A Small Bracket Tip
Well, the best time of year is upon us, with March Madness ready to tip off in fewer than 24 hours. If you haven't yet filled out your brackets, here's some statistical data that could give a small edge.
The tie-breaker in most pools is the the person who most closely predicted the combined score of the championship game. While ties are relatively unlikely in a small pool, it's still worth thinking about likely combined scores so that you can make as good of a prediction as possible.
I looked at the combined scores from the last 21 NCAA championship games (1985-2005). Scores ranged from an outlier-low 116 (2002) to 176 (1990).
The mean for the period was 149; the median was 148.
However, two things bothered me about that data. First, it covers a broad period during which the NCAA game has undergone many changes. Second, that ugly 2002 game between Maryland and Indiana is a completely statistical anomaly.
Narrowing the data to games played in the era when early draft entry changed the complexion of the college game (1997-2005), and throwing out the 2002 game, we get eight games with a score range from 145-165. Not surprisingly, the average is just under 155.
This seems like a good figure, but testing it against the data (in other words, how far off were you each year) shows that you get a high level of variance. Consequently, I think it needs to be adjusted down slightly.
In the end, 151 seems like a good number to me. It's fairly close to many of the games played in the mid- to high-140s, but isn't too far below the games in the upper 150s.
So I'll be putting 151 for my combined score in any pools I enter. If you want to use this number, feel free, unless you're betting against me.
The tie-breaker in most pools is the the person who most closely predicted the combined score of the championship game. While ties are relatively unlikely in a small pool, it's still worth thinking about likely combined scores so that you can make as good of a prediction as possible.
I looked at the combined scores from the last 21 NCAA championship games (1985-2005). Scores ranged from an outlier-low 116 (2002) to 176 (1990).
The mean for the period was 149; the median was 148.
However, two things bothered me about that data. First, it covers a broad period during which the NCAA game has undergone many changes. Second, that ugly 2002 game between Maryland and Indiana is a completely statistical anomaly.
Narrowing the data to games played in the era when early draft entry changed the complexion of the college game (1997-2005), and throwing out the 2002 game, we get eight games with a score range from 145-165. Not surprisingly, the average is just under 155.
This seems like a good figure, but testing it against the data (in other words, how far off were you each year) shows that you get a high level of variance. Consequently, I think it needs to be adjusted down slightly.
In the end, 151 seems like a good number to me. It's fairly close to many of the games played in the mid- to high-140s, but isn't too far below the games in the upper 150s.
So I'll be putting 151 for my combined score in any pools I enter. If you want to use this number, feel free, unless you're betting against me.
posted by N at 12:43 PM
0 Comments:
Post a Comment
<< Home