Rush Offence and team tactics

A technical breakdown and details of the model and data collected can be found at the end of the article.

Introduction

         Rush play is an important part of hockey, with important team level implications on both the defensive and offensive sides of the puck. While some rush plays like two on ones or breakaway are easy to appreciate for their obvious danger, most rush plays do not start out as obviously good scoring opportunities. In fact, most rush shot attempts look a lot closer to dump ins than actual meaning scoring chances. Sometimes a rush shot is a one person effort to get the puck in deep while four other linemates change after a prolonged shift in their own defensive zone. Other times, a rush shot can be a creative 2 on 2 play resulting in a high danger scoring chance. All this to say, there’s a wide range of different rush situations, and we can learn about teams and scoring tactics by analyzing them.

          With this in mind, I believe that we can learn a lot about a team’s offensive approach by observing their approach to rush shots. For example, teams that feature high volumes of rush shots from distance may represent teams that take a “any shot is a good shot” approach, while low volume rush teams with limited shots from distance may prioritize establishing in zone possession rather than take a low value shot. I also believe that there’s merit to different approaches, especially when we consider team composition and individual player ability. However, I also do believe that some approaches are better than others, for both short term performance, but also to support player development. Therefore, my goal with this small article is twofold: 1) To describe the general trends across various forms of rush chances, and 2) to examine the relationship between rush shot creation and overall offensive creation.

Methods

         All shot attempts in an OUA hockey game are screened by me, and all unblocked shot attempts are tracked. Importantly, one of the variables captured for each unblocked shot attempt is the play context. Specifically, the four play context categories in my tracking include a play being: In Zone, Rush, Advantaged Rush, or Neutral Zone/Defensive Zone shot attempt. In addition to the play context, I also include the specific details of an advantaged rush (e.g 2 on 1, 3 on 2, breakaway, etc.). In the case of rushes and advantaged rushes, I classify a rush shot attempt as any shot attempt taken within roughly 5 seconds of entering the offensive zone, though since I don’t have game clock time stamps in most of video tracking, this is not necessarily a hard cut off. There are also two exceptions where this may not be a perfect definition. First, if there is a quick rebound attempt off the initial rush shot attempt, I classify the rebound attempt as being a rush play as well even if the rebound shot is beyond 6+ seconds from the zone entrance. This also means that there could be a 2 on 1 shot attempt (called an advantaged rush), but the subsequent rebound is then classified as only a rush, since a defending player may have caught up within the 1-2 seconds it takes to get the rebound shot off. I believe in this system because it most accurately describes the state of play and can highlight opportunities in which teams reestablish some form of defensive positioning. The second exception to the earlier definition are counter attacks, where the attacking team has created a turnover at or around the offensive blue line. These plays typically more closely resemble a rush than they would an in zone offensive set up, and I include them as part of rushes rather than their own category, though this may change in my future collection. Importantly, these counter attacks starting high in the offensive zone make up a very small percentage of all rush opportunities, somewhere slightly less than 5 percent of all rush unblocked shot attempts.

Results

Question 1: What are the general trends to rushes across the different types?

I believe the best way to visualize where offence is being created is by using density maps, where darker colours indicate a greater percentage of shooting volume. Below are four sets of density maps that showcase four unique situations. The first density plot includes all rush attempts, which I don’t find to be particularly useful. Comparing a 1 on 3 zone entrance rush attempt with a 2 on 1 or a breakaway tells us next to nothing, thought it is interesting to see still such a strong amount of volume coming from higher than the faceoff dot line. What I find much more interesting is splitting up rush attempts into three categories: all non-advantaged rushes, all advantaged rushes, and all advantaged rushes without breakaways.

First, looking at advantaged rush attempts gives us a very tight density right around the crease through 554 attempts. However, this is heavily influenced by the data set containing 209 breakaways. When we remove those and look at non-breakaway advantaged rushes we see an extremely interesting distribution. While some advantaged rushes are forced to less than ideal zones by good defensive play, it is still very surprising to see such a wide and high volume range. In women’s hockey, shots from distance tend to be less valuable compared to the men’s game. That may be best seen specifically in advantaged rushes, where a shot from just inside the top of the circle on an advantaged rush is a high danger scoring chance in the men’s game, while it doesn’t hold quite the same weight on the women’s side. To more closely see that, here’s a shot map that includes the result of each shot attempt, where the size of the dot indicates the value in my expected goal model.

While there are certainly some goals from middle distance, I think this showcases the areas where teams benefit most from on advantaged rush plays. Examining goals only, the comparison feels even more stark. In non-breakaway advantaged rushes, we have seen 42 goals on 41.11 expected goals so far this season across 345 attempts, for an average value of 0.12 expected goals per rush attempt. If we filter by unblocked shot attempts from greater than 25 feet away, we get 9 goals on 7.66 expected goals across 153 attempts (0.05 xG per attempt). I think this critically highlights the importance of advantaged rush tactical approaches, because averaging 1 goal off 20 advantaged rushes will leave you scoring less than you’d like in those situations.

