Interactive Data Visualizations at Hockey Graphs

I’ve been spending the bulk of my time recently putting together interactive data visualizations to be hosted at Hockey Graphs (link). These include historical charts of team shooting percentages, as well as metrics I have derived such as 2-Period Shot Percentage (2pS%) and the percentage of team shots a player takes in the games they participated (%TSh). The visualizations are mostly franchise history and player career visualizations, so they hold a lot of interesting storylines and trends that you can look through. You can also compare Alex Ovechkin’s career %TSh arc to Wayne Gretzky’s, or Wayne Gretzky’s to Jari Kurri’s, or Wayne Gretzky’s to Dave Semenko’s…suffice to say, there are a lot of options with the player charts. As for the team charts, you can see how Boston falls off after they lose Esposito and Orr, or how quickly team possession spiked upward when Chris Pronger came to town. The team shooting percentages include league-average and standard deviations for perspective, while the 2pS% charts have similar comparative lines. I encourage you to check them out; feel free to download the docs as well, there is a button on there for that as well.

Team History in Shots-For Percentage in the First Two Periods (2pS%), 1952-53 to 2013-14

Doing some more experimentation with interactive data visualization, this time with Tableau. This is also the point where I find out the generic WP sites don’t let you embed Tableau, so you’ll have to navigate to this one for the fun. Forgive me.

Interactive Player Career Charting, Using Percentage of Team Shots (%TSh) – 1967-68 to 2012-13

I’ve been putting together this data for quite a while, but then it lay dormant for a bit. Nevertheless, I wanted to tinker, and give you something you can tinker with as well. This data uses the percentage of team shots a player takes in the games they participated (%TSh; explanation here) to give you a comparable mapping of players’ careers. Like I said, it’s been a bit since I’ve visited the data, so it only goes up to 2012-13. Also, for the sake of the work it’ll put on the graph, I only included personally selected players in the filtering options (mostly top players, and some bottom players for perspective). A metric like this gives you a sense of just one aspect of offensive contribution, but the way it’s presented gets closer to the idea of “contribution” than a lot of other measures.

Note: To filter in/out the players you’d like to see, click on the grey “Name” button, then click on “Filter,” then go wild. …In a perfect world, I’d have a nice wide blogroll so you didn’t have to squint. Sorry.

Percentage of NHL Players at Selected Heights (71, 72, 73 Inches), 1917-18 to 2014-15

Per Michael Lopez’s request, I broke the height data into 71, 72, and 73 inch measures to see if there was any indication of a 6-foot bias, or the tendency to settle on the round measure rather than necessarily using the true measure.

And the data:

I don’t see anything too fishy here, though there’s an interesting dip when the league ballooned out of the Original Six era.

NHL Forward vs Defensemen Size, 1917-18 to 2014-15 (Plus Goaltenders)

It’s pretty remarkable how uniform the relationship between these two player groups remains over the years, suggesting that the NHL has maintained pretty consistent attitudes about how size benefits players on the defensive side of the puck. Whether that has to do with physicality, or reach and blocking, or how it can contribute to shot power, is up for debate, though I would wager it’s more about the former three.

Most noticeable to me is that, while there are some minor fluctuations up through the 1970s, from 1980 onward there’s an incredibly steady difference between height and weight. In other words, the attitudes about defensemen, forwards, and relative size seemed to really solidify by the 1980s. Also notice that the plateau in height and the downturn in weight I observed in my last post on player size continues to hold true whether or not you break apart by player position.

For goalie fans who feel neglected:

Two things really jump out here: the dip in size in the 1980s and the fact that goalies are now, for the first time, taller than the skater population. Might the dip have been another contributor to the scoring fluctuation in the 1980s? Whatever the case, the linear upward trend since then suggests teams chose to go taller and taller. Original post.

NHL Player Average Size, 1917-18 to 2014-15

Using a combination of NHL and national data, it’s interesting to see the divergence between the general male population and NHL height. Notice how obesity keeps the national figures close to NHL figures when it comes to weight:

Can’t help but notice what’s happened recently? NHL players have been getting smaller. Whether that’s because of an increased emphasis on skill or better conditioning, I think this is a promising trend. Basically, the league reached a tipping point where skill couldn’t be traded off for size, and I think that’s good for the quality of the NHL game, and I think it might help make the league less injurious.

I like to also have the standard deviations in place for these measures, just to see where/when the diversity of player sizes might have changed. Today’s NHL, it seems, might be one of the most diverse physiologically since the early years – and it’s a more “true” diversity, as those early leagues only had around 50 players.

