The authors use data from fight logs during the 2016- 20 17 regular National Hockey League (NHL) season to test for a difference in the proportion of games with and without fights for each of the thirty NHL teams. Only one team (Toronto Maple Leafs) was more likely to be involved in a fight at a home game than at an away game. Teams that fought proportionally more often in the second half of the season made the playoffs; teams that fought significantly less often did not. And, long distance air travel (flights involving more than 1000 miles or trips that required crossing at least one time zone) resulted in disproportionately more games with fights for eight different NHL teams.
In 1945, George Orwell [
What constitutes a “fight” in the NHL? A “fight” is a one-on-one altercation or a bench brawl resulting in a five-minute penalty to two or more players or a ten-minute misconduct or game misconduct penalty to the instigator (who starts the fight). Such penalties can have a decisive effect on the outcome of a game. There are many reasons for fights: “bad blood” between players, retaliation, retribution, intimidation, disabling star players, among others. “Cabin pressure” on long flights might also take its toll on NHL players. That is, long distance air travel might be a factor that explains why weary short-tempered players drop their gloves.
In this brief research note, we compare for each NHL team during the 2016- 2017 regular season the proportion of games resulting in at least one fight in home games v. road games, the first half of the season v. the second half of the season, and back-to-back games (at home or away) v. all other games. Does air travel involving more than 1000 miles or crossing one or more time zones result in disproportionately more games with fights? Is the proportion of games with fights greater when an American team plays a Canadian team (than when two Canadian or two American teams play one another)? In short, are there any patterns to fighting in the NHL that might help ice hockey fans better understand the game?
The archive of hockey fights in the NHL [
The number of miles NHL teams traveled for the 2016-2017 regular season was compiled by Dirk Hoag [
The two data files were merged, matching the date of the game and the opponents for each of the 2460 observations on each variable of interest. For example, the Anaheim Ducks hosted the Columbus Blue Jackets on October 28, 2016. There was one fight that night between Josh Manson (Ducks) and Josh Anderson (Blue Jackets) at 01:50 of the first period. The Ducks’ distance was zero miles, as they were the home team. The distance traveled by the Blue Jackets was only 333 miles (far less than the distance between Anaheim, California and Columbus, Ohio), because the Blue Jackets had played a road game the night before 333 miles away in San Jose, California. The “distance” for the home team is not always zero. For example, the Anaheim Ducks played a home game (their 17th of the season) the evening of November 15, 2016. Yet, the “distance” for the home team that night is 1761 miles. The Ducks’ game 16 was played on the road in Nashville, Tennessee and the air distance between the two cities is 1761 miles. That is, the Ducks had to travel almost 2000 miles to return home in time to host the Edmonton Oilers in their next game. And, there was a fight that night between the Oilers’ Zach Kassian and the Ducks’ Josh Manson, at 07:01 of the second period.
A series of one- and two-tailed t-tests was run on the difference between two population proportions using STATA.
The comparisons involving the proportion of games with fights in 1) home v. away games or 2) games through the first half of the 82-game regular season v. the second half of the season are both two-tailed t-tests. For the team-by-team comparisons involving home v. away games, the null hypothesis (H0) is stated as follows:
H 0 : π Home = π Away (1)
while the two-tailed competing alternative hypothesis (HA) is:
H A : π Home ≠ π Away (2)
where π denotes the population proportion of games with at least one fight. Rejecting H0 in favor of HA would suggest that there is a large enough difference between the sample proportions to conclude (using α = 0.05) that the difference between the true population parameters, πHome and πAway, is not zero. The comparisons involving back-to-back games and games where the team had to fly over 1000 miles or cross at least one time zone to play their next game were one-tailed t-tests. For example, for the team-by-team comparisons involving back-to-back games, the null hypothesis (H0) is stated as follows:
H 0 : π Back − to − back = π Notback − to − back (3)
while the one-tailed competing alternative hypothesis (HA) is:
H A : π Back − to − back > π Notback − to − back (4)
where π is (again) defined as the proportion of games with at least one fight.
