This will be the first year of the WPA's for defenders. Last season we only had stats for offensive skill players, but this year we'll be looking at +WPA and to some extent +EPA, both of which measure defensive play-making, to select the All-WPA defensive squads.
Keep in mind what these stats measure. This is not an attempt to value a player in terms of his overall ability or his future expected performance. This is just a list of the guys who made the most big plays when it mattered most.
The envelope please...
- Home Archives for 2010
The 2010 All-WPA Teams - Defense
Game Probabilities - Week 17
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I put blame where blame is due for the prediction model's under-performing season...squarely on the shoulders of Norv Turner.
Washington Post: Don't Blame McNabb
Today's post at the Washington Post's Redskins Insider looks at where the blame lies for the Redskins' offensive struggles this season.
The 2010 All-WPA Teams - Offense
Last year I introduced the first All-WPA team, which was a fun exercise. This season every player's WPA, EPA and other advanced stats have been updated all season long, so this post is little more than ceremonial. (And here are the mid-season awardees for offense and defense.) Still, I'm sure the players appreciate the official acknowledgment that comes with being recognized by an obscure statistic that only about 20,000 people have heard of and fewer even understand.
To be clear, these are the All-WPA teams, and are not necessarily the best players in the league. They are the players who actually played the best when it mattered most. It's a narrative statistic, highly sensitive to the circumstances of score and time. For those who are uninterested in WPA, feel free to sort the columns however you like. Click on the position headers to go to the full lists of players and their stats.
With that, to the selections. Envelope, please. The 2010 WPA's go to...
Team Efficiency Rankings - Week 17
A very unlikely season. It's possible that 3 of the top 5 teams in overall efficiency will not make the playoffs. The number one team is already eliminated. The #12 and #22 teams will have byes in the NFC. The #29 and #30 teams are duking it out on Sunday night for a home playoff game, and the #26 team is theoretically still alive to win its division. That's the NFL.
SD's GWP is 0.81, which means they theoretically should have an .810 winning percentage over a large sample of games. After 15 games, the probability of such a team having at most 8 wins is only 1.4%. Special teams, bad luck, or whatever the adversity, 7.8 YPA and 5.4 defensive YPA should cure a lot of ills, especially combined with better than average interception rates on both sides of the ball.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
Offensive rank (ORANK) is offensive generic win probability, which is based on each team's offensive efficiency stats only. In other words, it's the team's GWP assuming it had a league-average defense. DRANK is is a team's generic win probability rank assuming it had a league-average offense.
Should the Falcons Have Punted?
Down 17 to 14 with 2:52 left to play, the Falcons faced a 4th and 6 from their own 43-yard line. With two timeouts and the two-minute warning in their pocket, they chose to punt, hoping to make a stop and get the ball back with enough time to tie or win in a final drive. Did head coach Mike Smith make the right call?
A punt in this situation typically nets 37 yards, which would give the Saints a first down at their own 20. This would give the Falcons a 0.15 Win Probability (WP).
Fourth-and-6 attempts outside the red zone are successful about 44% of the time. A success gives the Falcons a first down at (at least) their 49-yard line, worth 0.37 WP. A failure giving the Saints the ball at Atlanta's 43 would mean a 0.12 WP for the Falcons. All would not be lost. A stop or even allowing a FG still gives the Falcons time for a TD drive, which when all four downs are available are successful more often than many realize.
In total, the fourth down conversion attempt would be worth:
Players of the Week
With still games to be played this week, this might be a little premature. But there's enough attention on the prime-time games that great performances won't go unnoticed.
Aaron Rodgers kept his team's playoff hopes alive with his +0.61 WPA/+26.0 EPA performance against the Giants, topping all QBs for the week. 404 yards passing, no interceptions, with only 8 sack yards go a long way, especially when the opponent is keeping pace for much of the game. Runners-up are Tim Tebow (+0.54/+15.0) with an impressive comeback win and Carson Palmer (+0.43/+21.6) in a game that ended San Diego's shot at the post-season.
Matt Forte's 113 rushing yards at 5.9 YPC and 56 receiving yards on 4 catches against the playoff-bound Jets were good enough to top the RB list with +0.37 WPA and +5.9 EPA. Jamaal Charles (+0.21/+7.3) is runner-up with yet another impressive performance, including a 55% success rate. Correll Buckhalter is the EPA champ with +7.4.
Who's 'Clutch' in 2010?
Two of the most clutch QBs of 2010 face off tonight as the Falcons host the Saints. Although 'clutch' performance may not be a persistent skill in players, there undoubtedly exists clutch play itself. Due to the varying leverage created by the combination of score and time, some players will have their better moments when they matter most, and some players will have their worst moments at just the wrong time.
Here's one way to measure which QBs are most 'clutch' this season. WPA accounts for the leverage of score and time while EPA ignores it. We can plot each QB's Win Probability Added (WPA) against his Expected Points Added (EPA), creating a baseline expected WPA for each QB.
Love?
[Caution: Commentary below. No objective stats to be found in this post.]
It's the Christmas season, so it's a fitting time to ponder the concept of love. I've been taught a lot about love this season, mostly by people such as Chris Berman, Steve Mariucci, and Jon Gruden. According to them, Brett Favre plays football for the love of the sport. He loves the game so much, he's willing to go out there and risk severe injury, just so that we can enjoy watching him perform.
Brett Favre himself tells us how much he loves the game. His official personal website is titled For Love of the Game. And in his press conferences, he routinely and openly ponders how much left he has to give. Vikings fans aren't quite sure what he's been giving them, and I'm sure many of his teammates are now wondering what he has left them with.
We're told Brett Favre loves football, but what does that mean? What kind of love is that exactly? Love as in I love my kids? My dog? Pizza? Guacamole? English is a strange language when it comes to the word love. We use the term for so many things, from our desire for some McDonald's french fries right now to our favorite TV show to our undying selfless sacrifice, joy, and pride in our children.
Roundup 12/24/10
Mike Nolan regrets not benching Alex Smith sooner. Alleluia. Coaches cling to these top draft pick QBs far too long. Sunk costs.
We have a Doug Drinen sighting. Bring back the mid-week play-in game! (Well, bring back the idea anyway.)
Where did game theory come from?
Do teams that play on Thursday have an advantage the following week? Good question. I wonder if teams that play on Monday are at any disadvantage the following week. There would be a larger sample than for Thursday games.
Is Devin Hester Worthy of the Hall of Fame?
Hester recently broke the NFL record for total touchdowns on kick and punt returns with 14. There’s no question that’s a lot, but what kind of impact on game outcomes does a returner like that really make?
Aside from his touchdowns, Hester is an average returner. In other words, he either takes it to the house or he’s getting average return yardage, about 9 yds per punt return and 22 yards per kick return. He isn’t distinguished enough as a receiver to even begin meriting a discussion of Hall-worthiness with only 164 receptions and 2,132 receiving yards. Hester is distinguished by his return TDs and his return TDs alone, so let’s start to put those in perspective with a quick back-of-the-envelope analysis.
Hester’s total of 14 touchdowns is spread out over most of five seasons as a returner. He has 10 punt return TDs in 175 attempts and 4 kick return TDs in 112 attempts. Let’s compare those numbers to what an average team could expect in terms of returns.
In roughly the same period as Hester’s career, 1.3% of punt returns are touchdowns, and 0.7% of kick returns are touchdowns. The ‘expected’ number of touchdowns given Hester’s attempts would be 2.3 punt return TDs and 0.8 kick returns, for a total of 3.1 TDs. In essence, Hester has provided about 11 more TDs than we would expect over his five-year career.
Game Probabilities - Week 16
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I take a closer look at pivotal showdown between the Giants and Packers. One of these two very good NFC teams won't make the playoffs this year.
Washington Post: Measuring Offensive Line Woes
Today's post at the Washington Post's Redskins Insider looks at the Redskins' dreadful offensive line play, making the case that the line needs to be the team's primary focus in the off-season.
Team Efficiency Rankings - Week 16
Not much movement at this point in the season. The most significant exception is NE, which vaults from from #5 to #2.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
Offensive rank (ORANK) is offensive generic win probability, which is based on each team's offensive efficiency stats only. In other words, it's the team's GWP assuming it had a league-average defense. DRANK is is a team's generic win probability rank assuming it had a league-average offense.
AYPA Quiz
Most readers are familiar with Adjusted Yards Per Attempt (AYPA) as a very handy number for measuring the success of a QB. It's defined as net passing yards minus a 45-yard penalty for each interception, divided by total drop-backs. It correlates very well with winning.
Quick, without looking it up, rank the following passers in terms of their 2010 AYPA through week 15:
A. Drew Brees
B. Peyton Manning
C. Colt McCoy
D. Matt Ryan
E. Sam Bradford
Answer after the jump.
More Good Stuff at the Community Site
This past week the Community site received some interesting contributions.
Jim Glass continues his look at 'Big Win %' (A team's record in terms of beat-downs and close wins.) He also forecasts what this might imply for this season's playoffs.
Steven Buzzard asks if short passing success is the key to winning.
James Deyerle compares FG % vs Expected FG %.
Thanks to everyone who has contributed so far. We're getting some real quality stuff lately. Thanks again to Ed A. for editing the Community site.
Top Defenders of the Week
Top defensive playmakers of each week are now available at these links:
Defensive Ends
Defensive Tackles
Linebackers
Safeties
Cornerbacks
plus, All Positions
Week 15 Offensive Players of the Week
Just as everyone would have predicted, Drew Stanton tops all QBs in week 15. His 23-for-37, 252-yard performance was good for +0.80 WPA/+11.8 EPA and enough to lead the Lions from behind to beat Tampa Bay in overtime for their first road win in three years.