Change in Time On-Ice Percentage at Different Strengths – Forwards & Defensemen

I have done something similar in the past, but I wanted to take another shot at this and get it how I wanted it. The data’s old, and worth comparing to the past three seasons, but I think overall the trends will hold.

I have done something to this effect before, but I had always presented it as a multiplier rather than as a simple change in percentage (don’t ask me why, I was young and carefree). Looking at year-to-year change in these age groups, I wanted to present the average change to ice time when a player arrives at the age on the x-axis. I wanted to hold the axis values pat across these so that they could be compared to one another. The trends above are using lines of best fit, each having an r-square of 0.75 or better. You’ll notice 5v4 usage had a more exponential trend, which was suggested by the data.

It’s really fascinating to see how 5v4 patterns differ here; both groups stabilize after losing the increase in minutes through the early 20s, but forwards eventually start to fall the rest of the way while (presumably) defensemen with the big shot are among those that are kept around in their late 30s (and kept on the point for the powerplay). I think this also relates to why the defenders are dropping off more than forwards in their late 30s.

When it comes to 5v5, though, everything is slow and steady, and same for forwards at 4v5 (I’m guessing, because they aren’t really pressed to be as active on the opposite end of the rink, and tend to be conservative in their own zone).

Both groups are passing 0% change at about age 28, which suggests that is essentially the point that teams are noticing the talent is not quite as high as the player peak. A wholesale decline in production is protected somewhat by the plateau in 5v4, but 34 is nearing the point of no return for the forwards, while a good offensive defenseman might be able to maintain until 38 or so.

Data for the above:

Def Age Δ 5v5 Δ 5v4 Δ 4v5
18 to 19 2.0% 10.8% 7.6%
19 to 20 1.8% 8.4% 6.9%
20 to 21 1.6% 6.2% 6.2%
21 to 22 1.4% 4.3% 5.5%
22 to 23 1.2% 2.7% 4.8%
23 to 24 0.9% 1.3% 4.2%
24 to 25 0.7% 0.1% 3.5%
25 to 26 0.5% -0.9% 2.8%
26 to 27 0.3% -1.7% 2.1%
27 to 28 0.1% -2.4% 1.4%
28 to 29 -0.2% -2.9% 0.8%
29 to 30 -0.4% -3.3% 0.1%
30 to 31 -0.6% -3.6% -0.6%
31 to 32 -0.8% -3.8% -1.3%
32 to 33 -1.0% -3.9% -2.0%
33 to 34 -1.3% -4.0% -2.6%
34 to 35 -1.5% -4.1% -3.3%
35 to 36 -1.7% -4.1% -4.0%
36 to 37 -1.9% -4.2% -4.7%
37 to 38 -2.1% -4.3% -5.4%
38 to 39 -2.4% -4.4% -6.0%
39 to 40 -2.6% -4.6% -6.7%
Fwd Age Δ 5v5 Δ 5v4 Δ 4v5
18 to 19 1.5% 10.8% 3.0%
19 to 20 1.3% 8.2% 2.7%
20 to 21 1.1% 5.9% 2.4%
21 to 22 0.9% 4.1% 2.1%
22 to 23 0.8% 2.6% 1.8%
23 to 24 0.6% 1.3% 1.5%
24 to 25 0.4% 0.4% 1.3%
25 to 26 0.3% -0.4% 1.0%
26 to 27 0.1% -0.9% 0.7%
27 to 28 -0.1% -1.3% 0.4%
28 to 29 -0.3% -1.6% 0.1%
29 to 30 -0.4% -1.7% -0.2%
30 to 31 -0.6% -1.9% -0.5%
31 to 32 -0.8% -2.0% -0.8%
32 to 33 -0.9% -2.2% -1.1%
33 to 34 -1.1% -2.4% -1.4%
34 to 35 -1.3% -2.7% -1.7%
35 to 36 -1.4% -3.1% -1.9%
36 to 37 -1.6% -3.7% -2.2%
37 to 38 -1.8% -4.5% -2.5%
38 to 39 -2.0% -5.5% -2.8%
39 to 40 -2.1% -6.9% -3.1%

Possession & Shooting Percentage: How Much Can I Sacrifice in One & Still Succeed by the Other?

