Passing clusters: A Framework to Evaluate a Team’s Breakout

Quick breakouts – trying to move the puck out of your zone right after gaining possession – make up roughly 38% of possessions and account for 22% of all shots and 22.4% of Expected Goals (at least according to my possession and xG definitions). Therefore, understanding what does and does not work when breaking out the puck against present forecheckers is important. There is evidence that passes from the defensive half boards by wingers inside produce more offense than those straight up ice. But the puck is more often recovered elsewhere, so these passes by wingers aren’t the first pass in a possession and are therefore presumably influenced by the previous play. It should be interesting to find out how the inclusion of the pass(es) that came before affects this conclusion.

One way to do this is pass clustering. After all, if all you have is a hammer, everything looks like a nail.

An important note: The way the data is collected, it isn’t possible to know the intended location of a pass. All I know is where the puck ended up. Therefore I can’t really use failed passes in the analysis, since there’s no way of knowing where a failed pass was meant to go. I believe that there is enough value added by looking at successful passes, but any analysis that only focuses on successful events is inherently flawed. A good alternative example would be slot passes: Of course slot passes are very good when successful, but if you don’t know how frequently slot passes actually reach their target, the conclusions you can draw about decision making in the offensive zone are limited.

As a reminder, the blue passes are the pass clusters we’ll focus on:

They are also 3 of the 4 most frequent second passes in KHL hockey. To illustrate what kinds of passes are intended to be covered by the three wing pass clusters 7, 11 and 12, here are some clips:

Cluster 12: Cross Ice

This play clearly needs some space, but as you can see here, all three Omsk players collapse to Grigorenko’s (#25 red) side of the ice which gives Slepyshev lots of open ice.

Cluster 11: Straight Up

Telegin (White #7) is on his regular side at the start of the clip, but comes over to the strong side to give Kalinin (#21) an outlet up ice. The benefit here is that the chip along the boards can take multiple opponents out of the play, especially pinching defencemen.

Cluster 7: Inside

The intent here is to get the puck up the boards and then hit a streaking player (usually C) with a quick pass inside. Here, Andronov (#11, white) does a good enough job of not overcommitting to pressuring Petrov (#90, red, along the boards), so the play doesn’t fire perfectly. Like the chip along the boards, this is meant to quickly take a pressuring player out of the play.

First Passes

Now let’s take a look at the various first passes that lead to these plays. The green passes plotted above (clusters 4, 5 and 8) account for 65% of all set ups to the winger passes. And in general, these are three of the four most frequent first passes in KHL hockey, with the most frequent being an over/reverse pass behind the net.

In order to compare the general effectiveness of these first passes, we need to separate pass clusters 4 and 5 into rim and direct passes, as there is quite a difference in being able to play the puck directly to a winger at the half boards and having to rim it along the boards. Rim and Up are two of the common hockey breakouts, so let’s use these categories here as well. Rims are passes from clusters 4 and 5 that go along the boards, while Up passes are direct passes of cluster 4 and 5 and all passes of cluster 8. In addition to these passes as first passes of a possession, I also possessions where they were the second pass after an over/reverse play. This gives us the following picture (Stats are per 100 Possessions):

PassPossessionsDumpoutsPossession in NZPossession in OZShotsxG

The differences here are quite stark, Rims are a lot less efficient. If you take into account that rims are much more likely to result in a turnover (since the opponent has more time to pressure the receiving winger), they become a very inferior play option. This of course isn’t news to anyone, since for most teams, rims are more of a last resort. Now we can check what these numbers look like for each 1st pass/wing pass cluster combination:

So the 1st pass can indeed have quite a large impact not just on the success rate of wing passes but also the type of wing pass that is attempted. Rims are much less frequently followed by a stretch pass than Ups. The overall efficiency of stretch passes following rims is a little surprising, but my guess would be that this is largely a function of the space afforded to the winger.
Although the overall efficiency of stretch passes on rims looks tempting, any conclusion advocating for this breakout play presupposes that a) a rim is just as likely to be successful and b) a stretch pass is just as likely to be successful, which as I mentioned at the top, can’t be done with the data I’m working with here. Tracking data that includes intended pass locations on failed passes would allow for a more complete analysis.

This framework of pass clusters can be used to evaluate a team’s own breakout and preview an opponent’s breakout. For instance, without ever watching a minute of their games, you can already determine that

  • Admiral Vladivostok and Torpedo Nizhny Novgorod prefer having their wingers hit the center rushing through the inside (cluster 7 is used far more – over 38% – than any other pass clusters or carries following a rim or up pass)
  • Dinamo Riga, Traktor Chelyabinsk and Lokomotiv Yaroslavl prefer to have the weak side winger move over to provide support down the ice for cluster 11 passes (over 36% of the time)
  • while other clusters are slightly more frequent, Salavat Yulaev Ufa have the highest tendency to hit forwards for cross-ice stretch passes from the wings

This information can be used to improve opposition scouting and find video material to provide good examples for dos and don’ts for forecheckers and pinching defencemen on how to handle these plays.


In this article, I focused on one type of play: a pass up to the winger. But developing a taxonomy on all of the most frequent breakout patterns using pass clustering could greatly streamline the process of describing a team’s breakout preferences and abilities.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s