Ray Rice is the RB of the week. In the Ravens victory over the Saints, Rice logged 153 yards rushing on 37 attempts (4.9 YPC), plus 80 yards receiving on 5 catches. That's worth +0.50 WPA/+7.4 EPA.
Jamaal Charles gets the honourable mention (I'm in London today) as he continues his great season. Sunday he gained 126 yards on 11 carries. Let me repeat that...on 11 carries, good for 11.5 yards a pop and +10.0 EPA. I know to a lot of people EPA and WPA are just numbers, but they have practical significance. Yards and catches and completions are all good things, and we know more is almost always better. But what exactly do they mean in terms of winning? For example, how do we compare a RB with 100 yards of rushing and a fumble with a RB with 70 yards of rushing but no fumbles? Charles' +10.0 EPA means that his 11 carries generated 10 net points of the 14-point net differential in the Chiefs' 27-13 win over the Rams.
The Weekly League: Notes and Ideas for Week Fifteen
This week's edition of The Weekly League features:
1. Mind-altering previews of the Kansas City-Saint Louis, Green Bay-New England, and Chicago-Minnesota games.
2. An updated "luck" table, exposing the Atlanta Falcons for the frauds they really are.
and
3. No shortage of The Good Times™.
The Four Factors you see for each game represent each team's raw performance thus far in four important categories (pass and rush efficiency, pass and rush efficiency against) relative to league average (where 100 is league average and anything above is good).
Along with the Four Factors, you'll see two other numbers: Generic Win Probability (GWP) and Game Probability (PROB). The GWP is the probability a team would beat the league average team at a neutral site. It can be found for all teams here. The PROB is each respective team's chance of winning this particular contest. Your host, Brian Burke, provides PROBs to the New York Times each week, and those numbers (along with methodology) can be found here.
Finally, a glossary of all unfamiliar terms can be found here.
Kansas City at Saint Louis | Sunday, December 19 | 1:00pm ET
Four Factors
Notes
• Unless I'm mistaken, the prevailing view on Rams quarterback Sam Bradford is that he's having a good season.
• Probably because St. Louis is at the top of their division, is why, right?
• But let's also make sure to note that Bradford is second to last -- ahead of only Jimmy Frigging Claussen -- in yards per attempts among qualified QBs, with 5.4.
• That's (slightly) behind Derek Anderson, it should be noted.
• In somewhat related news, both Bradford and Anderson are way better at being quarterbacks than I am at being anything in the world.
Roundup 12/18/10
Matt Ryan for MVP. WPA agrees.
Is rushing the key to beating the Patriots?
Hidden gems drive the Patriots' success in 2010.
Worst late-season collapses.
PFR continues to add features.
Down by 2 scores late. Kick first and save time for later?
New Feature: Advanced Player Stats by Week
I've been meaning to implement this for a while. Want to know who were the real heroes and goats are each week? Now we've got WPA, EPA, and bunch of other stats all on one page to help figure it out. There is now a new version of the individual stats pages broken out by week.
For example, in week 14, the top QBs were Peyton Manning (+0.63 WPA/+17.2 EPA) and Jason Campbell (+0.51 WPA/+23.9 EPA).
Here are the top RBs, WRs, and TEs.
Next week I'll have time to add a page for defenders.
Game Probabilities - Week 15
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I take a closer look at Jets-Steelers match-up.
Almost Always Go for 2-Point Conversions?
In the Buccaneers-Redskins game this past Sunday, the Redskins were able to score a potentially game-tying touchdown at the end of regulation, only to fail to hit the extra point due to a mishandled snap. Gregg Easterbrook suggested the Redskins should have gone for the two-point conversion, which is a plausible strategy in many circumstances. But Easterbrook went on to add this little tidbit: "Rushing deuce attempts are about 65 percent successful in the NFL -- a better proposition than the 50/50 of advancing to overtime."
It's well established that 2-point conversion attempts are successful slightly less than 50% of the time, so could the 65% number for runs possibly be true? If so, what would that mean for NFL strategy?
There have been 718 2-point conversion attempts from 2000-2009, including playoff games. Overall, they've been successful 46.3% of the time. But this is slightly misleading because it includes aborted kick attempts. If we weed those out, along with some other mysterious plays, such as Josh McCown's kneel-down while trailing by 5 points in the final few seconds of the Cardinals-Vikings 2003 game, we get a different answer. For all normal 2-point conversions, the success rate is 47.9%.
Now look at the success rate broken out by play type:
Washington Post: Punting Stats Aren't Always What They Seem
Today's post at the Washington Post's Redskins Insider looks at the Redskins' recent firing of their punter and how misleading most punting stats can be. Also, here is last week's post, which I never got around to linking to. It breaks down how the Redskins have consistently declined since the beginning of the season and discusses what this says about their roster depth and coaching.
