If you're wondering if your favorite team the worst in the league, it probably is. But you should read on, anyway!
The New Top Dawg
The Ravens beat the Texans this past weekend and were able to jump up from outside the top-5 onto the top of the pile. While the numbers gave Houston a ton of respect, the Texans were without Mario Williams and Andre Johnson, so the Ravens' win probably shouldn't have allowed them to leap so far in the rankings. Even though they probably don't deserve the top spot, Baltimore is still a very good team on both sides of the ball. With Lee Evans and Torrey Smith keeping safeties honest, Anquan Boldin and the Ravens' tight end duo should be able to take advantage on intermediate routes. Oh, and don't forget about Ray Rice, the little engine that flew up the hill so fast it made your head spin.
Undefeated, but not Perfect
The Packers rolled through the Rams without any trouble on Sunday, but beating a sad sack wasn't enough to propel them to the top spot in the rankings. While their defense has been quite adept at intercepting the ball this year, their total defensive numbers have looked downright offensive. Yes, some of that is due to their offense putting up big numbers and allowing the defense to slack off a bit, but they're going to need to prove they can stop a good team when they need to. If I had to pick one team to win the Super Bowl this year I'd still take the Packers, but I would like to see their defense toughen up.
Ranked 49th
Okay, so maybe San Francisco isn't ranked outside the top-32, but some of you probably feel like they are. After all, they beat a top-10 team in the Detroit Lions to push themselves to 5-1, the same record as the aforementioned Lions. Heck, the Patriots are 5-1 and the No. 1 Ravens are 4-1, so what gives with San Fran being ranked 17th? First off, while they are a good team, they are by no means a great team, and because the way their offense is set up, they may never be able to even flash the ability of being a great team. Their pass defense is pretty good, but it barely ranks inside the top-10 in that category; and while Jim Harbaugh may have made Alex Smith in to a serviceable QB, the 49ers' pass offense ranks below those of Kansas City, Miami, and Arizona in terms of yards per attempt. The 49ers are almost a lock for the playoffs, and in a one game sample anything could happen, but you shouldn't be expecting big things in the Bay Area quite yet.
After you've digested my delicious narrative for dinner, here is your complete data for dessert.
RANK | TEAM | LAST WK | GWP | Opp GWP | O RANK | D RANK |
1 | BAL | 6 | 0.66 | 0.55 | 10 | 1 |
2 | DAL | 1 | 0.65 | 0.55 | 3 | 5 |
3 | PIT | 4 | 0.63 | 0.51 | 7 | 7 |
4 | GB | 5 | 0.61 | 0.47 | 2 | 25 |
5 | HOU | 3 | 0.61 | 0.53 | 11 | 9 |
6 | NE | 7 | 0.59 | 0.52 | 1 | 28 |
7 | NO | 2 | 0.59 | 0.50 | 6 | 13 |
8 | TEN | 8 | 0.57 | 0.52 | 5 | 14 |
9 | DET | 9 | 0.57 | 0.47 | 15 | 2 |
10 | NYG | 14 | 0.56 | 0.47 | 4 | 22 |
11 | PHI | 16 | 0.55 | 0.49 | 9 | 19 |
12 | WAS | 10 | 0.54 | 0.52 | 22 | 4 |
13 | SD | 12 | 0.53 | 0.45 | 14 | 17 |
14 | BUF | 11 | 0.53 | 0.51 | 12 | 18 |
15 | NYJ | 13 | 0.53 | 0.54 | 24 | 3 |
16 | OAK | 17 | 0.51 | 0.52 | 13 | 12 |
17 | SF | 15 | 0.51 | 0.50 | 18 | 8 |
18 | CHI | 22 | 0.48 | 0.50 | 20 | 11 |
19 | CIN | 18 | 0.48 | 0.46 | 17 | 21 |
20 | DEN | 21 | 0.46 | 0.54 | 25 | 15 |
21 | JAC | 20 | 0.46 | 0.54 | 32 | 6 |
22 | CAR | 19 | 0.45 | 0.49 | 8 | 30 |
23 | STL | 29 | 0.44 | 0.58 | 28 | 10 |
24 | MIN | 23 | 0.41 | 0.46 | 26 | 20 |
25 | CLE | 25 | 0.41 | 0.47 | 30 | 16 |
26 | MIA | 24 | 0.39 | 0.53 | 23 | 29 |
27 | ARI | 26 | 0.39 | 0.46 | 19 | 26 |
28 | IND | 27 | 0.39 | 0.48 | 21 | 32 |
29 | ATL | 32 | 0.39 | 0.47 | 29 | 23 |
30 | TB | 31 | 0.38 | 0.48 | 16 | 31 |
31 | KC | 28 | 0.38 | 0.49 | 27 | 27 |
32 | SEA | 30 | 0.36 | 0.50 | 31 | 24 |
TEAM | OPASS | ORUNSR% | OINT% | OFUM% | DPASS | DRUNSR% | DINT% | PENRATE |
ARI | 6.1 | 48 | 3.9 | 1.6 | 7.1 | 56 | 2.3 | 0.55 |
ATL | 5.8 | 41 | 2.8 | 1.3 | 7.6 | 55 | 4.2 | 0.40 |
BAL | 6.7 | 39 | 2.3 | 2.1 | 5.2 | 65 | 3.2 | 0.34 |
BUF | 6.8 | 45 | 3.4 | 0.0 | 7.6 | 52 | 5.4 | 0.36 |
