![]() As expected, the rankings were heavily skewed to modern teams ‐ 40 of the top 50 teams played in the 2000 season or later. We ran all of the NHL games for the last 74 years through our Elo ratings system without adjusting for the extra point (from the 2000 season forward) and lack of tie games (for the 2006 season forward). Thanks for the history lesson, what does this have to do with the rankings? A lot. Sounded good in theory, but the change didn’t have the desired effect and the NHL eliminated tie games completely, by introducing a shootout for the 2006 season. The theory was: if both teams in a tie game were guaranteed at least one point, they would be less likely to play it safe in overtime. The NHL determined that tie games were bad so, for the 2000 season, the league decided to award an extra point in tie games. ![]() Enough about methodology and why it makes sense, and on to what I learned. The best part about an Elo ratings system is the fact that it doesn’t treat all wins and losses the same ‐ it looks at the quality of the wins and losses (is there such a thing as a quality loss?) This fact makes it perfect to compare Wayne Gretzky’s Edmonton Oilers to Sidney Crosby’s Pittsburgh Penguins because it looks at the quality of the opponents each team faced, and converts this quality assessment to a number. The easiest way to think about Elo ratings is: you get rewarded for beating good teams, punished for losing to bad teams, and your score only changes marginally if you win (lose) games where you are the favorite (underdog). If Team A is heavily favored to beat Team B, Team A’s score is not going to improve a lot if it wins however, if Team A is heavily favored to beat Team B and Team A loses, Team A’s score is going to decrease a lot. The probability calculation is where the magic happens. A team’s performance is converted to a score by first calculating the probability of Team A beating Team B and then increasing Team A’s score if it beats Team B, or decreasing Team A’s score if it loses to Team B. An Elo rating system distills a team’s performance to a single score ‐ the higher the score, the better. You can skip this part if you already read our 50 most dominant teams in NBA history list ‐ you’re already up to speed on the benefits of the Elo ratings system.īefore getting into why, it’s probably helpful to at least mention how the Elo rating system works. We took the data and ran it through our Elo rating system in order to compare teams across different seasons. So, that’s what we did: we looked at the last 74 NHL seasons ‐ from 1943 to 2017 ‐ to try to determine the greatest NHL team ever. While we wait for Elon Musk to perfect time travel, we can use some math and analytics to tackle the question. It would be cool to grab our time machine, pick up some of the all‐time great teams in NHL history, and have a tournament to decide who is the best. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |