clock menu more-arrow no yes

Filed under:

2015 Passing Project Data Release: Volume I

New, comments

In this article, I lay the groundwork for how and when I plan to release our passing data. This first post, in what will likely be a long series, goes over basic player stats. Read on for the details.

If you buy something from an SB Nation link, Vox Media may earn a commission. See our ethics statement.

Kim Klement-USA TODAY Sports

Why do this?

That’s the first question I had before I even started tracking anything way back in October of 2013. What could I hope to learn from tracking passing plays? It started around the idea that as so much of hockey is random, passing represents a skillful play to link between players and generate shot attempts. By isolating that aspect of the game, you’re essentially isolating skilled, intentional plays made by the players. The other thing was a quote from Gavin Fleig in Soccernomics: "Teams that complete a higher number of passes in the final third consistently finish in the top four of the league."

So, that’s where it started. Could I find a passing metric that correlated well with teams winning games? Could isolating skilled plays teach us anything new about the game? The answers to both questions was a resounding "Yes."

Last summer, I wrote up my findings from the first full season of this project. This season, I wanted to increase the level of detail of the data I’d collect and, hopefully, would find a few people to help out. I never really thought people would stick with it for an entire season, but I was incredibly lucky to find a number of people willing to devote their time and efforts to this project. In total, we’ve tracked 492 games for this season, including every game for the Devils, New York Rangers, New York Islanders, Florida Panthers, Washington Capitals, and Chicago Blackhawks. If you’re wondering how we track, you can use the above link to my primer, or watch this video that we put together to give some transparency to the tracking process.

Now, there is a tremendous amount of data to be mined from our efforts and even more analysis that I have yet to do. So, here’s how this is going to work as far as releasing data. Every week or two, there will be a section of data I will either tweet out or post in a piece like this, likely the latter. Today will be the player stats for the entire season. This does not include Royal Road or One-Timer data as those will be discussed when I get into the on-ice data release. I still have some work to do to prepare that. But, there will be player data, on-ice data, team data, what leads to goal scoring, winning, losing, etc. There will be goalie and shot sequencing analysis. It's going to take months to post on everything, so bear with me. So, what will you see today?

Each time a player completed a pass that led to a shot attempt, shot, or goal is in the below chart. You’ll see whether or not it was a primary or secondary pass, whether or not it was generated in transition, and whether or not it was generated within the scoring chance (home plate) area or elsewhere in the offensive zone.

Dropbox Link to Data

You’ll see how efficient each player’s passing was (efficiency is defined as the proportion of shot attempts generated that resulted in actual shots or goals), how much offense they contributed to, or their offensive "impact" for lack of a better term, as well as how often they contributed. You’ll also see rates for each metric on a per 60 minute basis. All our stats are displayed in 5v5 and 5v5 Close formats. You can filter by team, player, position, and Time on Ice (TOI) groupings of at least 100, 200, or 300 minutes. Actual TOI is provided as well. You can also filter by number of attempts generated (at least 10, 20, or 30) to get rid of some players with minimal events.

Also, huge thanks to Sam Ventura and Andrew Thomas at War on Ice for working with me to get the on-ice and specific player/team data on only the games we tracked.

I won't go through everything here, but to quickly point out some of the better passers in the league is some familiar names and some surprising names. In terms of those players that generate shot attempts from primary passes per sixty minutes, with at least 200 minutes player, here are your top producers. All data in subsequent charts is of 5v5 data only.

SAG_Best

What matters?

So what, right? I’ll be getting into which metrics matter more than others in the coming weeks/months, but I’ll say a few words on both the descriptive and predictive value of this project now. From a descriptive standpoint, it is obviously worthwhile to know precisely how involved players are in a team’s shot generation. We often hear about possession drivers and passengers, but can only partly quantify that through Relative stats. That’s part of the descriptive benefit, moving from the macro, on-ice level of stats to the micro, individual-level of involvement.

