Of Passer Ratings and Power Plays: The Sports Data Viz Digest No. 1
Data visualization presents a rich opportunity across a variety of fields, perhaps none more so than the wide world of sports. Data is increasingly becoming an inextricable part of the sports experience from a fan’s perspective, but one look back towards history shows that it always has been. Perhaps that’s why, in response, the Sports Data Viz channel on the DVS slack has already been such a great source of inspiration and creativity. It’s provided us with exciting illustrations that challenge new propositions and show us ideas that we may never have considered before!
Being an avid sports fan is one thing, but I believe some of these pieces, and the future pieces to come, will attract interests from all avenues of life. The magic of this is that we can relate to each of these pieces in our own unique ways and take away our own unique learnings, maybe because we play the sport, watch as a fan, or are simply intrigued by a beautiful win probability visualization!
As has been the case from the start, it’s been really encouraging to see the contributions from all different sports. We’ve seen highly contrasted styles of analysis too, from granular charting of player activity to economics and team transactions. So let’s roll:
In this first viz, Peter Beshai brings us colorful and clean visualizations on NBA shooting percentages for players and teams. He has quantified both the change over time with graphs showing 2-point and 3-point percentages over an entire player’s career.
The above graph breaks a player’s shots into multiple buckets (each color representing a different bucket) based on shooting distance. The larger the area, the larger the percentage of shots that were taken in that range. The X axis shows the total number of shots (scale by the thousands). A lot of times in sports, analysis tends to be results-oriented, but it’s more instructive to step back and think about how those results came to be, which is what Shotline allows us to do, breaking down where players’ points are coming from.
Check it out here:
Next up, designer Paul Buffa has created two different graphs to display his change in heart rate on a run at specific locations. The side-by-side illustration of a scatterplot and heat map shows the viewer exactly where and when the heart rate changed based on elevation and accumulated distance. Paul’s pairwise matrix uses density contours to highlight trends based on the different variables. The combination of the two visualizations speaks volumes and tells a more complete story.
The additional context a heat map adds to the pairwise matrix uncovers deeper “relationships between elevation and heart rate.” I myself am not much of a runner, but for those out there, heart rate is a tricky vital sign to maintain. If you run similar routes or paths, this visualization would give you incredible insight: where your heart rate peaks and falls at what exact points on your run. It allows runners to better control their breathing and pace, based on the estimated heart rate.
Buffa provides an interactive version here: https://observablehq.com/@pstuffa/density-contour-matrix-for-geospatial-datasets.
Staying with the theme of tracking an athlete’s activity, designer Tom Worville has created a couple of soccer visualizations, specifically in this instance for Mesut Ozil of Arsenal (part of the English Premier League). Whereas Paul Buffa was previously helping track a runner’s route, Worville here is tracking the specific parts of the court, or field in this case, a player is most effective in!
The first graphic above denotes the change in Ozil’s total touches, stratified by area of the field. It is certainly interesting to note that Ozil’s activity declined by about 13 touches per 90 between seasons in the middle of the attacking third, ostensibly an area of the field that he should be dominating. This change is highlighted in even greater detail in the passing maps below:
The first map above shows the start points for Ozil’s most dangerous passes during his 2017-18 season. Compared to the second graphic below, which shows his most dangerous passes in the following season, the viewer can see the passes originated more from the sides, and deeper into the field, as compared to more directly through the middle in 2017-18. These graphics articulate the clear dip in effectiveness for Ozil’s passes and touches with minimal distractions.
We have a hockey visualization! Of all the sports out there, hockey intrigues me personally because I possess very little knowledge on the mechanics of the game. It is such a physically intense sport and so incredibly fun to watch live. Unfortunately, I’m from Atlanta where our hockey team left us almost a decade ago. That won’t stop me from taking in Bill Tran’s visualization on the Calgary Flames’ shooting last season, however.
The Flames scored an incredible 289 goals last year, second-most goals in all of the NHL. Tran broke down the goals by each specific player and distance. Unsurprisingly, he showed that the players at the top of the chart (who scored the most) tended to score most frequently from shorter distances. Still, players like Elias Lindholm, however, had a more varied distribution. But that’s just the forwards. How’d the defensemen do when called upon to contribute to the scoring fun?
The six Calgary defensive players scored most of their goals from beyond 25 feet or more! The shot attempts are skewed more to the right between the 35 to 70 foot range. In all likelihood, this is because defensive players tend to stay further back from the goal in general and as such are taking more speculative attempts.
This data could be of the utmost importance to not only the Calgary team but their opponents as well. Let us imagine both the Calgary team and opposing hockey teams have this graphic on hand. For Calgary, they could start confusing opponents’ schemes by either having defenders start to move up and take closer shots as well as having offensive players shoot from farther away! Likewise, opponents could start planning defensive schemes to anticipate who might be taking the shots and when they might be occurring. The predictive nature of this graphic ties it in with the adjacent ideas presented previously.
Check it out here:
Now we transition from patterns of play to economics of the game. This next piece is more of a survey on the way social media influences football fan interest. Vivid Seats and Opendorse partnered together to reveal the most popular NFL teams and players by United States county! Opendorse is a social publishing platform that allows over 7,000 athletes to share content on social. The viz was produced to explore trends in football teams and players based on location across the country.
The map is interactive, and the viewer is able to see, by county, which player is the most popular on the corresponding team. One would think that Odell Beckham Jr. or Rob Gronkowski are the most popular on social media, but Cleveland Browns wide receiver Jarvis Landry edges them out as the player who actually engages his fan base the most. Not to mention, America’s Team, the Dallas Cowboys, are also Oklahoma’s and New Mexico’s most favored team. In terms of pure land area, the Minnesota Vikings, Denver Broncos, and Dallas Cowboys cover the most territory for their respective fan bases.
While this piece takes on a bit of a different direction, it still contains one of the most widely used aspects of technology: social media!
Check it out here:
As an avid National Football League fan and fantasy football enthusiast, this next design was highly intriguing and engaging to me. Advaith Venkatakrishnan brought us a node network for the 2019 NFL free agency period, tracking virtually every piece of player movement between all 32 teams in the league. The nodes consist of all 32 team logos as squares and circles for player positions, while the edges represent the specific player movement between teams. A red edge means a player was leaving a team while a yellow edge means a player has joined a new team. A blue edge signifies a player staying on that team. The size of the node also corresponds to the size of the contract (i.e. larger nodes = more money).
Here is a quick snapshot of the network from a zoomed out perspective. Nodes on the outside represent players who are yet to get signed or retired, but still might have some money remaining on their respective contracts. This picture was taken with no filters being added, so every single player, no matter how small their contract was, is on this clip.
Some interesting insights are found through the use of the filter legend, which allows you to set limits for money as well as turn on/off each type of edge. For players who received deals $30 million and up, there were only three and all were quarterbacks: Russell Wilson, Ben Roethlisberger, and Carson Wentz (Wilson’s node was cut out but was at the bottom of the network). Advaith actually wrote about the design process behind his viz recently; you can also check out the app for yourself here:
And that just about wraps it up for now. With the NBA and NHL seasons kicking off now too, I can’t wait to see what other sports viz journeys we take!
