I tidied up a spreadsheet today with High/Low/Close game data going back to 1993. When I was done I figured that I would take a look at how accurate WTM has been in the long term. I ran WTM numbers for each team over the past 240 games (less for expansion teams) and the results were better than I expected.

Take, for example, the Atlanta Falcons. They’ve played 240 games since 1993 and have won 104 of them. Their WTM estimate over the same period of time was for 105.21 wins, a mere 1.21 win difference. Miami won 125 games and WTM estimated 125.14. The team with the biggest difference between actual and estimated wins was Green Bay. The Packers won 152 of their 240 games and WTM estimated that they should have won 141.16 during the same span, a difference of 10.84 wins.

The correlation between actual wins and WTM estimated wins turned out to be an astounding 0.984. The R-Squared for the same stats was 0.96. I’m curious to know how other methods have fared during the same period. If I get a chance I’ll run Pythag numbers…

Here is a chart showing each team, their actual wins, and their estimated wins during the period from 1993-2007.

Just about nailed Seattle and San Francisco on the head too. Not bad for a system that only takes into account three things: High, Low, and Close. Points. Nothing more.