So, the first thing you’d want to know is whether or not passing is something that’s repeatable. Well, to do this I took the number of shot attempts a player generates from primary passes per sixty minutes during the first half of the season and plotted it against their production from the second half of the season. Players who played at least 200 minutes in each half of the season were used from our six teams. This gave me 107 players. Here are the results.

SAG_60

Here there is a strong relationship that the ability to make a pass that leads to a shot attempt is predictive of itself and representative of player skill. For comparison’s sake, let’s see how a player’s individual shot attempts per sixty minutes fares.

iCF_60

Very similar. Nearly identical. What does this tell us? It tells us that the ability for an individual player to attempt a shot or generate one for a teammate is one deriving from skill and we can trust it as a means of evaluation going forward. I’ll be getting into more of this data as time progresses, but the important takeaway in this first post is this: if generating attempts via passes is just as reliable as a player’s own shot attempts in terms of repeatability, then it needs to be a part of how we evaluate players and how they contribute to offense. You can see this in the CC% (Corsi Contribution) column in the data (column BG). This represents the percentage of shot attempts a player contributes to via their own shot attempts, primary passes, and secondary passes that lead to shot attempts. Or, a player’s true offensive impact. Now, since percentages are only useful alongside rate stats, I’ve included this same concept as CC/60, or Corsi Contributions per 60 minutes. How reliable is a player’s total offensive contributions? Using the same criteria as above, let’s have a look.

CC%

CC_60

We see that how much offense goes through a player (CC%) and how often they contribute (CC/60), both are highly predictive of themselves and we would expect them to be reliable going forward.

These are the most basic and foundational passing metrics. Whose passes lead to shot attempts? How often do they generate passes? How much offense do they truly contribute to? There's so much more to come, but at the very least I wanted to establish the reliability of these foundational metrics. Okay, I'll show you one thing I've been working on.

If you can isolate shooting and passing, you can discover which has a greater impact on the other. What I mean by this is which is more likely to create a goal, since that is the, pardon the pun, goal of hockey. Passing increases a shot's likelihood of becoming a goal. I discussed this at length in my presentation in DC last month. So, if we wanted to try to project out how many points a player will score, is it better to look at a player's own shot attempts, or those he generates from passes?

I used the same criteria as above, but instead of a player's total points per sixty minutes, I used only his primary points per sixty minutes. Those would be goals and primary assists. This makes sense for two reasons: 1) The NHL cannot reasonably award a secondary assist to save its life; and 2) shot attempts and primary passes are primary events. Let's have a look.

iCF_60

So, a player's individual shot attempt rate from the first half of the season can predict about 26% of the points they'll score over the second half of the season. Certainly significant. Let's see how primary passing compares.

SAG_60_points

Based on these 107 players, it would appear if you want to forecast a player's primary point totals over the second half of the season, you should look at the rate at which they generate shot attempts from the first half of the season rather than their individual shot attempt rate. It would result in you predicting 8% more accurately. As linked above to my DC presentation, successful passes lead to more goals, so this would make sense.

Lots of good stuff to come down the pipe here at InLouWeTrust. Stay tuned!

To the Devils

For those of you reading this that do not care for Devils analysis, feel free to download the data and peruse on your own free time. Don’t hesitate to hit me up on Twitter (@RK_Stimp) if you have questions. For the rest of you, let’s have a look at some Devils passing numbers over the course of the season.

If you'll recall, in this article, I highlighted how our weighted data showed Jon Merrill to be better than we think. One aspect of a defenseman contributing to offense and quality sequencing of events is the breakout, or first, pass that you often hear people talk about when evaluating defensemen. I decided to take a look at this phase of the game and see if it gave us any additional statistical support to Merrill's positive play. The reason for this is simply: on shot sequences that begin with a secondary pass in transition, teams shoot at 5.4% on all shot attempts on the end of that sequence. This is higher than a secondary pass in the offensive zone (5.0%), a single pass in transition (3.1%), and a single pass in the offensive zone (4.8%).

