- Home Archives for September 2012
The Redskins pass defense actually may be worse than it appears, for a couple reasons. First, the pass defense deficiencies have been obscured by their relatively high interception rate. You might think that that’s a good thing, and it is. Or I should say, it has been. Rarely do defensive interception rates persist throughout the season. The Redskins current interception rate is 3.5%, significantly higher than the league average of 2.7%, but it’s bound to regress. Interceptions are relatively rare and notoriously random, particularly from the defense’s perspective. They’re thrown by a quarterback more than they are taken by a defense.
Second, the Redskins’ opponents so far have been well below average when facing teams other than the Redskins...
As you may have noticed by the chart at the top of the front page, the visualizations unveiled late last season are now up and running for 2012. As a refresher, there are four sets of visualizations. Each set can always be accessed through the Graphs|Visualizations menu. Each visualization is interactive. You can select seasons, players, teams, weeks, etc. Hovering over a team symbol or player will show enhanced details.
The team stat viz plots each team's performance on an x-y graph. Good defense is up, and good offense is to the right. So teams on the bottom left are your thoroughly bad teams. There are tabs to see Expected Points, Win Probability Added, and Success Rate. Plus, there's a tab that just looks at team offense and one that just looks at team defense. On those two plots, running is measured on the vertical axis and passing is measured on the horizontal axis. Like the other plots, up and right are good. Below each main plot you'll see a chart of week-to-week performance. You can choose any season since 2000 to look at. You can also narrow down the span of weeks to examine. For example, if you want to look at the 2011 Texans with only T.J. Yates starting at QB, you can move the left slider from week 1 to week 13. Or if you only want to see regular season performance move the right slider to week 17.
The next visualization is for franchise EPA since 2000. You can select a team and see how they've trended in offensive EPA and defensive EPA in the past 12 years. For example, you can see the Giants' best teams over the past several years were not their 2007 or 2011 championship squads.
Game probabilities for week 4 are up at the New York Times' Fifth Down.
If you happen to be a new reader, it might be worth your while to go through my archive there. Most weeks feature a lead-in discussing or applying one of the many concepts written about in these spaces.
In general, I don't like penalties at all. It breaks up the flow of the game, and most of them are so subjective that fans of the penalized teams always feel aggrieved. And even when penalties favor your team, there's something about it that feels cheap. Plus, with all the complex rules and reviews, the NFL seems like it's run by a bunch of lawyers. Wait, what?...Oh. It all makes sense now...)
Here's what I think the problem is. The NFL has a yardage inflation problem. Offenses fly up and down the field with ease in the modern NFL. A generation or two ago, a 10 or 15 yd penalty was a more severe blow. These days, not so much.
So the level of pain might need to be increased. For example, holding might need to be a 15 yd penalty now. Pass interference might need to include an extra 10 yds. Perhaps personal fouls should be 20 or 25 yds. Helmet-to-helmet violations might need to be 30 yds. Yes, those are stiff consequences, but we'd ultimately see far fewer flags, and the game would flow so much better. On net there might be fewer total penalty yards despite the stiffer yardage consequences, because players would be more reluctant to hold, interfere, etc...
Think of it this way: In the 1970s a speeding ticket might have cost about $30. If it were still just 30 bucks in 2012, people would be far more reckless on the road because of inflation. A $30 fine just isn't as painful in 2012 as it was in 1972. The NFL has a yard-inflation problem, so just as the states have to raise the price of a speeding ticket to maintain the same level of deterrence, the NFL needs to raise the pain of rule violations.
Just thinking out loud.
It's time to launch the first rankings of the season. The rankings and prediction model was redone prior to last season, but just like always, it's a team-efficiency logistic regression model. It's based on passing, running, turnover, and penalty efficiency. Since last season running is represented by Success Rate (SR) rather than Yards Per Carry (YPC). SR correlates far better than YPC with winning games. YPC is too susceptible to a handful of relatively rare break-away runs and wrongly penalizes successful plays in short yardage situations. I believe the revised model better reflects the true inner workings of the sport.
