How Inherited is Athletic Ability in Ice Hockey?¶

There are a lot of things that factor into athletic performance: skill, determination, reflexes, and more. When we think about successful players and what makes them so great we tend to focus more on the things they earned rather than the things their genetic gifts.Of course, body type doesn’t determine performance, and players work on making themselves stronger and faster. It’s still something to consider. According to a medical journal, “body composition and body weight are two of the many factors that contribute to optimal exercise performance. Taken together, these two factors may affect an athlete's potential for success for a given sport” (Rodriguez).

This leads me to wonder what the ideal body type is for a hockey player. In sports like basketball, we can easily say that being very tall is preferred. Ice Hockey is a contact-heavy sport that requires a lot of agility and endurance. According to Jack Han, a hockey analyst, “hockey is not really a vertical game”, which means height isn’t that important (Han). Long arms and a heavy torso are two characteristics that can provide an advantage in this sport. Another thing you want is short legs because “shorter legs means better acceleration” (Han).

Off the top of our heads however, we can't really give numerical values for the ideal height and weight for a player. It’s true that a player can succeed even if they don't have the ideal body type, but there has to be a range of values that contain the highest number of successful players. That is the question I am looking to answer in this project.

In the process of finding the answer to this question, I hope to gain insight on some other interesting topics as well. How has the ideal height and weight of a hockey player changed over time? Humans have gained more insight on nutrition over the years and it has led to taller, more fit players. We also need to consider how changes in the game can affect the ideal body type. Another important thing to consider is that we can't lump all the players together. Forwards, defenseman, and goalies have different jobs and it's likely there are different ideal body types to best support these jobs.

In [1]:
# Data
import pandas as pd

data2018_2019 = pd.read_csv('/Users/diyajhamtani/Desktop/DS 2500/Project/PlayerInfo2018-2019.csv')
data2018_2019.head()
Out[1]:
Player S/C Pos DOB Birth City S/P Ctry Ntnlty Ht Wt Draft Yr Round Overall 1st Season HOF GP G A P
0 Matt Cullen L C 1976-11-02 Virginia MN USA USA 73 202 1996 2 35 19971998 N 71 7 13 20
1 Zdeno Chara L D 1977-03-18 Trencin -- SVK SVK 81 250 1996 3 56 19971998 N 257 14 38 52
2 Joe Thornton L C 1979-07-02 London ON CAN CAN 76 220 1997 1 1 19971998 N 221 33 79 112
3 Patrick Marleau L C 1979-09-15 Aneroid SK CAN CAN 74 215 1997 1 2 19971998 N 204 31 37 68
4 Derek MacKenzie L C 1981-06-11 Sudbury ON CAN CAN 71 177 1999 5 128 20012002 N 1 0 0 0

I loaded data from the 2018-2019 season into a pandas data frame shown above.

Here's a data dictionary.

Name Type Description
Player string Player Name
S/C string Side Skater Shoots or Side Goalie Catches
Pos string Player Position (c --> center(forward), d --> defense)
DOB string Player Date of Birth
Birth City string Player City of Burth
S/P string Birth State or Birth Province
Ctry string Birth Country
Ntnlty string Player Nationality
Ht int Player Height
Wt int Player Weight
Draft Yr int Year Player was Drafted
Round int Round in which Player was Drafted
Overall int Overall Pick # in Draft Round
1st Season string Year of First Season Played
HOF string Is Player in Hall of Fame (yes or no)
GP int # of Games Played
G int # of Goals Scored
A int # of Assists
P int # of Points

The data set contains the height and weight of all players who played during the season, along with the points and games played. We can calculate the points per game played as a measure of success for a player. The dataset also contains the positions that players play, allowing us to have an idea of how position (forward or defenseman) affects the results. We can also check if players from a certain geographical region have different average height and weights. This data set is from one season, but I will also gather data from multiple seasons to compare my results over time.

There is a small concern for the accuracy of the heights and weights listed. Jack Han talks about how “almost every player on the Habs’ (hockey team) roster is an inch or two (sometimes more) shorter than advertised on the official team media guide” (Han). For the project I’m going to assume the heights are accurate. Another potential issue is that a player's points can be influenced by their linemates. For example, if a player plays on the same line with a very good player, they may put up more points than they would on a different line. This project uses the data of over a thousand players so hopefully this won’t have a significant impact on the results.

Machine Learning Aspect:

I would use regression and input the heights and weights to predict the points per game played. Then I would compare the estimates to the true values. The heights and weights that lead to the highest points/games played would be the ideal heights and weights. I’d then compare across multiple seasons to check for any meaningful changes.

Works Cited

CathySquires. “How Big Are the Leafs?” Pension Plan Puppets, Pension Plan Puppets, 12 Oct. 2022, https://www.pensionplanpuppets.com/2022/10/12/23383386/average-height-weight-age-32-nhl-teams-2022-2023-season.

Jack_Han. “The Anatomy of an Ideal NHL Player.” Eyes On The Prize, Eyes On The Prize, 19 Aug. 2015, https://www.habseyesontheprize.com/analysis/2015/8/19/9171871/anatomy-of-the-ideal-nhler-montreal-canadiens-carey-price-size-scouting-measurements-nhl.

“Nutrition and Athletic Performance.” Medscape, 1 Mar. 2010, https://www.medscape.com/viewarticle/717046_6.