Also, a big thank you goes to the readers and commenters, who make this a fun and lively site. A special thank you also goes to Ed Anthony, who edits the Community Site. We had some very good contributions this season. Keep 'em coming.
It was another big year for ANS. It has become even more of a full-featured site. It started with the blog, and then came the win probability graphs. Last year, we added advanced stat pages for teams and players. Now this year, we've added two major features: the advanced stat box scores and some really interesting visualizations.
Like I did for 2009 and 2010, I'll highlight some of the best posts and features of the year. Keep in mind this list is only a small fraction of the 276 posts of 2011. We'll start last January and go in chronological order.
Billy Cundiff was a kickoff machine last season, and all those touchbacks had the equivalent effect of 20 sacks.
I contributed an article to Slate about how the new OT format changed the equation when it comes to onside kicks.
Foreshadowing the visualization feature, I plotted team Success Rates and used it to develop a strategy for defeating mighty NE. Here's more on plotting SR. Note that running and passing correlate positively on either side of the ball, but team offense and team defense correlate negatively.
Carson somehow managed to merge literary criticism with football analysis. And it worked.
ESPN The Magazine featured ANS prominently in a great article on win probability--the 'killer stat'.
Great offenses > great defenses.
We finished up in January by taking a look at the positive and negative components of defensive WPA.
Beginning with last year's Super Bowl, I upgraded the server capabilities around here and was able to offer a WP graph widget that any website could freely add. Since then it's been upgraded. And thankfully, no more server crashes!
Here's the analysis of SB 45, in which I suggested it would have been better for GB to miss than make a FG.
This was one of my favorites of the year. It was a no-idea-is-a-bad-idea free-for-all on how to change professional football and make it more interesting. Don't miss the comments with many more great ideas.
John Morgan, Seahawks writer extraordinaire, joined ANS in March and contributed some analysis on the most exciting games of 2010. John also shared some thoughts on how to improve statistical analysis.
In the run up to the draft, I took a look at each team's needs according to their Expected Points Added generated at each position. And here's defense.
April wrapped up with a look at the demise of the first-round running back.
May only saw one lonely post, announcing my self-imposed personal lockout from the NFL. It was my first extended break from almost daily attention to football in years, and it was well worth it. Read the post if only for the references to Michael Crichton's speech on "speculation" and to Rolf Dobelli's excellent essay on how corrosive the "news" treadmill can be. Here's a snippet from Dobelli:
“Out of the approximately 10,000 news stories you have read in the last 12 months, name one that - because you consumed it - allowed you to make a better decision about a serious matter affecting your life, your career, your business - compared to what you would have known if you hadn't swallowed that morsel of news.”While I was wondering the football analysis wilderness like Kwai Chang Caine, refining my razor sharp skills while fighting for justice, ESPN was hard at work combining Win Probability, Expected Points, and Air Yards into Total Quarterback Rating. I liked it, but over the season it came across as forced, as if the on-air folks were at gunpoint in a hostage video when referring to it. In retrospect, for mass-market consumption of advanced stats, I'd recommend sticking with concepts like AYPA or interception rate, or maybe even the old Success Rate rule of thumb from Hidden Game of Football (4+ yds on 1st down, half the yardage needed on 2nd, conversion on 3rd down). The simpler and more transparent you can make a stat, the more likely it will gain mass acceptance.
I returned from the desert in August, newly schooled in the way of the Tiger and the sign of the Dragon, and I immediately attacked football analysis with lethal clarity and purpose. My first post back was a devastating roundhouse kick to the back of conventional wisdom's fat, stupid head, titled Don't Pay CJ. ...By the way, how'd that work out? I haven't been paying attention...Wait, what's that? He had one of the worst seasons ever for a starting RB? You don't say.
By the end of the month, the WP graph pages received a major upgrade with the addition of advanced stat box scores for every game since 2000. To be completely honest, I'm tinkering with football stats almost every day, and very little of what I do really strikes me as very cool any more. Sometimes I wonder, if this weren't my own site, would I be spending my time reading it? The box scores are one of those things I frankly love. Every Sunday, I find myself eager to see the box scores after each wave of games.
August wrapped up with some detailed research on how QB performance is related to age. I summed it up by writing, "Successful, established QBs will generally continue to be successful until one day they're not. We won't see it coming. But of course, everyone will pretend they did."
September began with the Internet echo chamber wondering how many wins Peyton Manning is really worth to his team. WPA has its limitations, but it seems like it was custom made for exactly a question like this.
Carson Cistulli returned with his Weekly League and Notes feature. We also welcomed Keith Goldner and his very cool Markov drive model to ANS.
Success Rate never really made sense for defenders. We can't criticize a safety for making a tackle 10 yards downfield when his only other alternative is to allow a touchdown. So I introduced Success Count for defenders as a replacement.
I asked When Will Ray Lewis Slow Down? Sadly, it turns out the answer might turn out to be week 10 of the 2011 season.