The results of the various t-tests are reported in Tables 1-7.
Team | Proportion | p-value on differencea | |
---|---|---|---|
Home | Away | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.415 0.244 0.268 0.195 0.049 0.195 0.341 0.098 0.317 0.293 0.317 0.390 0.268 0.293 0.268 0.195 0.317 0.366 0.195 0.171 0.268 0.293 0.098 0.220 0.195 0.341 0.341 0.122 0.268 0.122 | 0.463 0.171 0.293 0.293 0.098 0.317 0.268 0.195 0.220 0.390 0.171 0.366 0.122 0.317 0.171 0.195 0.268 0.415 0.146 0.146 0.366 0.317 0.146 0.317 0.146 0.317 0.098 0.244 0.268 0.220 | 0.656 0.414 0.806 0.304 0.396 0.206 0.472 0.211 0.319 0.352 0.123 0.820 0.095 0.810 0.286 1.000 0.627 0.651 0.557 0.762 0.343 0.810 0.500 0.319 0.557 0.814 0.008 0.153 1.000 0.240 |
ap-value for a two-tailed t-test where H0: πHome = πAway vs. HA: πHome ≠ πAway. bNames in italics are playoff teams.
Team | Proportion | p-value on differencea | |
---|---|---|---|
First half | Second half | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.463 0.317 0.293 0.268 0.098 0.341 0.268 0.073 0.390 0.366 0.366 0.341 0.146 0.317 0.220 0.171 0.220 0.293 0.146 0.073 0.341 0.268 0.146 0.220 0.195 0.317 0.268 0.220 0.244 0.098 | 0.415 0.098 0.268 0.220 0.049 0.171 0.341 0.220 0.146 0.317 0.122 0.415 0.244 0.293 0.220 0.220 0.366 0.488 0.195 0.244 0.293 0.341 0.098 0.317 0.146 0.341 0.171 0.146 0.293 0.244 | 0.656 0.014 0.806 0.607 0.396 0.077 0.472 0.061 0.013 0.641 0.010 0.495 0.265 0.810 1.000 0.577 0.145 0.070 0.557 0.034 0.635 0.472 0.500 0.319 0.557 0.814 0.286 0.392 0.618 0.078 |
ap-value for a two-tailed t-test where H0: π1st half = π2nd half vs. HA: π1st half ≠ π2nd half. bNames in italics are playoff teams.
Team | Back-to-back away games? | p-value on differencea | |
---|---|---|---|
Yes | No | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.364 0.273 0.111 0.167 0 0.214 0.111 0.429 0.400 0.444 0.182 0.250 0.182 0.357 0.091 0.375 0.417 0.444 0 0.250 0.250 0.357 0 0.636 0.429 0.154 0 0.231 0.273 0.222 | 0.517 0.133 0.355 0.345 0.129 0.370 0.290 0.147 0.161 0.375 0.138 0.394 0.100 0.308 0.207 0.156 0.214 0.406 0.188 0.121 0.394 0.307 0.194 0.200 0.091 0.393 0.107 0.250 0.267 0.194 | 0.808 0.147 0.920 0.873 0.872 0.846 0.863 0.044 0.056 0.353 0.364 0.776 0.239 0.375 0.806 0.083 0.095 0.419 0.908 0.178 0.776 0.375 0.500 0.004 0.012 0.937 0.881 0.553 0.485 0.425 |
ap-value for a one-tailed t-test where H0: πBack-to-back = πNot back-to-back vs. HA: πBack-to-back > πNot back-to-back. bNames in italics are playoff teams.