Team Efficiency Rankings - Week 15
New England's offense claims the number one spot this week, but their early season defensive woes are holding them back.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
Offensive rank (ORANK) is offensive generic win probability, which is based on each team's offensive efficiency stats only. In other words, it's the team's GWP assuming it had a league-average defense. DRANK is is a team's generic win probability rank assuming it had a league-average offense.
Advanced Team and Player Stats Updated
The database has been successfully repaired. Advanced team and individual player statistics are updated following each wave of games and currently include results through Monday night's two games.
Advanced Team Stats
Opponent Adjusted Team Success Rate
Quarterbacks
Running Backs
Wide Receivers
Tight Ends
Offensive Lines
Players By Team
Defensive Ends
Defensive Tackles
Linebackers
Cornerbacks
Safeties
Defenders By Team
Checking in on Tackle Factor
With 12 games under our statistical belts, I thought it's a good time to check in on a new concept I unveiled in the off-season. Tackle Factor (TF) is a stat that tries to make better sense of the NFL's tackle numbers. Let's see how this season's top defenders are faring in TF. If you're already familiar with TF, skip down to the tables of top players.
You might think that more tackles is always better for a player, but it's not that simple. Some positions simply get more tackles by their nature, especially linebackers. Plus, some defenses are worse than others, allowing more 1st downs, more plays, and by definition, more tackles to be had.
TF begins to untangle tackle stats by looking at a defender's share of his team's tackles. This accounts for the number of plays (and tackles) allowed by each defense. It then compares each defender's share of team tackles to the expected share of tackles by a player at his position. TF excludes special teams tackles and offensive tackles on turnovers. A TF of 1.0 would theoretically be average, but only if a defender has played in all of his team's snaps. TF is also adjusted for games played, so a player who has missed a few games due to injury isn't penalized.
One other additional benefit of TF, one that I did not even think of when I first implemented it, is that it normalizes for hometown scorer tendencies. Tackles are technically not considered official stats by the NFL. Each team has its own scorer for home games, who decides whether each tackle is recorded as a solo or who, if anyone, gets credited with an assist. So tackle stats are not purely objective. (This is not to suggest the scorers are biased in any way. Rather, the point is there are 32 slightly different standards for what qualifies as an assist.) Because TF is based on a share of team tackles, scorer tendencies are mitigated.
Jamaal Charles
I need a new post to drop the semi-nude picture of Tom Brady down off the front page, so I thought I'd chime in on the Jamaal Charles re-signing. The Chiefs inked (if you're a sportswriter you're supposed to say 'inked' ...or sometimes 'locked-up' when it's a re-signing) the RB to a 5-year deal with $13 million guaranteed. Is Charles really a top back worth top money?
Yes, he is. (You're supposed to make readers wait to the end of the article for a wishy-washy non-answer, but I'll cut to the chase.) Charles' numbers are great, and the seem to be getting better each year. Currently in his 3rd year, Charles is averaging 5.8 YPC for his career and 6.0 YPC for 2010. (4.3 YPC is average for RBs.) How good is 5.8 YPC? It's better than any 3-year span or any single year for LaDanian Tominson, Marshall Faulk, or even Erik Dickerson (not to be confused with the Bruce Dickinson, who could easily run for 5.8 YPC if he wanted.)
The Weekly League: Notes and Ideas for Week Fourteen
Along with the Four Factors, you'll see two other numbers: Generic Win Probability (GWP) and Game Probability (PROB). The GWP is the probability a team would beat the league average team at a neutral site. It can be found for all teams here. The PROB is each respective team's chance of winning this particular contest. Your host, Brian Burke, provides PROBs to the New York Times each week, and those numbers (along with methodology) can be found here.
Finally, a glossary of all unfamiliar terms can be found here.
New York Giants at Minnesota | Sunday, December 12 | 1:00pm ET
Four Factors
Notes
• True Fact: Americans are holding their collective breaths, waiting to find out whether Brett Favre or Tarvaris Jackson will start this afternoon's game.
• Other True Fact: Americans should probably stop holding their breaths, lest they suffer some sort of oxygen deprivation-related brain damage.
• As for which is the better option -- Favre or Jackson -- our host Brian Burke suggests the latter might have an edge on the former, although it's not entirely clear.
• A notable thing is that, despite what must be considered a disappointing season on the face of it, the Vikings are actually ranked tenth per GWP, with a 0.59 mark.
• They're also the NFL's fifth-unluckiest team (as you can see below in the GWP Luck table).
Interesting Posts over at the Community Site
This past week the Community site received some interesting contributions.
Jim Glass looks at whether the close-game/clutch-play narrative is backwards.