There are always new readers each year, so here is a quick and dirty refresher on how the model works. (Most of this write-up is taken from last season's first rankings post.) A logistic regression is fed net YPA, run SR, and interception rates on both offense and defense, plus offensive fumble rate. Team penalty rates (penalty yds per play) and home field advantage are also included. These particular aspects are selected because they are predictive of future outcomes, not because they explain past wins. This is a distinction overlooked by most experts and even other stats-oriented sites.
There's a distinction between the WP model’s empirical methodology and its automatic output without any intervention or input from a human. When I do a detailed analysis for any specific play, I have the luxury of time and logic to dig directly into the data. The “auto” model that spits out WP estimates without any human input is based on lots of assumptions and interpolation on top of extrapolation etc. There are literally billions of combinations of game states (yd lines, downs, to go distances, seconds remaining, score difference, time outs). It’s just a matter of how much time I can put into coding the calculator to handle special cases like “a team's very last desperation play.”
With all the attention on that final play in the GB-SEA game, I thought it would be useful to look at Hail Mary success rates.
To put last night's Russell Wilson-to-Golden Tate Hail Mary That Shouldn't Have Been in perspective, consider these sets of plays from Week 3 that add up to +0.99 WPA, the total awarded for the Seahawks' game-winning play.
- JAX @ IND: 1st and 10, 0:56 Q4, 16-17: B. Gabbert pass short right to C. Shorts for 80 yards, TOUCHDOWN, 0.73 WPA
- NE @ BAL: 3rd and 9, 2:28 Q4, 30-28: T. Brady pass short right intended for W. Welker INTERCEPTED by L. Webb at NE 40. L. Webb to NE 40 for no gain. PENALTY on L.Webb Illegal Contact 5 yards enforced at NE 32. 0.22 WPA
- BUF @ CLE: 2nd and 6, Q2 0-14: T. Richardson right guard for 6 yards TOUCHDOWN. 0.04 WPA
- DET @ TEN: 2nd and 10, 0:06 Q4, 34-41: Sh. Hill pass deep middle to T. Young for 46 yards TOUCHDOWN. 0.42 WPA
- KC @ NO: 4th and 5, 2:26 Q4, 21-24: M. Cassel pass short left to D. Bowe to NO 39 for 7 yards. 0.27 WPA
- NYJ @ MIA: 3rd and 7, 10:53 Q3, 10-10: M. Sanchez pass short right intended for S. Hill INTERCEPTED by C. Clemons at MIA -5. Touchback. 0.20 WPA
- NE @ BAL: 2nd and 15, 6:43 Q3, J. Flacco pass deep left to T. Smith pushed ob at NE 47 for 38 yards. 0.10 WPA
- JAX @ IND: 1st and 10, 1:33 Q4, 14-16: A. Luck pass short left to D. Brown pushed ob at JAX 28 for 39 yards. 0.39 WPA.
- NE @ BAL: 2nd and 7, 1:26 Q4, 28-30: J. Flacco pass deep right to D. Pitta pushed ob at NE 35 for 17 yards. 0.34 WPA
- DET @ TEN: 1st and 15, 3:23 Q4, 27-27: J. Locker pass deep left to N. Washington for 71 yards TOUCHDOWN. 0.19 WPA
- HOU @ DEN: 3rd and 6, 3:05 Q4, 18-31: P. Manning pass short left to J. Dreessen for 6 yards TOUCHDOWN. 0.07 WPA.
On a crazy Sunday with upsets out the wazoo, the most surprising win had to be Christian Ponder and his lowly Vikings taking down the San Francisco 49ers. Minnesota outplayed the Niners from the very start of the game; their opening drive lasted 7 minutes and 40 seconds and went 82 yards over 16 plays for the first score of the game.
