Who's Really Ahead?

The winding down of the current baseball season seems an appropriate time to take a look at a curious inconsistency that sometimes appears in team standings. Teams are generally ranked on the basis of both winning percentage and "games behind" the leader. Here's a recent example:

EAST	W	L	Pct.	GB
Baltimore	92	59	.609	--
New York	88	63	.583	4
Detroit	74	77	.490	18
Boston	74	78	.487	18 1/2
Toronto	71	80	.470	21

There's no inconsistency here between the ranking according to percentage and the ranking according to games behind, and that's typical. Once in a while, however, the team with the higher winning percentage may be at least one-half a game behind in the standings. That anomaly intrigued Peter A Rogerson, a geographer at the State University of New York at Buffalo, and he went to the trouble of analyzing the underlying mathematics and identified the conditions necessary for such an occurrence. His results appear in the latest issue of the Journal of Recreational Mathematics. In formal terms, Team 1 is said to be "k games behind" Team 2 when one-half the difference between the number of wins by each team plus one-half the difference in the number of losses by each team is equal to k: [(w2 - w1) + (l1 - l2)]/2. In effect, the "games behindÓ statistic represents the number of games that Team 1 would have had to win, instead of lost, in order to be tied with Team 2. Rogerson provides a real-life example of the type of inconsistency in rankings that can arise (Triple A baseball standings in the American Association on May 14, 1991):

Team	W	L	Pct.	GB
Buffalo	14	9	.609	1/2
Indianapolis	18	12	.600	--
Nashville	14	14	.500	2 1/2
Louisville	12	20	.375	6 1/2

He concludes that such an inconsistency is most likely to occur when one team has played substantially fewer games than the other. "In the case of baseball, this can often happen early in the season, when one team has had many games postponed by adverse weather conditions,Ó Rogerson says. Usually the cause is rain, but in Buffalo, that may sometimes be snow, he adds. The inconsistency can also come up later in the season, especially when two teams have relatively high winning percentages. A table showing the winning percentage of Team 2, p, and how many more games Team 2 must have played than Team 1for the inconsistency to arise, c, illustrates why.

Winning percentage, p	Critical number of games, c
.510	51
.530	17
.550	11
.570	9
.590	7
.610	7
.630	5
.650	5
.670	3

"Having two teams with extremely high winning percentages is also more common early in the season,Ó Rogerson notes. He adds that if Team 2 has a winning percentage of exactly .500, the inconsistency can't arise because Team 1 could only be at least half a game behind Team 2 if its winning percentage was less than .500. Similarly, the inconsistency cannot come up when Team 1 has a winning percentage of exactly .500. Of course, baseball isn't the only sport in which teams are ranked according to winning percentage and games behind. The same anomaly could occur in football, basketball, and hockey team standings. In which sport are they most likely to come up?

References:

Rogerson, Peter A. 1997. Inconsistencies in league standings. Journal of Recreational Mathematics 28(No. 2):81-84.