Ian Simcox measures just how big an effect strength of schedule can have on a team's regular season W-L record.
Karl Berthold looks at the effect Mike Martz has had on his teams over his career, both as head coach and as offensive coordinator.
Bruce D. introduces a novel type of team-rating system. It's based on points scored and allowed, adjusted for strength of schedule and for lucky plays. Bruce has also been keeping us up to date on which teams have benefited the most by lucky and unlucky plays. That is, high-impact plays that tend to be non-repeatable, such as kick returns for touchdowns blocked field goals, fumble returns, etc.
Many thanks to Ed A. for editing the Community site.
Fisher's Blunder
Trailing by 6 to the Colts with under 5 minutes to play, the Titans faced a 4th and 1 from their own 34-yard line. Coach Jeff Fisher chose to punt, and the Titans went on to give up a field goal on the Colts' subsequent drive and lose the game. How much did that decision cost the Titans?
A successful conversion would give the Titans a 1st and 10 on at least their 35, giving them a 0.23 WP. A failed attempt drops their chances to a 0.06 WP. Conversion attempts with 1 yard to go are successful 74% of the time, making the total WP for going for it:
Weekly Game Probabilities
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I take a closer look at Sunday Giants-Vikings match-up.
What Proportion of Football is _______?
How much of the game is passing, running, kicking, or punting? How much of football are interceptions? Fumbles? How much do penalties impact the outcomes of games? One way to answer that is to simply add up how many of those kinds of plays occurred and divide by the total number of plays. But that’s not going to tell us much, because each kind of play tends to have a different magnitude in terms of its effect on the outcome of games.
When we talk about “how much of football is” something, we need a good definition of what football is. The essence of football, like any other sport, is about competing and winning. So when we ask how much of football is passing, we want to know the impact of passing plays on the outcomes of NFL games compared to other types of plays.
Win Probability can provide the answers. In each game, the WP of each team starts at 0.5, but must end at 1 or 0, for a net total of 0.5 WPA. But between the first kickoff and the final whistle, the WP can swing up and down, traveling far more than the net 0.5. To calculate the total movement, we just need to add up the absolute value (7th grade flashback) of each play’s Win Probability Added (WPA).
EPA by Pass Depth and Down
In response to the recent post about the comparative success of short and deep passes a few readers asked for a break out of the results by down. Some suggested the advantage of deep passes might only due to unsuccessful check downs on 3rd down, which is certainly plausible. The numbers are going to be slightly different here because I neglected to exclude red zone attempts in the original post. (Please read the original post for definitions and caveats.)
Here is the break out by down. It appears that the advantage exists on all downs, and the advantage is not significantly greater for 3rd downs than for 1st or 2nd downs. But that doesn't necessarily rule out the check-down effect, as they are a common tactic on all downs.
Team Efficiency Rankings - Week 14
San Diego's strange season continues, but they lose their #1 spots on offense and defense. NE now owns the offensive honors while Pittsburgh owns the defensive title.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
Offensive rank (ORANK) is offensive generic win probability, which is based on each team's offensive efficiency stats only. In other words, it's the team's GWP assuming it had a league-average defense. DRANK is is a team's generic win probability rank assuming it had a league-average offense.
Advanced NFL Stats Coach of the Week
This week's COW is none other than Raiders coach Tom Cable. The Raiders beat the heavily favored Chargers on the road yesterday, mostly by virtue of two unconventional but statistically favorable decisions. Oakland built a 14-0 lead by the end of the 1st quarter, and both scores came courtesy of 4th down conversion attempts.
On 4th and 1 from the SD 9, Jason Campbell ran a play fake naked bootleg in for a TD. Later, on 4th and 1 from the SD 21, Cable made no hesitation. As soon as the measurement came in, he just gestured 'go' at Campbell. No timeout. No challenge. No try-to-sucker-the-defense-into-off-sides-then-call-timeout-and-kick nonsense. Just a QB plunge for 3 yards. The Raiders went on to score their second TD of the day, putting SD into a hole from which they never could dig out.
The Weekly League: Notes and Ideas for Week Thirteen
This week's edition of The Weekly League features:
1. Rapture-inducing previews of the Cleveland-Miami, Dallas-Indianapolis, Pittsburgh-Baltimore, and New York Jet-New England games.
2. A quiz!
and
3. Another quiz, right after the first one!
The Four Factors you see for each game represent each team's raw performance thus far in four important categories (pass and rush efficiency, pass and rush efficiency against) relative to league average (where 100 is league average and anything above is good).
Along with the Four Factors, you'll see two other numbers: Generic Win Probability (GWP) and Game Probability (PROB). The GWP is the probability a team would beat the league average team at a neutral site. It can be found for all teams here. The PROB is each respective team's chance of winning this particular contest. Your host, Brian Burke, provides PROBs to the New York Times each week, and those numbers (along with methodology) can be found here.
