Is It Time to Increase the Use of the Long Change in Hockey? A proposal for the NHL
The game of hockey has gone through many changes throughout its lifetime. Both the equipment and the rules have radically altered the game, and with it, the strategy teams employ to win. No longer do goalies instinctively play a “stand-up” style in order to avoid a puck to their unprotected face. Similarly, the team with a lead is no longer allowed to freely shoot the puck down the ice, as they did before icing was introduced in 1937–38. (In one game in 1931, the winning team “iced” the puck 50 times and fans were so enraged they threw trash on the ice).
In more recent history, a host of new changes were introduced in 2005–06 in response to the 2004–05 lockout which was one of the longest lockouts in American professional sports’ history. These changes affected everything from the size of the offensive and neutral zones to the ability to make two-line passes to the size of goaltender equipment. Although these changes seem far-fetched, even more drastic changes were proposed. One that is often debated is the size of the goal. Goalies, as you can imagine, are not amused by this idea. As well as many fans who feel that this is one, or several, steps too far.
Therefore, the NHL must be careful in not overstepping its bounds with drastic rule changes. With this in mind, I have a proposal that will increase scoring, shots, and odd-man situations while not changing anything about equipment or gameplay, per se: make the long-change more frequent. If teams have their defensive zones further from the net for two, or even three periods, line changes will be more difficult, and the game will be less predictable. This has already been done to a small extent for overtime since the 2015–2016 rule changes. After 2015–16 overtime is played with a long change. This, along with overtime shifting to 3-on-3, quadrupled the number of games decided in overtime, compared to the previous short change, 4-on-4 structure. Some argued then, as I’m arguing now, that the league can go further and make the long change more frequent during regulation play, too.
Data
To make my proposals, and to review the current state of the NHL short and long change structure, I use detailed game event stats from the 2011–12 to the 2018–19 seasons. The data originally come from the NHL stats API, but were collected in SQL tables here. I stored this data on the Google Cloud Platform and queried and transformed the data in SQL with the mysql.connector library in Python. This data was visualized in a Jupyter Notebook using both Python (Seaborn) and Tableau (the Tableau report can be viewed here).
Current NHL Stats
In order to make my proposal, I will start by reviewing current game play stats. This will give us an idea of how play differs by the type of change occurring.
Overall, nearly every stat differs noticeably by period. In most cases, the second period has the most of a stat, followed by the first, and then the third period. At first this is somewhat puzzling: why do the first and third periods differ in the same stat category? They both have the same type of change, after all. The difference, however, is due to different in-game strategies: often, at the end of the game, the winning team is playing much more defensively than they would when the score is tied, as they always are at the start of the game.
In what follows, I show the period averages over all games in my dataset.
Goals
Probably the most important stat, goals, is clearly in favor of having more long changes. The second period has, on average, the most goals, followed by the third and then first. The fact that there are more goals in the third may be because empty net goals almost exclusively occur in the third period. Still, the second period has 5% more goals than the first period, on average. Therefore, it is clear that the long change has more goals than periods with a short change, even when counting empty net goals.
To find out why there are more goals in the second, it helps to take a look at how some of the other stats change in the second period.
Shift Length
Each position has a longer average shift in the second period than they do in the first or third periods. This is quite intuitive: having a further distance from the defensive zone to the bench means that its harder to change players on the fly. The shifts in the second period are 7% longer than the first for forwards, and 10% longer for defensemen. With more tired players on the ice, it makes sense that play is more open and that more mistakes are made.
Penalties
Penalties also show a clear difference between periods. On average, 0.13 more penalties are called in the second period than in the first period. The length of the shifts, and by extension, tired players, probably cause the increase in penalties in the second period. Futher, penalties are clearly tied to more goals: in the 2008–09 season, odd man situations produced more goals than 5-on-5 play, about three times more goals! Penalties clearly happen more often in the second period.
Shots
Shots, too, happen more often in the second period. Given what we know about shift length and penalties, this result should not be too surprising. Although the quality of shots taken is not considered, shots can be taken as a proxy for how offensive the play is. Here, we see that the second period is the most offensive, while the third, where one team is most likely protecting a lead, has the fewest shots on average.