Finally, we note that for 716 of all 2016-17 regular season games involving two American teams the proportion of games with at least one fight was 0.244; for the other 514 games the corresponding proportion was 0.255 (the two-tailed p-value on the difference was 0.554). For the 454 games when an American team played a Canadian team the proportion of such games with at least one fight was 0.240; for the other 776 games the corresponding proportion was 0.254 (p = 0.446). Yet, for the 60 games when a Canadian team played another Canadian team, the proportion of games with at least one fight was 0.367; for the other 1170 games the corresponding proportion was statistically smaller at 0.243 (p = 0.002). Fighting appears to be much more common in games involving only Canadian teams. That is, Canadian, not American, teams seem to have a stronger taste for violence. Of the 60 games between Canadian teams, 22 resulted in at least one fight. Edmonton was the host five times; Calgary, Montreal, Ottawa, Toronto, and Vancouver were the hosts three times each; and Winnipeg twice.
Fighting in the NHL has been commonplace. Since its inception in 1922, the league’s Rule 56 regulates but does not ban violence and imposes a minor, major, misconduct or game misconduct penalty on ice-skating pugilists. Fighting occurs, if for no other reason than as a wake-up call for a tired team. But no one heretofore has investigated whether fights occur more often at home than away, or more often in the latter half of the season than in the first half, or whether long distance air travel is a determinant of fighting. Do only good teams (read: teams that make the playoffs) fight?
The results presented here for the 2016-2017 regular season indicate that only
Team | Did the home team travel at least 1000 miles before the home game? | p-value on differencea | |
---|---|---|---|
Yes | No | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.444 0.429 0 0.500 0.333 0.667 0.300 0.500 0.200 0.333 0.667 0.300 0.625 0.571 0 0.500 0.333 0.250 0.333 0 0.500 0 0 0.143 0.500 0.333 0.500 0.100 0.333 0.667 | 0.406 0.206 0.297 0.179 0.026 0.158 0.355 0.077 0.333 0.286 0.289 0.419 0.182 0.235 0.275 0.179 0.316 0.378 0.184 0.179 0.243 0.316 0.103 0.235 0.179 0.343 0.324 0.129 0.257 0.079 | 0.419 0.106 0.899 0.132 0.009 0.016 0.625 0.025 0.726 0.406 0.088 0.750 0.006 0.038 0.730 0.132 0.475 0.694 0.265 0.745 0.135 0.876 0.683 0.705 0.132 0.518 0.241 0.596 0.349 0.001 |
ap-value for a one-tailed t-test where H0: πAt least 1000 miles = πLess than 1000 miles vs. HA: πAt least 1000 miles > πLess than 1000 miles. bNames in italics are playoff teams.
Team | Did the visiting team travel at least 1000 miles before the home game? | p-value on differencea | |
---|---|---|---|
Yes | No | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.308 0.067 0.444 0.125 0.143 0.333 0.333 0.250 0.111 0.250 0 0.357 0.222 0.308 0.125 0.250 0.125 0.250 0.167 0.125 0.200 0.333 0.333 0.222 0.152 0.500 0.333 0.222 0.182 0 | 0.536 0.231 0.250 0.333 0.088 0.316 0.241 0.189 0.250 0.448 0.200 0.370 0.094 0.321 0.182 0.182 0.303 0.455 0.143 0.152 0.389 0.314 0.114 0.344 0.125 0.241 0.057 0.250 0.300 0.237 | 0.914 0.911 0.129 0.877 0.329 0.475 0.273 0.385 0.813 0.882 0.886 0.533 0.149 0.535 0.649 0.331 0.846 0.854 0.439 0.576 0.794 0.463 0.080 0.756 0.576 0.053 0.018 0.568 0.775 0.830 |
ap-value for a one-tailed t-test where H0: πAt least 1000 miles = πLess than 1000 miles vs. HA: πAt least 1000 miles > πLess than 1000 miles. bNames in italics are playoff teams.