Finally, a glossary of all unfamiliar terms can be found here.
Cleveland at Miami | Sunday, December 05 | 1:00pm ET
Four Factors
Notes
• This game features two of the league's most interesting cases: Peyton Hillis and Cameron Wake.
• Wake, who began his professional career in Canada, is either second or tied for first in sacks, depending on whose numbers you use.
• He's also first in QB Hits, with 21.
• As for Hillis, he's third among all running backs in WPA (0.96) and and seventh in EPA/P (0.09).
• This, from a player who was traded for Brady Quinn this past offseason.
Washington Post: Returner Brandon Banks' Impact
Today's post at the Washington Post's Redskins Insider looks at the Redskins' recent loss to the Vikings.
-Why kick two field goals down by 10 in the 4th quarter?
-How big has return specialist Brandon Banks' impact been?
-LaRon Landry's absence really hurts the Redskins' D.
-The 'Skins are now 5-6. Do they still have a shot at the playoffs?
-The upcoming Giants game is a statistical uphill battle.
Weekly Game Probabilities
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I take a closer look at Monday night's Jets-Patriots match-up.
Roundup 12/4/10
Quantifying the importance of the Jets-Patriots game. I suppose many of the same insights could apply to the Steelers-Ravens game.
Is John Elway really the comeback king?
The Patriots are a second-half team this year. Last year they were a first-half team. I wonder if it's just a 'splits happen' phenomenon or something more?
Peyton Hillis and one of the most lopsided trades in recent memory.
Weekly playoff projections from NFL-Forcast.com.
Which is the better team, the one that's 7-4 or the one that's 4-7? Or is it a team that's 2-9?
Deep vs. Short Passes
Over the past couple years, we've learned that passing well is more important than running well in terms of winning games. We've learned that passing has become more and more lucrative over the years. And we've learned that offenses should pass more often, particularly outside the red zone and on 1st down. But 'passing' is a large category, encompassing everything from a screen pass 2 yards behind the line of scrimmage to a 50 yard bomb.
Beginning in 2006, the NFL classified every pass attempt as either 'short' or 'deep,' where deep means anything past 15 yards. About 19% of pass attempts are classified as deep. Unfortunately, that's all we get, so we can't tell a screen from a 14-yard down-field pass attempt. Still, it allows us to begin to pull apart different types of passes and examine them in one more layer of detail.
In normal football situations, in which the clock is not yet a factor and the score is relatively close, pass plays, including sacks, yield an average of +0.08 Expected Points Added (EPA), while run plays yield an average of +0.01 EPA. Which type of pass is more responsible for that advantage, risky deep passes or safer short passes?
Motivation
Much of statistical sports analysis is rooted in economics. Many of the statistical tools used in sports research are the same tools used in economic research, such as multivariate regression, utility theory, and game theory. Professional sports offer researchers a somewhat controlled labratory to examine more general economic principals.
But there is an unarticulated question at the heart of sports research: What is the point? What’s the bottom line? What are the athletes we study trying to accomplish? Making sense of any activity requires an understanding of the ultimate goal. To many economists, who have a financial orientation, the bottom line is quite literally the bottom line. To fans like me, the bottom line is winning, not revenue or profit. But economists often see professional sports in terms of financial gains and losses. I think researchers like me, who focus on wins, and professional economists, who focus on profit, are both missing a very large part of the picture.
If there is one thing that economics teaches us, it’s that incentives matter. The real question becomes, what are the incentives? For most of the history of economics, money was treated as the only important motivator. After all, money obviously does motivate nearly all people, plus it’s easy to measure. But there is far more to motivation and incentives than money, and economists would be the first to admit that. But just like how we admit measuring individual player performance is flawed and incomplete, it's done anyway simply because it can be measured.
Team Efficiency Rankings - Week 13
The toughest schedules so far this year belong to AFC East teams, Miami and New England in particular. Jacksonville has also faced a tough slate of opponents to date. Aside from the NFC West teams that get to play each other twice, the softest schedule so far belongs to San Diego.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
New Offensive Line Page
Aboutt a month ago, I introduced a method for valuing offensive line play using advanced statistics. The full explanation can be found in the original post, but the basic concepts goes like this:
The offensive line's job is, ironically, defensive. Lineman protect the quarterback or the rusher from tacklers. The best an offensive line can do is prevent their opposition from making plays, either by hitting or sacking the QB, by tackling a rusher for a loss or short gain, or by deflecting a pass. The impact that an opposing front seven defenders make can be measured by their +EPA or +WPA. These stats count only plays in which the defense scores a 'victory' by forcing a setback on the offense. An offensive line can be measured by how few 'victories' they allow, measured by how little +EPA or +WPA they forfeit to their opposing front seven.