Takeaways, Giveaways, Hits
I should mention the stats that don’t follow the trend we saw for goals, length of shifts, penalties, and shots.
Stats that decline slightly throughout the game
- Takeaways
- Giveaways
Both stats mostly show bad decision making on the part of the offending team and are not necessarily correlated to the distance they have to skate to get to the bench.
Stats that noticeably decline throughout the game
- Hits
It’s unclear how this relates to distance from the bench.
Stats that remain the same throughout the game
- Blocked shots
- Stoppages
Blocked shots probably relate to the type and quantity of shots taken in that period: point shots are much more likely to be blocked than rebounds. Stoppages seem like they should increase in the second period, but rules incentivizing free-flowing play (no change icing, delay of game for shooting puck out of play in defensive zone, etc.) probably are why we see no change by period.
The Proposals
Knowing that important offensive stats vary by period, how would we expect these stats to change if the long change was increased? To answer this question, I used the averages for the first and second periods, and recombined them as if the game is already played with two or three long changes. I also multiplied the third (long change) period by the ratio we see between the third period and first period. This was done in order to simulate the difference in game strategy that already occurs in NHL games in the third period. Lastly, I summed the values in the baseline (current) scenario, the two long changes scenario, and the three long changes scenario. Here is how the stats compare:
Most importantly, we see that the number of goals would increase in both proposals. Shots, and penalties would also increase in both proposals. These stats mean that there would be more exciting, back and forth, unbalanced play in both long change scenarios.
Giveaways, takeaways and blocked shots would slightly decrease in each scenario. Hits would more noticeably decrease, by about two hits a game in the three long changes scenario. In contrast, stoppages would very slightly increase in each scenario.
Drawbacks to Changing
So I’ve made my case. Let’s say you buy it. Great! The question now becomes whether the benefits outweigh the costs. This is where the question is less about numbers and more about values.
The major reason for keeping things the same is that people are tied to how things are now. It just feels wrong. Although this change is not as big as changing the net size (pun intended), it still requires an adjustment, from players and fans alike. Along the same lines, it does require minor, visual adjustments for all parties involved. For example, fans who attend games choose their seats based on whether the home team’s offense more often occurs close to the fan, or whether the home team’s defense more often occurs on their side of the ice. For example, if a fan has the away team goaltender near them in the first period, they will see more home team goals on their side of the ice. If the situation all of a sudden gets reversed, many will be unhappy. I wouldn’t blame them. That’s why I propose that the benches switch sides, but teams still shoot in the same direction. For example, if fans are used to the TV showing the home team bench being on the right and shooting on the left during the short change, under the new proposal, the home team will be on the left and still shoot left in the first and third periods. This way, the direction of play doesn’t change, only the placement of the benches changers.
Additionally, there are minor logistical challenges to solve. The national anthem(s) at the beginning of the game usually have the starting line up stand on the blue line closest to the bench. The team then starts the game on that same side. In my proposal, if they stand for the national anthem near the bench, they will then need to start play on the other side. Although minor, one does have to consider these issues when weighing the benefits and drawbacks of any proposal. In my humble opinion, the benefits to gameplay outweigh these minor logistical challenges, for the league, the networks, players and fans.
Conclusion
The rules of hockey, on a grand scale, have not varied significantly since the beginning of the game: score more goals and you win the game. This is something fans appreciate and make the game timeless. While the equipment and use of TV slightly affect gameplay the game has always stayed close to its roots. With the proposals made here, the game becomes more exciting, with minimal changes. By increasing the number of periods teams make a long change, we can increase the number of goals, shots, and odd-man play. At the same time, takeaways, giveaways, hits, and stoppages remain very close to their current values per game. With a minor change, we can make the game more exciting with minimal investment; a win-win for all. The question is: will the positives be enough to sway purists and change the status quo? Only time will tell…
P.S.: The code used to make this project can be found in the project’s GitHub repo here.
