Basketball Statistics Blog

Statistician's Digest: Matchup Probabilities ~ Applicability & Formula

Raymond Cheong designed a computer ranking system (CRS) - see it at
www bme.jhu.edu/ - to predict the outcome of a hypothetical matchup between two basketball teams. The CRS relies on the observation that average score margin is highly correlated to the ability to win, but that average score margin can be adjusted for strength of schedule and random variations in day-to-day ability.

The CRS gives the expected score difference in a hypothetical matchup, and assuming Gaussian variation in day-to-day ability, gives a closed formula for the probability one team will defeat another. This system could be easily adjusted for other sports, but is particularly suited for NCAA Division I men’s basketball because of the large number of games played and the finely graded scores.
The two-thirds winning record of home teams is often cited as evidence that a home court advantage exists. However, the HCA controlled for differences in home and away team abilities is zero. This is partially accounted for by the trend that strong teams invite weak teams in the first half of the season to warm up for conference play.

On the other hand, the Home Court Advantage calculated from paired home-away games is 3-4 points, suggesting the HCA is real. Raymond Cheong currently do not have a satisfying explanation for the discrepancies between these analyses.
On balance, this evidence neither confirms nor denies that the HCA exists, but at minimum that if an HCA is included in a CRS the way it is calculated must be carefully evaluated - see www bme.jhu.edu/ -.

You must be logged in to read or post a comment. Register

tags:  basketball  ~  statistician's digest   ~  basketball analysis  ~  statistics
visit:  eBA Portal  ~  eBA Forums  ~  eBA Encyclopedia  ~  eBA Clinics