I continued contributing to both the Washington Post and New York Times (online editions) in 2011. The Post also licensed the WP model to create some of their own visualizations but they were never able to make their data feed play nice.
Marc took a look at Cam Newton's early season success.
Longtime reader Jeff Clarke asked why would any team kick deep from the 50 following a penalty?
Jack showed he has a real knack for graphs/visualizations in his first post, along with a love of Kafka.
Go for it on 4th and goal from 15 down by 3 in the 4th quarter? Yup.
A preemptive response to the all the harping the efficiency model gets during the season. Also, check out how out to lunch the very first rankings of the season were in a couple cases--WAS, SF, and ATL in particular. Otherwise, not bad after only 3 weeks.
It's conventional wisdom that you never want to take points off the board. Problem is, it's wrong.
In October I did a podcast with Carson for FanGraphs.
The problem with all the power rankings out there is that they are far too influenced by wins to-date.
The Carson Palmer trade was costly for OAK and a big win for CIN. As it stands today, Palmer is 17th in WPA/G, EPA/P, and AYPA each, which happens to be right where his career numbers predicted him to be. Was it worth it for OAK? It might seem like it does to OAK fans for a few days if they make the playoffs, but 2 first round picks?
One of the most widely read posts of the year was this report on the game probability model's accuracy and a discussion of the issues surrounding it.
Keith crunched the numbers on Cutler's struggles in the pocket.
I added offensive line stats to the advanced box scores.
I looked at how much better Sebastian Janikowski really is at long distance FGs.
This was not a very widely read post at all, but I thought it was interesting: The worst single game WPAs on record.
Jack compared Aaron Rodgers and Matt Ryan following their teams' matchup.
Josh Katz has been doing an amazing job with the playoff probabilities all season. He explains everything really well and cuts right to the critical upcoming games. The best part of each of his posts is the high-leverage games of the week. Here's a great example from week 14.
Zach Sanders' pithy write ups for the weekly efficiency rankings are always entertaining. Here is a good example from week 11.
Stats often do measure the intangibles.
November is really when the Tebow-mania took off. Here are our posts on the subject.
-Another Argument for Tebow
-Detroit at Tim Tebow
-Introducing the Trent Dilfer Club
-How is Tebow Winning?
-Tebow by Quarter
-Tebow (Titled simply to test the Internet frenzy that Tebow had become.)
-Narrative-Chasing Intuition Monkeys (You'll have to click the link to understand.)
Even some posts on completely different subjects got derailed onto Tebow.
The concept of team identity is one of my pet peeves.
I took a look at which receivers were target most often in interceptions.
November is when the visualizations feature became cemented here at the site. Like I mentioned above, there are some features around here I personally enjoy more than others. The visualizations are one of those that I am always eager to check out as soon as they're updated each week. I'm continuing to add some really cool stuff, specifically for player stats, which will be officially unveiled soon.
It happened to be a good weekend to unveil the Fourthdownulator feature, as Mike Smith decided to go for it on 4th and inches on his own side of the field in OT against NO. This analysis became the most popular post of the season. Almost as popular was a post about how to talk to a 4th down skeptic, which was distilled from this conversation I had on Fox Sports radio. The incident also prompted this analysis on RB dives vs. QB sneaks.
I added a game-by-game stat page for all offensive skill position players. Each game will also link you back to the WP graph.
In one of my first foray's into salary and salary cap analysis, I crunched the numbers on how much money the Redskins put into their notoriously thin offensive line.
A discussion on the purposes of punishment, in light of Ndomukong Suh's suspension.
By the way, in November we hit an ANS record for most posts in a month, averaging over 2 per day.
At the Community Site Michael Beuoy derived the team rankings implied by the weekly point spreads.
Early in December, Jason Garrett was blamed for icing his own kicker. His real mistake was not taking the timeout much sooner. I think he fell victim to the mythical siren song of field goal range.
Player stat visualizations were added in December. I'm still not sure which ones I'll keep. Also, based on a suggestion from Chase Stuart, I added a franchise performance visualization.
I thought it might be useful to filter some of the randomness out of EPA. I simply capped any single play at an EPA of 2 points and threw out trash time plays to create Expected Points-Experimental (EPX). It got some mixed reviews and needs some work and further tinkering, but I think the idea has potential.
In a big step forward, I started crunching salary numbers in earnest to estimate the fair market value of safety performance. I then used that analysis to estimate a value for impending free agent LaRon Landry.
Keith analyzed KC's schizophrenic decisions on 4th down.
I computed just how important playoff seeding is.
Merry Christmas!
Drew Brees broke Dan Marino's single-season total passing yards record, and this was my take.
I pitched-in on an NFL Network/Freakonomics feature on Incentives.
And to round out the year, the All-WPA teams were announced: AFC Offense, NFC Offense, AFC Defense, NFC Defense.
Thanks for a great twenty-eleven, and we'll see you in twenty-twelve.
Advanced NFL Stats: The way of the tiger. The sign of the dragon. How is it that you do not hear these things?