In this implementation, the +WPA/+EPA of opposing defensive lineman are counted for all plays, and the +WPA/+EPA of linebackers are counted in only runs and pass rushes. In other words, pass defense plays by linebackers are excluded because the offensive line has no influence on linebackers in pass defense roles.
For example, the Patriots' offensive line leads the league with opposing defensive front sevens totaling the least +EPA with 84.6. The league average allowed +EPA for offensive lines is 127.6. That makes the NE offensive line 43.0 EPA better than average.
New Matchup Pages
You've probably noticed the win probability scoreboard at the top of the page. Compared to the old drop-down menu, it should be a much easier way to navigate from game to game to check out the graphs or leave comments. On Sundays it's a great way to stay on top of all the simultaneous games.
Clicking on a live or final game will take you directly to the WP graph for the game, but until now, there was nowhere to go by clicking on an upcoming game. Starting today, there is a matchup page for all forthcoming games, complete with one-stop shopping for a comparison of each team's stats. There's team efficiency stats, advanced team stats, plus conventional and advanced player stats for each team. I plan to add more features in the near future.
The matchup page started as something I thought might help Carson with his Weekly Notes feature or me with my contributions at the Post or Times, but I quickly realized it would be something everyone would like. It's a bit of an eye chart, but there's lots of good stuff in there. The efficiency table, which is the first one, might be the simplest and most useful.
For example, looking at Sunday's game between MIA and OAK, I can quickly see that MIA has the edge in passing, but OAK has the edge in running. On defense however, OAK has a hard time stopping the run. Both team have slightly better than average pass defenses. The difference in this game might come down to penalties, as OAK has one of the worst penalty rates in the league.
Roundup 11/27/10
Reader Gautham Venugopalan alerted me to this article that describes how Packers defensive coordinator Dom Capers grades out his defense. It sounds similar to a simplified success rate stat.
Behind the net on the randomness of NHL win-loss records.
What are the most prominent "wrong football beliefs?" Punting all the time, maybe. Running back overuse? The importance of running?
Bruce D shares his breakdown of lucky plays at the Community Site. He tracked each team's "lucky" plays, defined as non-repeatable things such as kick returns for TDs or fumble recoveries. He scores out each team in terms of these plays. Bruce then demonstrates that these plays do not correlate from one half of the season to the next, at least for 2009, suggesting they really are lucky plays. Awsome. I'd love to see more from Bruce and more from all you guys crunching numbers on your own.
This site appears to have some sort of EI algorithm for telling you which games on your DVR you should watch and which ones you should delete. Interesting idea, but how many people would use this? Helmet-knock: Tech Crunch and premium subscriber Borat. (For Christmas this year, I'm renewing Borat's premium subscription for half-price. I was raised to be very generous like that.)
Manning vs. Brady using EPA and WPA.
There's a great new ESPN stats blog, simply called Stats & Info. According to the site, it started as an in-house source of content for producers and analysts, but now it's publicly available. Nice blend of conventional and advanced statistical insight. Here are some example posts. Look for more advanced stats, including for football, from ESPN in the future. It's a niche they're looking to fill, and they have some very bright guys working on their stats team.
The Weekly League: Notes and Ideas for Week Twelve
This week's edition of The Weekly League features:
1. Previews of the Green Bay-Atlanta, Philadelphia-Chicago, San Diego-Indianapolis, and San Francisco-Arizona games.
2. A photo of a player relevant to the article -- always a good technique to attract the attention of readers.
and
3. Over 100 points of joie de vivre.
The Four Factors represent each team's raw performance in four important categories (pass and rush efficiency, pass and rush efficiency against) relative to league average (where 100 is league average and anything above is good).
Generic Win Probability (GWP) is the probability a team would beat the league-average team at a neutral site. It can be found for all teams here. Game Probability (PROB) is each respective team's chance of winning this particular contest. Those numbers (along with methodology) can be found here.
Finally, a glossary of all unfamiliar terms can be found here.
Green Bay at Atlanta | Sunday, November 28 | 1:00pm ET
Four Factors
Notes
• Pop quiz, America.
• What's more surprising:
• That (a) Green Bay is third in the NFL, with a 0.74 GWP, despite a rushing attack fronted by Brandon Jackson, or
• That (b) Atlanta is a very pedestrian 19th overall, with a 0.47 GWP, despite a glistening 8-2 record?
• In any case, as the Increasingly Renowned Brian Burke noted mid-week, the latter shouldn't be very surprising at all.
Roethlisberger's 2010
In several aspects, Ben Roethlisberger may be having a career year. 6.5 AYPA, +0.25 EPA/Play, and 2.3 sacks per game (which is good for him). Six games is not a big sample, but the Steelers offense has been firing on all cylinders since his return.
What's the Deal with the Falcons?
The Falcons are 8-2, tied for the best record in the NFL, but in terms of efficiency they are below average. The efficiency model has them ranked 19th out of 32 teams. So what's the story?
