at least performed, worse than this team.
Therefore a component of the measurement should be the uniformity of its futility.