The Cardinals dominated the Eagles all game long, but after back-to-back field goals to bring the game within 18 points (why do teams still kick field goals down 21 in the third or fourth quarter?), Arizona needed to seal the win. They did so with a 13-play, six-and-a-half minute drive (including two Eagles timeouts), which resulted in a field goal.
Using our Markov model, let's look at these two game-changing drives.
In week 1 the average HFA was 3.2 net points, and in week 2 it was 8.7 net points. As I write, Week 3's average is -1.9 net points through 14 games. Week 1 was obviously well within the long term trend of 2.5 net points for the home team, and Week 3 tilted slightly toward the visitors. But week 2 showed a large advantage for home teams. We know there is week-to-week variance in net scores for home teams, but how far from ordinary is 8.7 net points for the home team in a single week?
Since 2000 the standard deviation for weekly average HFA is 4.0 points, and the average is 2.4 points, which means week 2's 8.7 is not 2 standard deviations from the mean (p=0.40). Week 2 featured only the 15th largest HFA in the past 12+ years. The regular refs were on the field for the 14 weeks with larger HFAs.
The chart below shows the distribution of weekly average HFA in net points for the home team.
The Lions almost accidentally beat the Titans. In the new OT format, the Lions were in the position where a FG ties the game and continues it, but a TD ends the game with a win. Head coach Jim Schwartz said he intended to try to draw the Titans off side on the critical 4th and 1, but the ball was snapped accidentally.
DET faced a 4th and 1 on the TEN 6. Let's look at the FG option first. A FG ties the game, but gives the ball to TEN in a sudden death format, which means a 0.43 WP (they'd win 43% of the time). FG attempts are good 97% of the time, so the FG option is worth a total of:
0.97 * 0.43 = 0.42 WP
Conversions on 4th and 1 are good 68% of the time. A 1st and goal at the 5 results in a TD 60% of the time and result in a FG attempt 28% of the time. The math gets a little tricky here because the TD has a WP of 1, but the FG gives us a 0.43 WP, very similar to what we calculated above.
A successful conversion would mean a total WP of:
You wouldn't know by reading this study. It sets up a multiple regression model that uses a kitchen-sink approach to estimating the time needed in the end-game to get to the 35-yd line, commonly accepted as FG range. It uses QB rating, time remaining at the start of the drive, number of all-pro players on the offense, time outs remaining, starting field position, home field advantage, and whether the 2-minute warning is still available. The dependent variable is the time taken to reach the 35.
There are numerous fatal problems with this study. First, the model assumes linearity of the effects of predictor variables. I can tell you from my intimate familiarity with the variables involved that they are not linear at all. The model also assumes a normally distributed outcome variable, which is not investigated, and I doubt could be possible because games are bounded by the expiration of regulation time.
The study uses 3 seasons of data, which only yields 92 example situations to analyze.
The authors find enormous multi-collinearity problems with their model, and I'm not surprised. The model specification looks like this:
time taken = constant + field position + time outs+ ...a bunch of other stuff... + game time at drive start + game time when reached 35
But doesn't time taken = time at start - time when 35 reached? Of course. You can't have a regression model where the dependent variable is always the exact sum of two of the independent variables. The model's r-squared is 0.97, because it's one giant tautology.
Former Broncos GM Ted Sundquist has been a longtime fan of football analytics and has been a friend of Advanced NFL Stats for several years. Despite being an Air Force grad, Ted's a great guy. Ted has peppered me with questions from time to time and has always been supportive of the larger stats movement. He's getting his own site (TheFootballEducator.com) off the ground these days, so I'm pleased to host a guest-post.
“Statistics are for Losers”? Only for NFL teams that don’t use them
By Ted Sundquist – General Manager, Denver Broncos Football Club (2002-2008)
Most might assume that for a sport so valued and valuable in the world of professional athletics that each and every aspect of its decision making processes would be “cutting edge”. Game planning and personnel acquisition would be prepared, executed, and analyzed with the top technological tools of the time. With so many sources of information and data across the internet, the game of professional football is sliced, diced, examined, scrutinized, evaluated, probed and otherwise dissected by every inch of the playing field, and every down & distance measured.
