This is MLS specific, 1996-2010, using the MLS Library Database provided by scaryice on his blog: http://usasoccer.blogspot.com/2010/11/download-mls-lineup-database-1996-2010.html He counted shootouts as draws, which for these purposes works just fine. To account for the different lengths of seasons, I took 30 as the base number, and multiplied that and Points Per Game, Goals For Per Game, and Goals Against Per Game for all 180 MLS team seasons available. These were rounded to nearest goal. Because of the rounding, we ended up with a goal differential of +3, as opposed to the expected 0. Then, for every point total that existed, we took the average Goals For, Goals Against, and Goal Differental, whether there were 1, 2, or 15 teams that had that point total for a season. Obviously, if there was only one team with a certain number of points, that was also the average. These averages were taken to the 2nd decimal place. From there, we took the point totals and ran that array against our GF, GA, and GD averages to look for a correlation. Unsurprisingly, goal differential had the highest correlation to points at .985. However, it was interesting to see that Goals For had a correlation of .893, while Goals Against was only (-).833. The rudimentary claim then can be made that 5-4 is a "better" scoreline than 1-0. Below is a spreadsheet. Note that I left the "team seasons" on the left, but they are not correctly sorted with the numbers on the right. I only listed the "team season averages" on the first instance that the season point total showed up, then sorted accordingly. Hence, the 30 game season averages go from 66 to 16, then restart at 59 down to 19. That 59 through 19 are all duplicates that are counted for in the team season averages. All in all, not much to see here, but something to mess around with at least. Download: [ame="http://www.zshare.net/download/83346349304d48c1/"]zSHARE - MLS Correlation.xls[/ame]