Alex Novet Hockey Graphs ghostofnyles Teams with similar total talent levels can look very different Red Team Blue Team PA Parenteau Sidney Crosby Joel Ward PA Parenteau Boone Jenner ID: 1044297
Download Presentation The PPT/PDF document "Strong and Weak Links: Talent Distribut..." is the property of its rightful owner. Permission is granted to download and print the materials on this web site for personal, non-commercial use only, and to display it on your personal computer provided you do not modify the materials and that you retain all copyright notices contained in the materials. By downloading content from our website, you accept the terms of this agreement.
1. Strong and Weak Links: Talent Distributions within TeamsAlex NovetHockey Graphs@ghostofnyles
2. Teams with similar total talent levels can look very differentRed TeamBlue TeamP.A. ParenteauSidney CrosbyJoel WardP.A. ParenteauBoone JennerJoel WardDan GirardiBen HuttonDmitry OrlovBen HuttonGAR: 6.8GAR: 17.2GAR: 5.4GAR: 6.8GAR: 7.1GAR: 5.4GAR: -3.2GAR: 6.0GAR: 7.2GAR: 6.0Total GAR: 32.2Total GAR: 31.1My goal today is to convince you to pick the red team
3. Strong Link Game: The team with the best player usually winsWeak Link Game: The team without the worst player usually wins
4. Strong Link vs. Weak Link is a way of understanding how a system works and how to improve performanceOROR
5. It is not immediately clear if hockey is a strong link game like basketball or a weak link game like soccerStrong Link GameWeak Link GameEach player has the ball a small fraction of the game1 goal can win a game, so mistakes matterOne player can easily get possessionFew players are on the court at onceThe team with the best player usually winsThe team without the worst player usually winsFew players are on the ice at onceWeakest player can be hidden with limited ice time1 goal can win a game, so mistakes matterBest skaters only play a fraction of the game
6. We can begin to test this by examining the best and worst player on each team as measured by GARThe Strong and Weak Link are UncorrelatedThe Strong Link is More Related to SuccessCorrelation = -0.121Strong Link Correlation = 0.348Weak Link Correlation = -0.037
7. More formally, we can use a regression to show the strong link is the more explanatory player
8. The regressions aren’t perfectly leveraged but seem suitable
9. If we think about talent as a resource, there are several techniques from other fields we can applyThese have been used in sports before, but not in quite this wayMany fields have ways to measure the distribution of resourcesStatistics: Standard Deviation or Range1Business: Herfindahl-Hirschmann Index (HHI)2Economics: Gini Coefficient3
10. Each of these measures looks at the distribution of talent throughout the entire team
11. Each of the regressions indicate that more extreme talent distributions do better than balanced ones
12. In the current draft system, tanking worksTradesChoose quality over quantityFree AgencyIf you have to choose, spend money on the star, not the benchDraftImplicationsHockey is about creating plays, not avoiding mistakesCoaching
13. Next StepsReview by a real statisticianIncorporate goalies Improve on ranking system and address colinearityIncorporate marginal cost of improving a player given their roleThank you to Dom Luszczyszyn, Micah Blake McCurdy, and Dawson Sprigings for data collection help