Team | Did the home team travel across at least one time zone before the home game? | p-value on differencea | |
---|---|---|---|
Yes | No | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.400 0.200 0 0 0.200 0.333 0.182 0.167 0.267 0.250 0.400 0.545 0.400 0.600 0.200 0.250 0.333 0.643 0 0.167 0.800 0 0.167 0.091 0.250 0.400 0.333 0.091 0.154 0.667 | 0.419 0.269 0.297 0.216 0.028 0.171 0.400 0.069 0.346 0.310 0.306 0.333 0.250 0.194 0.290 0.189 0.316 0.222 0.211 0.171 0.194 0.333 0.086 0.267 0.172 0.333 0.342 0.133 0.321 0.079 | 0.543 0.691 0.899 0.850 0.047 0.178 0.904 0.169 0.701 0.650 0.335 0.109 0.239 0.007 0.712 0.385 0.475 0.004 0.812 0.511 0.002 0.938 0.269 0.886 0.284 0.384 0.512 0.644 0.870 0.001 |
ap-value for a one-tailed t-test where H0: πCrossed at least one time zone = πCrossed no time zones vs. HA: πCrossed at least one time zone > πCrossed no time zones. bNames in italics are playoff teams.
Team | Did the visiting team travel across at least one time zone before the home game? | p-value on differencea | |
---|---|---|---|
Yes | No | ||
Anaheim Ducksb Arizona Coyotes Boston Bruins Buffalo Sabres Carolina Hurricanes Columbus Blue Jackets Calgary Flames Chicago Blackhawks Colorado Avalanche Dallas Stars Detroit Red Wings Edmonton Oilers Florida Panthers Los Angeles Kings Minnesota Wild Montreal Canadiens New Jersey Devils Nashville Predators New York Islanders New York Rangers Ottawa Senators Philadelphia Flyers Pittsburgh Penguins San Jose Sharks St. Louis Blues Tampa Bay Lightning Toronto Maple Leafs Vancouver Canucks Winnipeg Jets Washington Capitals | 0.500 0.250 0.273 0.167 0.111 0.364 0.316 0.160 0.136 0.318 0.100 0.389 0.100 0.300 0.217 0.273 0.182 0.455 0 0 0.167 0.250 0.222 0.278 0.048 0.250 0.308 0.190 0.222 0.250 | 0.440 0.095 0.300 0.345 0.094 0.300 0.227 0.250 0.316 0.474 0.194 0.349 0.129 0.333 0.111 0.167 0.300 0.368 0.188 0.188 0.448 0.349 0.125 0.348 0.250 0.349 0 0.300 0.304 0.207 | 0.354 0.094 0.568 0.873 0.438 0.349 0.262 0.761 0.917 0.846 0.753 0.393 0.596 0.591 0.185 0.224 0.775 0.288 0.920 0.920 0.956 0.724 0.233 0.684 0.967 0.724 0.001 0.793 0.722 0.381 |
ap-value for a one-tailed t-test where H0: πCrossed at least one time zone = πCrossed no time zones vs. HA: πCrossed at least one time zone > πCrossed no time zones. bNames in italics are playoff teams.
one NHL team was involved in disproportionately more fights at home or at away games. The Toronto Maple Leafs appeared to scuffle more often at home. Teams that scuffled significantly less often in the second half of the season failed to make the playoffs. And, surprisingly, teams that scuffled significantly more often in the second half of the season did make the playoffs. Three NHL teams― all of which made the playoffs―were revealed to be more belligerent when they were scheduled to play road games on consecutive nights. Long distance air travel appears to make a difference. Eight different teams that traveled more than 1000 miles or traveled across one or more time zones before their next game were involved in fights more often than when distant travel was not an issue.
Finally, fighting appears to be more frequent in games involving only Canadian teams. In the words of Don Cherry, Canadian ice hockey commentator for CBC Television: “Anybody who says they don’t like fighting in the NHL has [sic] to be out of their minds.”
Castillo, H.L. and Sommers, P.M. (2017) An Analysis of Fights in the National Hockey League. Open Journal of Statistics, 7, 676-688. https://doi.org/10.4236/ojs.2017.74047