Sunday, January 2, 2011

Don't Throw Out The Records In This Rivalry

It's something we like to say in any rivalry event, and we certainly hear it every time Beat Hope Week comes around:

"Throw out the records, it's a rivalry game!"

The idea here, I think, is that we don't need to consider the two teams' win-loss records coming into the game because both teams have an equal shot at winning. The boosted emotion and adrenaline level the playing field so that any gap in talent level is significantly reduced.

Is this old adage true? No. At least it's not true in the case of the recent Calvin-Hope Rivalry. If it was the case, we'd expect the team with the better record to still only win 50% of the time. Or, at least, a reasonably close percentage to that.

The two schools have met 31 times over the last ten seasons (since 2001-02). Four of those times Calvin and Hope came into the game with identical winning percentages (Calvin won two, and Hope won two); we'll remove those games from consideration leaving us with 27 contests.

Of those 27 games, the team with the better winning percentage coming into the day won 18 times. That's a 0.667 winning percentage for the "better" team. That's really not at all close to 0.500, especially when we're basically comparing it to 27 coin flips.

If the win probability truly was a 50-50 split, and we had thousands of sets of 27 games, we'd expect this result (the team with the better record winning 18 times or more) only 6% of the time.

So, I guess it's wrong of me to completely dismiss the idea. There is a 6% chance that the 0.667 winning percentage was a product of random variation of a 50-50 probability, but for now, I'm going to consider it 94% false.

No comments:

Post a Comment