Loading scores...
Hockey Analytics

Breakouts & Rushes Via Passes

by Gus Katsaros
Updated On: October 4, 2018, 4:04 pm ET

The singular greatest adaptation of what was commonly referred to as ‘advanced stats’ was in the introduction of proxies. Possession grew from the initial parsing of NHL data that became the basis for modern analytics.


These proxies have proved adequate to determine results, judge player and team performance while adding more layers to predictive power. Modern work is based on the proxy for puck possession, a shot-metric focused analysis tool. Corsi (shot attempts) and Fenwick (unblocked shot attempts) have proven to have some predictive power and newer models are using expected goals.


The Passing Project, organized by Ryan Stimson and crew (with their second data release), has some potential with actionable data for game play and could become proxies themselves. Ryan expanded on this already, releasing an attempt at using passing data to present in the shot quality and introducing a passing metric called Passing Shot Contribution (PSC).


For now, we’re going to focus on data that isolates breakouts and solo rushes.


Before getting into that, limitations exist.


Data in the Passing Project is predicated on an eventual shot event. Passes not resulting in a shot event aren’t reflected in the data and it’s a bit of an opportunity lost. Tracking is of course very difficult and there has to be a basis for mass tracking. Analyzing passing data that led to a non-shot event has advantages to analyze game-play, especially for systems play, but there’s still a lot of actionable game-play information.


For now, this is how the breakout picture looks as of the latest data release.




The raw data in the passing file looks like this, parsed in categories across the top.


View post on imgur.com


To determine defensive zone breakouts the data has to be conditionally filtered. Fortunately, the data is presented to determine game flow through sequencing. Three potential passes are tracked prior to the shot event (with SOG being marked as well as if the shot came from within the scoring chance home plate).


We are strictly using 5v5 as game strength for all items in this writing. Filtering the A3 zone beginning with a ‘D’ signifies the third pass began in the defensive zone, signifying a breakout originating from the defensive zone. (If the passing project tracked location – even in sub-zones of the defensive zone – we could determine if the pass originated from behind the net (controlled breakout), or from the top of the zone (regroup). I haven’t lost hope in isolating regroups, but it’s a work in progress.)


The image below represents that data with the first column representing breakout per games played (with the stipulation the breakout ended with a shot attempt). The second column differentiates a breakout differential that ended up with a shot on goal – not just an attempt.


The Gms Tracked column shows the imbalance across teams tracked. Uniformity for analysis skews the data among different teams. Los Angeles seems to have the best breakout per game ratio (that leads to a shot attempt), however they only have seven games tracked. Half of their attempts end in a shot on goal.


Highlighted teams represent above average games tracked. Project Lead Ryan Stimson, being a New Jersey Devils fan has tracked the most (37 – we will get back to the Devils in a moment), with Washington, Chicago, Toronto and Tampa Bay topping 20 tracked games.


Of all the above average tracked teams, the Maple Leafs have the best ratio of breakout to shot on goal (0.689).


View post on imgur.com


It’s unfortunate the previous year’s data doesn’t contain the same sequencing information. It will be interesting to see how this all shakes out in the end with a more complete data set.


Directionality can also be ascertained with each zone marked where the pass originates.


Something else that appeared once looking at breakout data.


Solo Rushes


The concept of a ‘solo’ rush is that of a single player attacking in the offensive zone. For this purpose, last player to accept the pass in the neutral zone and execute a shot on goal is deemed a solo rush.


The Randy Carlyle led Toronto Maple Leafs gained a reputation for being easy to play against. The term ‘rush team’ was coined to describe their style of play. Lots of teams can be categorized as rush team, but the Leafs exemplified a ‘one and done’ offensive strike, scoring off the rush more so than off a cycle or with sustained zone pressure. I broke down the Carlyle led Leafs offensive zone systems, dubbing them ‘aggressively passive’.


Rush teams aren’t very effective. The style is conducive to capitalizing on a spatial shooting situation and reversing quickly into a defensive stance; quick strikes, shots then defensive posture.


David Johnson of Hockey Analysis provided some analysis


His definition of a rush shot:


To test this theory I defined a shot off the rush as the following:

  • A shot within 10 seconds of a shot attempt by the other team on the other net.
  • A shot within 10 seconds of a face off at the other end or in the neutral zone.
  • A shot within 10 seconds of a hit, giveaway or takeaway in the other end or the neutral zone.


Using the Passing Project data we can pinpoint passes that ended up as shots on goal off the rush, when a player gets a pass in the neutral zone.


The theory behind this is the passing sequence ends up in a shot on goal, whether inside the scoring chance area, or outside. To do this, I’ve eliminated the first two passes (A3 & A2), but focused instead on the A1 pass originating the neutral or defensive zone and ending in a shot on goal. This rush chance could be a solo attack or in conjunction with teammates, but the last player to the get the pass in the neutral zone takes a shot on goal. Without any other passing and no turnover event, the shot on goal demonstrates a solo effort not originating from a cycle or extended zone time (some assumption is required there).


The Toronto Maple Leafs (20 tracked games), across the top is the A1 zone (zone from which the pass preceding the shot originated) with the shooter in the first column. The A1 zone is broken down into two categories. The ‘y’ column represents shots taken within the scoring chance home run plate. The ‘(blank)’ denotes a shot on goal outside of the scoring chance area.


Using Toronto below, there’s a lot of solo effort via Nazem Kadri. If he receives a pass in the neutral zone – and mainly through center – he’s likely to end up firing the puck on net. Is this a Kadri trait, or is this exhibited throughout the lineup? Toronto in general likes to use the center part of the ice to distribute passes before quick strikes. This could be extremely helpful information.


The takeaway? Limit Kadri’s space where he’s likely to receive passes in the neutral zone. His tendency is to finish plays more to himself than passing it off… defensemen should likely expect that and step in his way; put a stick in his shooting arc, get into his skating lanes to force changes in direction, do something to just get in his way.


If there was non-shot event data, we could compare to how many times (or types) the play breaks down in this instance.


Teams can place a roaming defensive forward in the center ice area whenever Kadri is on the ice. Limiting the pass reception potentially thwarts any shot attempts – while forcing play to either wing.


View post on imgur.com


New Jersey had 37 games tracked as of the latest data release, 17 more than Toronto. Their rush table looks significantly different from the Buds. Five shots have been generated off the rush in the scoring chance area, while most of the shots are being taken from the outside.


It’s no wonder the Devils rank among the bottom in 5v5 goals.


Toronto, with more shot production from the scoring chance area has barely scored above New Jersey and don’t rank much higher than the Devils.



View post on imgur.com




The data disseminated through the passing project can be used in myriad ways.


New companies now tracking data using video breakdowns are taking advantage of the lack of sophisticated public data, but it’s highly suspect any data is shared outside of proprietary concerns.


With more sophisticated data and many more games tracked, there are many different avenues here to distinguish more accurate game play.

Gus Katsaros
Gus Katsaros is the Pro Scouting Coordinator with McKeen’s Hockey, publishers of industry leading scouting and fantasy guide, the McKeen’s Annual Hockey Pool Yearbook. He also contributes to popular blog MapleLeafsHotStove.com ... he can be followed on Twitter @KatsHockey