I'm excited to introduce two new features of the stat visualizations, and I think you'll find them very clever, fun, and useful. They are by far my favorites.
The first new graph, titled "Nth Best", plots each selected QB's career in order of his best through worst seasons. As opposed to a straight chronological plot by year, this presentation provides a very continuous and smooth curve, and it allows an easy comparison of the total career performance of each QB. A chronological plot produces a much more erratic and confusing display of peaks and valleys, making comparisons difficult.
The horizontal axis represents the Nth best season of each selected QB. The vertical axis represents the total EPA for each season. As you view from left to right, the graph will always appear to decline, but don't interpret this to mean the QB's performance is trending downward. It's just that the seasons are sorted from best to worst. Also keep in mind that graph plots total season EPA and not EPA per game or per play, so some of the worst seasons of a QB will be those marred by injury or prior to when he became a starter.
The second new graph, titled "Career WPA," plots cumulative career WPA. In short, it illustrates the running total WPA of a QB's career. The horizontal axis represents each season of a QB's career in chronological sequence--his 1st season, 2nd season, and so on. As each QB progresses through his career, he'll accumulate WPA (if he's good). Like the first graph, this usually produces a nice smooth curve, and it's a great way to compare QB careers.
If either of these styles of graphs sound familiar, it's probably because they're inspired by the WAR Graphs (Wins Above Replacement) at FanGraphs. Ok, not inspired, more like copied. I hope David Appelman accepts my flattery in lieu of licensing fees!
There are a few drawbacks. I wish I could go back much further than 2000, as these kinds of graph are obviously best suited for historical comparisons. Even for modern QBs like Peyton Manning or Kurt Warner, we don't get to see their earliest seasons. Still, even with that limitation we have lots of QBs to compare. It's still fun and useful to compare a 2011 Porsche Carerra with a C6 Corvette, even if we can't compare a Porsche 356 with an original Corvette. And at the very least the path has been paved for when digital data for earlier seasons becomes available.
Also, in the Nth Best graph, I can't use custom colors of the QB's actual team. For QBs who have switched teams, the visualization software splits the player's career line into separate lines, which scrambles the seasons out of order.
- Home
- Unlabelled
- QB Career Comparison Visualizations
QB Career Comparison Visualizations
By
Brian Burke
published on 1/03/2012
Subscribe to:
Post Comments (Atom)
Just spent 30 mins playing with the Career WPA graph - Great tool. Had lots of fun playing around and comparing the QBs by generation (Brees/Brady/P.Manning, Rivers/Eli, Flacco/Ryan,Sanchez) was very insightful. Congratulations for this !
Just for kicks I wanted to see Matt Flynn's career average numbers after that Lions game... but I guess those games where he came in and took a knee are hurting his stats.
Is Flynn even in there? I think I put in a filter for the 40 players with the most game appearances in any one yr.
One other thing I still need to fix: When a guy changes numbers, his ID changes and his career splits into two parts. I'll have to go into the data to fix that.
Any chance for a way to select different stats for Nth-best, or is that tough to do in that software? Great stuff.
Not hard to do. There just has to be a separate tab for it. Ideally there would be an option to pick any of the stats to chart, but the viz software (I believe) doesn't have that feature. I'm going to weed out some of the other EPA by game/by year tabs, so there will be room for some additional charts. Alternatively, I could just have one composite chart of graphs of every stat, but that's really inelegant.
If anyone out there is a Tableau Jedi, drop me a line.
This is brilliant, thanks Brian.
I just did EPA by Year for the three big QBs taken in the '04 draft (Manning, Rivers, Roethlisberger), it's amazing how similar their careers have been.
When you look at the career WPA added tool, looks like Roethlisberger is a win or two ahead, and all three are on a Brady-like career track so far.
It took about two minutes of playing with the Career WPA graphs to realize that Peyton Manning has been far and away the best QB of the last ten years. Plot him against Brady, Brees, and Rodgers to see what I'm talking about. I've never seen such compelling evidence in the Manning v Brady debate. Awesome tool Brian thanks so much!
A couple caveats on the Manning vs Brady issue. One thing is that the stats and charts start at 2000, after Manning had 2 yrs of experience. Notice how Brady's roughest years were his earliest. Also, Brady virtually missed an entire year in '08 due to injury. But, the season appears in the chart because he did have 1 game appearance.
Second, and perhaps more importantly, is that WPA favors a really good QB with a bad defense. If you gave IND a better defense over the past decade, Manning would have had fewer opportunities for big WPA drives. Some of his +0.75 WPA game-winning drives would have been trash-time TDs for little WPA.
How much did Brady's better defense "hurt" him in terms of WPA compared to Manning? Well, we already know the answer--that's what EPA tells us. And looking at the career EPA chart, it's still pretty clear Manning was consistently superior. Brady does have the highest single season EPA though, actually the two highest between the two QBs, one of which is this season.
The book is not closed on Manning vs Brady. Given the way this year has played out, it would be easy for Brady to maintain his high level for a few years longer than Manning.
If you go by who is further to the left on the chart in career WPA to determine "greatness", then plot Matty Ice against anyone. Forget Brady or Manning, he's better than/ahead of everyone! That's where the charts get maddening because people can use them to make ridiculous claims. But people are going to continue to do that regardless of evidence anyway I guess. Check out Favre's career WPA from 2000 on. That's not a very steep slope on that man's chart. My grandma could navigate that in her wheelchair.