The game probabilities will begin next week at the New York Time's Fifth Down. I've been getting lots of email from antsy readers, so I thought a quick note here would answer them all. Team efficiency rankings will begin next week also. Because both the game predictions and rankings rely on empirical data, they need a minimum number of weeks of stats before they become meaningful or reliable.
This week's article at the Post looks at whether the numbers favored the Redskins' 62-yd FG attempt or going for a conversion after Josh Morgan's insanely stupid unsportsmanlike penalty call.
After every game, each NFL coaching staff evaluates the performance of players on each and every play. Coaches also grade themselves on play selection and other decisions, such as challenges, clock management and fourth down decisions. The Redskins staff should take a hard look at the final decision of Sunday’s game to attempt a 62-yard field goal rather than try to convert a 4th and 16...
...There isn’t the luxury of time and a computer full of stats on the sideline during a game, so it’s hard to fault the Redskins’ coaches. The value of analyses like this one is not to prove there is always a right or wrong answer, but to recalibrate a coach’s intuition so that he’s armed with the best information next time.
Mike Shanahan disagrees with me, but he is wrong. Already the post has hundreds more comments than their typical post gets, so it seems it struck a nerve. The comments there have come a long way from a couple years ago, but a quick scan of them tells me most of them miss the point.
By the way, these posts now appear in the print version of the Sunday Post each week.
For the most part, the Jets did just that on Sunday. Pittsburgh's running backs were completely ineffective. Jonathan Dwyer needed 12 carries to get 28 yards; Isaac Redman needed the same number for 25 yards. The Jets defense held Pittsburgh to just a 34.5 percent success rate on runs.
The Steelers did hit on one notable deep ball -- an incredible catch by Mike Wallace for a 37-yard touchdown. But Roethlisberger's Steelers only attempted four other deep passes, resulting in two completions for 43 yards. One other deep attempt saw an intentional grounding penalty. This is an outcome, I would think, most defensive coordinators would walk away from with satisfaction.
But Mike Pettine and Rex Ryan's defense left Heinz Field drubbed for 27 points by the Steelers offense. Pittsburgh whittled down the Jets with an array of successful short passes. Observe:
Andrew Luck got his first of what will likely be many career wins on Sunday, with the Colts taking down the Minnesota Vikings on a game-clinching field goal with eight seconds to play. Luck played extremely well throwing for 224 yards on 31 attempts and adding two scores; he notched a formidable +12.9 EPA and +0.67 WPA. What killed the Vikings, however, was not just Luck's play, but penalties. They were penalized 11 times for 105 yards, which may not sound terrible, but they continuously allowed the Colts to continue drives. The biggest of these follies came on the Colts first drive of the second half.
Up 17-6, the Colts took the ball from their own 20 and would ultimately kick a 45-yard field goal. Below we can see the development of the drive using our Markov model.
This week's post at the Post looks at the very un-Shanahanian play calling in Sunday's Redskins' upset of the Saints.
Usually on fourth and short, offenses send in goal line packages of players and plays because the situations are similar. Both scenarios call for just a short gain. But goal line situations are vastly different because additional yardage beyond the amount required is useless. The yardage beyond what is required on a fourth down in the open field is extremely valuable. Plus, forcing defenses to respect the possibility of a deep pass and defend the entire field makes it easier to convert on future fourth downs...
Additionally, these developments change the strategy equation in the end game. When offenses are in what they call the four-minute drill, trying to run out the clock with a single-score lead, they can no longer count on a run-only strategy. We used to talk about the two-minute drill, in which heroic quarterbacks lead their teams on a winning march down the field. Now, two minutes sounds like a quaint eternity. Offenses need only seconds to steal a win. In the Jaguars-Vikings game on Sunday, the teams needed only 1:18 to manufacture not one but two full, lead-changing drives totaling 10 plays and 11 points.
