The NHL is undergoing a statistical revolution. The availability of data is increasing each season, and with player and puck tracking becoming standard across NHL arenas, this data will only increase.
While having large data sets has led to the development of complex predictive models like those created by Micah McCurdy and EvolvingWild among others, the evaluation and collection of certain pieces of data is still subjective, and inconsistent.
Last week on Twitter, Dimitri Filipovic commented on the inconsistencies in counting takeaways.
He’s not wrong. For certain statistics, the NHL just seems to get it wrong. Takeaways is one example, and another is hits.
If you dabble in the world of fantasy hockey, you know the importance of category coverage, and getting players who can run up totals across the board is key to winning matchups. This is what makes players like Matthew Tkachuk, Tom Wilson, J.T. Miller, and Andrei Svechnikov so valuable – they can score, but also dish out hits.
However, there is a clear bias in how hits are tabulated in different NHL arenas, and while this has zero impact on the actual on-ice product, it has a massive effect on the record book, and fantasy hockey.
Hit count bias by arena
Using data from the previous three seasons (2017-18, 2018-19, and 2019-20), we conducted the following analysis for all teams.
- The total number of hits the team recorded in all home games was calculated.
- The total number of hits for the team’s home opponents was calculated.
- The above two totals were summed to give the “total home hits” value, and this value was divided by the total home games played across the three seasons.
- The total number of hits the team recorded in all away games was calculated.
- The total number of hits for the team’s away opponents was calculated.
- The above two totals were summed to give the “total away hits” value, and this value was divided by the total home games played across the three seasons.
- The differential was calculated by taking the total home hits and subtracting the total away hits. This gives the hits bias for each arena in the NHL.
The resulting differentials for each team are plotted below, split by division.
The reason for calculating it on a team by team basis was to minimize the effect of team play in the hit counts. The premise is that a heavy hitting team should, on average, have the same hit counts on the home and on the road. Over three seasons of data, changes in style of play should even out, and teams should be hitting at the same rate regardless of the arena they play in.
However, we can clearly see below that this is not the case. Certain arenas award hits more judiciously than others, and certain arenas see higher hit counts than others.
Of the seven teams in the North division, there is a clear positive hit bias in three arenas, and a clear negative hit bias in two.
In Ottawa Senators and Edmonton Oilers home games, teams see an increase in hits, both home and away. The differential here is 9.0 for both teams, which means that games featuring the Senators and Oilers will have higher hit counts when those teams are at home versus on the road.
Conversely, the Calgary Flames and Winnipeg Jets see a decrease in hits for both home and away teams. Again, this means that games featuring the Flames and Jets see more hits when those teams are on the road versus at home.
This has major fantasy hockey implications. If you’re looking to add a player to increase your hit count for a particular week, it’s crucial to know what the schedule looks like for that team. If you’re choosing a Canuck, chances are the hits per game you’re looking for will be accurate regardless of schedule.
If you’re adding a Senator or Oiler, you need to make sure that player has a slate of home games instead of away games, or you’ll be losing out on hits due to the arena bias. An Oiler playing in Winnipeg will record fewer hits than the same Oiler playing against Winnipeg at Rogers Place.
This affects road teams as well. Picking up team hits leaders from teams like Calgary and Winnipeg will yield higher than average hits on weeks they are on the road simply due to the arena they’re playing in.
Takeaways: Players with games in Edmonton, Ottawa, and Montreal should be favoured, and players with games in Calgary and Winnipeg should be avoided, purely looking at average hit counts per game.
The East division has the largest spread of any division in the NHL. Pittsburgh is the team with the most significant positive hit bias in the league, with a whopping 16.6 more hits recorded in each Penguins game at home versus on the road.
Because Pittsburgh is at the very top of the differential column, let’s take their hits leader, and NHL hits leader, Brandon Tanev, as an example. He’s played four home games and four away games for eight total games this season.
Tanev currently has 44 hits for the Penguins. That’s an average of 5.5 hits per game in total, which is downright bonkers. He’s fantasy relevant just based on that stat. However, looking at his home and away splits tell a different story.
At home, Tanev has recorded 26 hits compared to just 18 on the road. That’s 6.5 hits on average at home and 4.5 hits on the road. That’s a difference of almost 45%, and greatly alters what you should be predicting for his hit counts if he’s at home versus on the road.
On the flip side, let’s take a closer look at the Buffalo Sabres. They have the most significant negative hits bias in the division at -11.6. Their hits leader is Rasmus Ristolainen, and his home and away splits paint a similar picture.
At home, Ristolainen has 11 hits over four games, compared to 18 over four road games. He averages 2.75 hits in each home game and 4.5 hits in every away game. That’s a very significant difference.
Takeaways: Players with games in Pittsburgh should be favoured, and players with games in Buffalo should be avoided, purely looking at average hit counts per game.
The same reasoning applies to all teams in all division. In the West, there are more negative hit bias arenas than positive. A significant positive hit bias exists in Vegas where there are 10.7 more hits per game, and a slight positive hit bias in Arizona where there are 5.7 more hits per game.
A significant negative hit bias exists in Minnesota where there is an average of 11.7 fewer hits per game, and a slight negative hit bias exists in St. Louis and Colorado where there are 7.2 and 5.6 fewer hits per game respectively.
No significant bias exists in San Jose or Anaheim, so per game hit rates are likely what you’ll see for each player on these teams.
Takeaways: Players with games in Vegas should be favoured, and players with games in Minnesota and St. Louis should be avoided, purely looking at average hit counts per game.
The Central division has two teams with significant positive hit bias and two with significant negative bias. The other four teams have relatively no bias, with a maximum magnitude of 3.3 hits per game differential.
Chicago is the major positive bias team in this division, with 14.3 hits per game more for games held in the United Center. Carolina is the other positive team with 8.2 hits per game higher when the Hurricanes are at home.
On the negative side, games in Nashville see 12.5 fewer hits per game, and games in Columbus see 10.0 fewer hits per game.
Takeaways: Players with games in Chicago and Carolina should be favoured, and players with games in Nashville and Columbus should be avoided, purely looking at average hit counts per game.
The impact of hits biases
The fantasy impact of arena hits biases has already been discussed, but the overall impact is greater.
In terms of player valuation, stats like giveaways, takeaways, hits, and blocks can play a major role. T.J. Brodie was criticized for years in Calgary for his giveaways. Chris Tanev is lauded for his shot blocking. Bruising fourth liners see a high percentage of their value derived from how often they dish out hits and how often they drop the gloves.
This arena bias should offer additional context in how we value these types of players. Maybe adjustments need to be made, or a more standard definition of what constitues a hit needs to be developed by the league.
That said, until the league develops a standard definition for a hit that is implemented league-wide, knowing which arenas have higher than average hit counts is very valuable for fantasy owners. On weeks where you are down in hits, looking at whether a big hitting player is playing on the road or at home can be the difference between winning or losing the week