         Moving onto non-advantaged rushes, I believe that this is the area where we see the opportunity for coaches and teams to express not only tactical philosophy, but also showcase skill. A team’s ability to take a non-advantaged rush and turn it into a scoring chance is one way in which a team can maximize their 5v5 offence. Conversely, teams and coaches may also appreciate that player ability may limit the ability to attack successfully on non-advantaged rushes, and instead choose to maintain in zone possession unless they have a clear opportunity to attack in a dangerous setting. At the group level, we see heavy shot density from distance on non-advantaged rushes, with a bunch of density at the net front. When we take a closer look, again comparing shot outcomes and expected goals, we see 87 goals on 78.71 expected goals. We’ve seen 34 goals on 28.17 expected on shots from greater than 25 feet away (1545 attempts), and 53 goals on 50.54 expected on shots from within 25 feet (697 attempts). While closer to the net is obviously better, these non-advantaged rush plays from distance also include those situations where a player is throwing a puck to the net and getting off for a change, and sometimes this is absolutely the correct play to make. I’d also argue that there’s certain situations where a rush shot from distance can be used to create an offensive zone faceoff, allowing a team to get their strongest offensive line out for an offensive zone start. There’s nuance to decision making, but the overall scoring results do speak for themselves. Lastly, no matter the rush type examined so far, you can notice that not a single goal is scored from outside the dots, and very few from within the dot line lane.

Question 2: What is the relationship between rush shot creation and overall offensive creation?

Through the first 101 games of this season, rush shot attempts have accounted for 36.91% of all unblocked shot attempts. We see a wide range in of the average value of each rush attempt, and very wide ranges for how much of a team’s overall offence comes from the rush, both from an expected vs actualized perspective. Knowing that advantaged rushes are more likely to lead to greater scoring opportunities, I’ve also included a team’s total advantaged rush numbers to showcase the relationship between advantage rushes and total xG per rush attempt.

The correlation between xG per rush attempt and total advantaged rushes is 0.58, which indicates that advantaged rushes obviously play into how teams’ rushing averages look. Teams below the regression line are getting good overall rush value despite relatively fewer advantaged attempts, and those above are the opposite. In each case, this can be explained multiple ways, but each way is likely influenced by rush philosophy. With that in mind, let’s look closer at if teams approach rush attacks and in zone attacks similarly.

To examine team philosophy, I think the simplest approach is looking at the relationship with slot shots. Given the importance of getting the puck to the net front, if rush philosophy mirrors overall team philosophy, we would suspect a high correlation between the two, assuming that teams are roughly equally skilled at getting to the net front in both situations. In this situation, breakaways are removed from the analysis because they are almost a guarantee to get to the slot/net front. However, what we find instead is an extremely low correlation of 0.09 between rush slot shots, and in zone slot shots. The results are displayed below graphically, this time with team indicators completely removed. In this case, there’s no clear consensus amongst teams. More rush slot shots don’t clearly lead to more in zone slot shots. 

Now, what if teams are forgoing rush shots generally in favor of a greater number of in zone slot shots. This is where we would expect to see teams who are choosing a more selective approach. This time, we find an extremely low correlation of -0.02, indicating no consensus at a league level. Graphically, we see something pretty interesting. 

In small samples like this one (14 teams), a single outlier has a lot of weight. Removing that team reverses the trend, and instead finds that teams with more rush shot attempts actually have more in zone slot shots. This is a little surprising, though it is most likely indicative of better teams having the puck more, and therefore getting more shots generally. With that in mind, a few teams still do stick out as having a lower rush shot volume and notably higher numbers of slot shots.

The last area that I examined is team shot selection from beyond 40 feet on the rush and overall. We see a correlation of right around 0.5 between in zone shots from beyond 40 feet, and rush attempts beyond 40 feet. If we look at where that distance occupies, it does give us an idea of where rush offence is most related to team philosophy. That is, teams who are more likely to shoot from distance in zone may also view that same shot location as valuable on the rush.

Using a single team (identifying information removed), we can examine this phenomenon more closely. In the first image, red areas indicate locations where this team has a greater density of shots relative to league average, with blue indicating less shot density. This team has higher shot density along that 40 foot line overall (in all situations), and when we look at the same team’s non-advantaged rush shot density map, we can see some of the same patterns emerging.

Some teams choose to play by being more aggressive with their shot volume from distance, and that can be an effective strategy under certain circumstances. However, I’m less convinced of that approach holding true on non-advantaged rush chances given the lower likelihood for screens, tips, and rebound opportunities compared with in zone play. At the end of the day this overall line of thinking can probably be simplified to a single question; do you like to shoot from far away or not.

Conclusion

         Rush offence remains a critical component of game play and team success, with some teams relying on it for nearly half their offensive output or more. Undoubtedly, creating more high danger chances off rush and transition offence will make you a hard team to play against, and there’s incredible value in being that type of team. This breakdown highlights large differences in approaches to rush offence and in zone offence, though there’s some reason to believe that teams who take higher volumes of shots from distance off the rush will employ the same approach in zone.

Technical Background

         The data collected and reported in this article is retrieved as part of the WHIMS (Women’s Hockey Insight Model for Scoring) project. So far, this project includes over 25 thousand unblocked shot attempts collected in women’s Ontario University Athletics (OUA) hockey. The WHIMS model has also been tested out of sample at the international level (World University Games), and at the national USports level, demonstrating accuracy in both out of sample environments. The current WHIMS model being applied to the 2025-2026 season is an elastic net logistic regression model, trained on roughly 15000 unblocked shot attempts from the 2023-2024, and 2024-2025 OUA women’s hockey seasons. Model testing using 10 fold cross validation has the AUC at 0.815. For more information on the model, please check out my broader description in the WHIMS tab of this website.

Through the first half of the 2025-2026 OUA season, 7575 unblocked shot attempts have been recorded using An Nguyen’s shot plotter web app. All of my analyses and visualizations are done using R and R Studio, and due to my role currently working with one of the organizations included in this data set, neither the data nor code is currently available. This report is based entirely on the current results from the 2025-2026 OUA season, and all names and identifying information of teams has been removed.