In all the excitement of the new season I forgot to make a post linking to last week's article on how preseason statistics for rookie starting QBs are not predictive of first year performance.
There was an unfamiliar air of uncertainty surrounding Peyton Manning's debut as a Denver Bronco on Sunday night. From 1999 through 2010, Manning defined the quarterback position, throwing for over 50,000 yards with an adjusted net yards per passing attempt (per Pro-Football-Reference) of 7.24. Only Philip Rivers's 7.26 mark bests Manning, and Rivers did it in 112 fewer starts and with a higher percentage of his play coming in the recent League of the Quarterback.
For the first time since Manning opened 2010 with the Colts, we really just didn't know what to expect. Sure, "he's Peyton Manning," but multiple neck surgeries have the potential to derail any sports career, uch less one as fragile as NFL quarterback. There was talk he had become a reverse Zoolander -- instead of losing his ability to turn left, rumors mounted that Manning had lost the ability to throw to his right.
In that case, Sunday night's performance against the Steelers was Manning's Magnum.
In an analysis of the previous overtime format, I noted the following:
In 1974, the league FG% was 60.6%. This year, it was 84.5%. And that even masks how much kickers have truly improved. In 1974, 36% of all FG attempts were from 40 yards or beyond. In 2008, the figure was 41%. These days, teams aren’t looking to get inside the 25 for a field goal attempt, they’re just hoping to get inside the 40. Getting a quick score in overtime has become a far easier proposition. Field goals have gradually warped NFL football. In 1974, there were 3.0 FG attempts each game compared to 3.9 in 2008, a 30% increase.
RGIII was the only rookie quarterback to win on Sunday, when his Redskins took down the Saints 40-32. Brees was pressured heavily all game and while it's extremely difficult to quantify the value of coaching, people are wondering what kind of effect the loss of Sean Payton will mean.
One immediate impact is strategic decision-making. Sean Payton is known for his risky decisions. And by that, I mean he is known for those decisions that are publicly perceived as "risky," even if they are the statistically correct decision. Last year, the Saints were the No. 8 team in terms of win probability forfeited on 4th down (meaning they were in the top quarter of the league in 4th-down decision-making). So, what did we learn on Sunday?
With just about 2:00 to play in the 3rd quarter, down 30-14 already, the Saints were faced with a 4th-and-goal from the 3-yard line. Below you can see the results from the 4th-Down Calculator:
Take a look at Adjusted Net Yards Per Attempt (courtesy of PFR), which with just one number incorporates passing efficiency, interceptions, and sacks. Since the dawn of the modern passing era, passing has become steadily more lucrative. But since 2004, the rate of increase in average ANY/A has accelerated. The 2011 season featured the most successful passing game ever.
For context, compare the graph above with the next one. This shows the same trend but for rushing yards per carry. There is a very shallow increasing trend since a trough in the mid 1990s, but it pales in comparison. The jump in net passing last year alone is larger than the increase in rushing over the entire period. (I've kept the scales of both graphs identical for a pure comparison.)
It's a common practice in the NFL that when your team is backed up against your own goal line, you run the ball to "give some breathing room." Often you'll see either a QB sneak or a handoff right up the middle. In fact, since 2000, teams have run the ball on 1st down from inside their own 3-yard line on 72.4% of all occasions.
We know that passing is the much more efficient option in today's NFL, so should teams actually be running the ball, using up one of their precious downs? A quarterback still has around 10 yards to work with when backed up against his own goal line, whereas a running back needs to ensure positive yardage -- plus he starts 5-7 yards behind the line-of-scrimmage on most occasions.
We're going to look at the track record in the league on passing and rushing plays using both EPA per play and safety percentage (% of plays on which a safety was recorded).