Tradeoff Between Percentages and Possession

The blue and black lines represent the upper and lower bounds of what you can expect a team to sustain percentage-wise (save + shooting percentage). Generally speaking, what we’re seeing here is that a even a team that shoots well will struggle to make the playoffs if their possession is below 50%. Conversely, a team that shoots poorly needs to achieve a very high possession (55%) to make the playoffs. The key: it costs less to add possession than goal-scoring and goaltending, because the latter is conspicuous value. What’s more, chasing percentages is more risky than possession…which means if you convince yourself to chase the percentages over possession, you assume greater risk across the board.

The full piece for which I created this graph is viewable here over at Hockey Graphs.

Recent Publications

I’ve had a spate of success recently, getting the opportunity to publish in online and print at The Chicago Tribune and regularly at The Hockey News. Though some of The Hockey News pieces are still forthcoming (for both the upcoming World Junior Championships and Money & Power special issues), I do have links to most of the since-published-online work.

– The Hockey News, “THN Analytics: An Introduction,” October 8th, 2014, Link    *this piece also appeared in the THN “Fear” Issue, under the title “Truth in Numbers,” November 3rd, 2014

– The Hockey News, “THN Analytics: Slow Start? Firing the Coach Might Not Be the ‘Fix,'” October 16th, 2014 Link

– The Hockey News, “THN Analytics: Visualizing the Trade,” October 23rd, 2014 Link

– Chicago Tribune, “Possession Arrow Pointing Up for Blackhawks,” November 1st, 2014 Link    *this piece also appeared in the Sunday Morning Tribune, November 2nd, 2014

– The Hockey News, “THN Analytics: The Statistical Argument Against Fighting,” November 7th, 2014 Link    *this piece also appeared in the THN “Fight” Issue, under the title “Busting Heads, Busting Myths,” December 8th, 2014

– The Hockey News, “THN Analytics: Comparing NHL Greats With New Historical Data,” November 27th, 2014 Link

Needless to say, this takes away from work I’d otherwise be doing here or at Hockey Graphs, but they’re also pretty exemplary of the work I would have been doing here or at HG in any case, hence why I’m linking them.

The Literary Stanley Cup: Conclusion

Photo by “Mafue”, via Wikimedia Commons

What can you say about an underdog that loses? That it was deserved? That all is right because of it?

Or maybe we get too wrapped up in team affiliation to realize that it’s all amazing. Anze Kopitar made it here from Slovenia, Henrik Lundqvist made it here from the 7th round, Justin Williams had a series of injuries that nearly jeopardized his career – Marc Staal, too, Mike Richards and Jeff Carter were cast-offs with media-created character issues, and Martin St. Louis and Mats Zuccarello are supposed to be too small for the pro game. The lineups of these two teams are filled with underdogs, people with enthralling stories, both devastating losses (Dominic Moore’s wife, St. Louis’s mother) and personal triumphs (Alec Martinez the Hero, who was given less ice-time than a top 6 forward in the regular season; Willie Mitchell, whose career was nearly over a year ago).

The Kings were such heavy favorites going into the Finals we lost sight of the fact that this was a battle between a 5- and 6-seed. Both teams cut an arduous, impressive path to the Finals, and we were treated to four great, 1-goal games. The closing overtime periods of the 2014 Stanley Cup Playoffs were some of its most exciting – posts rang, Quick and Lundqvist were unbelievable, and Martinez’s goal, celebration, and humble interview were hockey at its purest.

Jonathan Quick, the placeholder for Jonathan Bernier four years ago, made a trio of overtime saves I will never forget. That toe save on a deflected shot, that blocker save through opaque traffic, and again, the blocker, on an excellent wrister from Chris Kreider. And then there’s Brian Boyle’s goal to put the Rangers ahead 2-1; in a flash of dexterity and speed, he gained the outside, then went top-shelf across his body. If you’ve played the game, you know how hard that shot is to pull off. Boyle, the gargantuan former King, the Blueshirts’ 4th-liner. So many players. So many stories. Underdogs all.

If we can just experience the game this way, how awesome it all is, I don’t think we’ll ever lose.


At the end, I couldn’t help but think how far we’ve come from hockey’s roots: that the champions are from Los Angeles, their best player is from Slovenia, the game-winner was score by a guy named Martinez, and they got here combining the best of scouting and analytics. Sometimes, when we live through major historical shifts, we don’t realize how important they are until much later. Even today, when I ask people I know who’ve lived through 1968, I’m struck by how difficult it is for them to attach themselves to the significant events of the time. It’s like they can’t see how they’re connected to these epochal giants as they stomped over the Earth.

So, can you see it?