Def_A2

Here we see that Merrill generates two shot attempts with secondary passes in transition every sixty minutes, a higher frequency than all other Devils defensemen. The Y-axis shows that when Merrill is on the ice, he contributes far more to the team's total offensive output via these types of passes. This is significantly higher than the other defensemen, doubling some of them. What does this mean? Simply this: When Merrill makes an outlet or breakout pass, the Devils are more than likely to generate a shot attempt off of his than any other defensemen. To lead the breakout or attack in transition is valuable. With this data we quantify that oft-referenced talking point of a "first pass" and see just how good the Devils defensemen are.

For the forwards, you often hear about "keeping plays alive" and "loose puck recoveries" in order to generate offense once the team is inside the offensive zone. Well, we can quantify that as well. There are some passing sequences that go beyond two passes before a shot attempts is taken, but it's obviously more difficult to do so. Once a team loses possession, they must regain it and we often see the player recovering the puck is almost always a forward and they dish back to another forward of a defenseman on the point. Either a shot attempt or another pass and then a shot attempt are likely coming as a result of that puck recovery (as an aside, I think tracking loose puck recoveries is something that would be incredibly worthwhile). For now, we can use a forward's offensive zone secondary passes leading to shot attempts as a metric that helps quantify these talking points of "keeping plays alive" and "loose puck recoveries." You can also think of this as sustaining offensive pressure. Let's have a look.

Fwd_A2

It's unsurprising to see Scott Gomez atop any passing list seeing as how he choose to pass more than shoot; however, he consistently generates sustained offensive sequences once the team sets up in the offensive zone. There is a drop off to the next player: Adam Henrique. Henrique is likely a top-six complementary player at his ceiling, but suffers from having very little in the form of quality linemates in Jersey. That doesn't stop him from generating more sustained offense from secondary passes than anyone not named Gomez, including the player right behind him on this list: Jaromir Jagr. After these two, there's another drop off to a group of players that includes Travis Zajac, Jacob Josefson, Martin Havlat, and Dainius Zubrus.

So, this gives us a sense of which players are keeping the puck alive or recovering and then dishing to a teammate to allow the team greater sustained possession.

Meet the Team!

As in many of my articles or presentations, I'd like to acknowledge the many people who have helped out with this project. Their names are below. Even if they just tracked a single game, it all matters and I appreciate their time. Everyone tracked a team of their choice or pitched in close to the end in an effort to finish certain teams. We all tracked the opposition in each game, so that explains why we have data on every team.

I tracked the Devils games, most of the Blackhawks games, and a few others here and there as we wrapped up the season.

ILWT's very own Brian Franken tracked the Rangers this season (@onepasthunter)

Shane O'Donnell tracked the Panthers, Bruins, and Blackhawks (@shane13420)

Jesse Severe tracked the Capitals (@jessesevere)

Ryan Stoll tracked the Islanders (@rybee824)

Cameron Taylor tracked the Flames (@cofstats)

Francis Mahoney tracked the Lightning and Capitals (@vipfrancis3)

Benoit Roy tracked the Sharks (@Benroy_)

Robert Fitch tracked the Bruins and Blackhawks

Andrew Glazer tracked the Blackhawks (@amglazer)

Brandon Dedrick tracked the Sabres (@iatemybrother)

Craig Wercynzski tracked the Avs

Kevin Higginbotham tracked the Stars

Nick Biss tracked the Blue Jackets

Spencer tracked the Blackhawks (SpenceIce)

Your Requests

As I mentioned above, there's going to be a tremendous amount of data and articles written on this site over the next several months. As I release it, I'll write a bit more on various Devils players, but I also want to hear from you. What would you like to see analyzed now that you can look at this data yourselves? What makes sense? Doesn't make sense? What goes against what you thought about a player or supports what you thought about a player? Sound off below or ask me on Twitter.