1 / 18

Making the NHL Playoffs

This analysis by Dustin Schneider and Lawrence Mulcahy aims to predict NHL team playoff probabilities based on average game statistics using logistic regression. The model will assess data from the three seasons prior to the lockout, focusing on factors such as goals per game, goals against per game, and special teams performance. The study evaluates the significance of various predictor variables in determining a team's likelihood of making the playoffs, ultimately guiding betting decisions on teams like the New York Rangers, Columbus Blue Jackets, and Montreal Canadiens.

kelli
Télécharger la présentation

Making the NHL Playoffs

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript

  1. Making the NHL Playoffs Dustin Schneider Lawrence Mulcahy

  2. Objective • To predict the chance of a NHL team making the playoffs based on average game stats. • Logistic regression will be used to model and predict which teams have the best chance of making the playoffs. • The data set used will be limited to the 3 seasons prior to the lockout.

  3. Logistic Regression • Type of statistical model used to work with a response variable that has two outcomes: success or failure. • If we allow the response variable to take on the value of 1 or 0, the mean is the proportion of 1 and P(success) = p. • Logistic regression, however, is based on odds instead of the proportions of the outcomes.

  4. Odds • The population odds are defined as: where p is the proportion of the event. • If we have a proportion of .3333, then the odds are

  5. The form of the logistic regression model is: • The proportion in this model is actually the proportion of failure. • The model can use more than 1 predictor variable.

  6. Concordant Data • Begin by creating all possible pairs of observation with different responses. • Each pair is then classified as a discordant pair or a concordant pair. • The more concordant pairs the better the model.

  7. Predictor variable: pointsln(p/1-p) = 29.6968 – 0.3416(points)

  8. Predictor variable: Goals against per gameln(p/1-p) = -18.0235 + 6.8572(GA/G)

  9. Predictor variable: Goals per gameln(p/1-p) = 18.6687 – 7.2638(G/G)

  10. Predictor variable: Power Play %ln(p/1-p) = 7.1231 - .4530(PP%)

  11. Predictor variables: Penalty kill %ln(p/1-p) = 35.701 – 0.4278(PK%)

  12. Predictor variables: G/G GA/G PP% PK% ln(p/1-p) = 14.5743 – 9.3229(G/G) + 8.3898(GA/G) – 0.4983 PP% - 0.0521 PK%

  13. Predictor variables: G/G GA/G PP%ln(p/1-p) = 9.4912 – 9.3015 G/G + 8.6026 GA/G - .4906 PP%

  14. Predictor variables: G/G GA/Gln(p/1-p) = 5.3741 – 10.1836 G/G + 8.0720 GA/G

  15. Activity Professor Hartlaub wants to place a bet on one of the following teams making the playoffs. Which team should he choose to bet on? New York Rangers(NYR) Columbus Blue Jackets(CBJ) Montreal Canadiens (MTL) http://www.nhl.com/ice/teamstats.htm?navid=NAV|STS|Teams

  16. Hint

  17. Resources • NHL.com • http://www.jstor.org/stable/pdfplus/2685041.pdf

More Related