I often hear the refrain that Matt Ryan had a down year last year, but he's back as one of the best QBs in 2010. But he's throwing at a 6.2 net YPA pace, exactly average this season. The defense is giving up 6.7 net YPA, over a standard deviation worse than average. Their running game is nothing special either, gaining an average 4.1 YPC, while the defense gives up 4.3 YPC.
Atlanta hasn't faced a terribly tough schedule either. Their opponent GWP is slightly below average at 0.48.
So could it be a consistency thing? Do their SR numbers indicate they are deadly consistent, inching the ball up the field on offense and forcing frequent 3-and-outs? Nope, their opponent-adjusted team SR is the same as their efficiency ranking--19th.
Washington Post: Another Game Comes Down to the Final Play
Today's post at the Washington Post's Redskins Insider looks at how the Redskins were able to pull out another close one. Yet another Redskins game comes down to the final play.
-Where to the Redskins' games rank in terms of excitement this year?
-How is Donovan McNabb performing compared to his career averages?
-How big a role did penalties play in overtime?
-The 'Skins are 5-5. What are their chances at making the playoffs?
NYT: Weekly Game Probabilities
Weekly game probabilities are available now at the nytimes.com Fifth Down. This week I also put a number on Santonio Holmes' 2010 heroics.
Team Efficiency Rankings - Week 12
There's a ranking I never thought I'd see. Peyton Manning's offense is ranked 21st in the league in efficiency. Last week, I got a few questions about how that could be , and the answer is opponent strength. The Colts have put up some slightly above average passing numbers, but they've been against some very weak defenses, including HOU twice, JAC, WAS, and DEN. Even NE's defense is ranked 23rd, giving up 7.0 net yards per pass attempt.
The team rankings below are in terms of generic win probability. The GWP is the probability a team would beat the league average team at a neutral site. Each team's opponent's average GWP is also listed, which can be considered to-date strength of schedule, and all ratings include adjustments for opponent strength.
How Random Are Interceptions?
One of the most random events in football is the interception. We can say that not just because of tipped balls or freak gusts of wind, but because we know that interception rates for teams and for individual quarterbacks varies widely from season to season. They also vary greatly from one half of the season to the next, which suggests that more is at play than player ability. In a recent post I estimated the proportion of team win-loss records that can be attributed to sample error in an effort to demonstrate the technique.
I looked at the interception rates (int per attempt) for all ‘qualified’ NFL quarterbacks since from 2002-2009. To qualify, a QB must have thrown a certain proportion of his team’s pass attempts. The overall average interception rate is 2.9%. The standard deviation, from QB to QB, is 0.94%, which means about 2/3 of the QBs will have interception rates that fall somewhere inside 2% and 4%.
To calculate the variance due to randomness, I couldn’t directly use the formula from the binomial distribution. Unlike team seasons, which all have 16 games, QBs have varying numbers of attempts during a season. Instead, I used a random simulation to estimate the random variance. I started with the premise that all interceptions were completely random. What if every quarterback’s pass attempts each had a 2.9% chance of being intercepted, regardless of who he is or who the opponent is. What would the distribution look like then? How different would this purely random distribution be from the actual distribution we observe in the real NFL?
The Weekly League: Notes and Ideas for Week Eleven
This week's edition of The Weekly League features:
1. Previews of the Green Bay-Minnesota, Oakland-Pittsburgh, Indianapolis-New England, and Denver-San Diego games.
2. Responses to three readers on matters sundry.
and
3. Tons of whimsy.
The Four Factors represent each team's raw performance in four important categories (pass and rush efficiency, pass and rush efficiency against) relative to league average (where 100 is league average and anything above is good).
Generic Win Probability (GWP) is the probability a team would beat the league-average team at a neutral site. It can be found for all teams here. Game Probability (PROB) is each respective team's chance of winning this particular contest. Those numbers (along with methodology) can be found here.
Finally, a glossary of all unfamiliar terms can be found here.
Notes to Three Readers
To the Reader Who Asked for Fewer NFC East Games
Your wish is my command. Or, at least, this wish is my command. Other wishes, probably not so much.
To the Reader Who Asked Why the Unlucky Teams on the GWP Table Are Green
I'm drawn towards -- and, I'm guessing, many readers here are drawn towards -- teams that are better than public perception might suggest. The green -- as opposed to the red -- reflects bad luck as a "virtue."
To the Reader Who Asked About My Impressive Jawline, Whether It's Natural
Yes. It is. Cento per cento.
Green Bay at Minnesota | Sunday, November 21 | 1:00pm ET
Four Factors
Notes
• This is what you might call "a clash of NFC North foes."
• So that's one thing.
• But another thing is: have you ever frigging seen Clay Matthews play?
• He's 24th among linebackers in +WPA and sixth in EPA/G.
• But he's more like first or second overall in the category of "being scary."