CAR | 7.4 | 45 | 3.9 | 0.3 | 7.7 | 54 | 1.9 | 0.54 |
CHI | 6.1 | 40 | 2.0 | 0.7 | 6.6 | 58 | 1.7 | 0.46 |
CIN | 6.3 | 43 | 2.5 | 0.3 | 5.5 | 57 | 1.0 | 0.39 |
CLE | 5.1 | 43 | 1.4 | 0.8 | 6.3 | 58 | 2.2 | 0.42 |
DAL | 7.7 | 37 | 3.9 | 1.5 | 5.8 | 59 | 3.2 | 0.45 |
DEN | 5.8 | 39 | 4.2 | 2.3 | 7.3 | 61 | 1.8 | 0.38 |
DET | 6.7 | 35 | 1.7 | 0.0 | 5.2 | 63 | 3.6 | 0.51 |
GB | 8.8 | 40 | 1.9 | 1.0 | 7.1 | 52 | 4.6 | 0.31 |
HOU | 7.1 | 39 | 2.5 | 0.3 | 6.0 | 56 | 3.0 | 0.38 |
IND | 5.7 | 43 | 1.0 | 2.8 | 7.6 | 55 | 1.6 | 0.27 |
JAC | 4.5 | 38 | 3.6 | 1.1 | 6.2 | 61 | 3.1 | 0.27 |
KC | 5.8 | 36 | 3.5 | 2.1 | 7.5 | 56 | 3.1 | 0.38 |
MIA | 6.0 | 42 | 4.1 | 1.2 | 7.9 | 53 | 1.2 | 0.35 |
MIN | 5.3 | 43 | 1.1 | 0.7 | 6.5 | 55 | 2.6 | 0.46 |
NE | 8.5 | 50 | 3.4 | 0.6 | 7.9 | 53 | 3.4 | 0.44 |
NO | 7.5 | 44 | 3.0 | 0.9 | 6.2 | 56 | 1.3 | 0.33 |
NYG | 7.9 | 38 | 2.6 | 1.3 | 6.3 | 54 | 3.3 | 0.42 |
NYJ | 6.1 | 36 | 2.5 | 2.2 | 5.9 | 57 | 4.8 | 0.44 |
OAK | 6.8 | 42 | 2.2 | 1.0 | 6.1 | 55 | 1.9 | 0.61 |
PHI | 7.2 | 49 | 5.1 | 1.9 | 6.5 | 55 | 3.8 | 0.40 |
PIT | 7.0 | 44 | 3.0 | 2.0 | 4.5 | 55 | 0.5 | 0.41 |
SD | 7.4 | 45 | 3.8 | 1.0 | 6.4 | 57 | 3.8 | 0.31 |
SF | 5.8 | 39 | 1.2 | 1.4 | 6.0 | 64 | 3.3 | 0.53 |
SEA | 5.2 | 35 | 2.8 | 1.7 | 7.5 | 64 | 2.9 | 0.45 |
STL | 4.7 | 40 | 1.0 | 2.1 | 6.8 | 55 | 2.4 | 0.51 |
TB | 6.1 | 46 | 2.7 | 0.7 | 7.6 | 55 | 2.4 | 0.56 |
TEN | 7.2 | 33 | 2.2 | 0.8 | 5.5 | 53 | 3.1 | 0.51 |
WAS | 6.1 | 44 | 5.0 | 0.8 | 5.7 | 57 | 2.9 | 0.39 |
Avg | 6.5 | 41 | 2.8 | 2.1 | 6.6 | 57 | 2.8 | 0.42 |
Confirmation bias be damned, you must be happy to see the Ravens at the top of the heap.
Forgive me if I have my terms wrong, but hopefully my question will make sense.
There's such a small data set (6 games) and the differences between teams is so small (Bal is .66 while Dallas is .65 and both defenses are .55), are win probabilities more meaningful then noise or randomness? Also, how does the effect of noise or randomness change as the data set grows?
My guess is that with 6 games, randomness probably has a greater effect then win probabilities, but at some point, those lines should cross. Is there a way of capturing that?
Philosophical debate time! Which is the better division according to the rankings: the NFC East or the AFC North?
Top to bottom the NFC East is better with all teams ranked in the top 12, the best average GWP (.575 to the AFC North's .545), and the lowest standard deviation (0.04) of any division (AFC North highest at 0.10).
However, it's reasonable to argue having three equally matched good teams and one terrible team benefits a division as it maximizes the three good teams' playoff chances as they can all go 4-2 in division (see: 2010 NFC South). So by only judging the top 3 teams of each division, the AFC North just barely edges out the NFC East 0.590 to 0.587. But the AFC North also boasts the #1 and #3 overall teams, which is darn impressive and should count for something.
Ultimately, what do you value more - overall quality and parity, or individual greatness?
I understand SF isn't an elite team yet, at least not offensively. But how do they drop in the rankings after beating a top-10 team like the Lions?
Fair question. It's due to how the 49ers won the game, which was basically in the least repeatable way possible.
-a couple long breakaway runs
-last second 4th down TD
-Smith had a terrible game: -14.4 EPA and a 26% SR
-Conventionally, he had a 53% completion rate, many of which were easy screens, with only 125 total yds averaging just 1.9 Adjusted YPA
How do your rankings account for special teams? SF has had great special teams in all facets. They've won the field-position battle in every game they've played.
It's also true that SF wouldn't likely be spotting Detroit 120 yards of penalties if they were to meet again on a neutral site.
I totally get why the 49ers are ranked so low. Their best trait on offense is avoiding interceptions, which is sure to change soon. They don't have a good QB and their WRs stink, so I think their 26th-placed rank in NY/A is appropriate. And they're good at run D in an era where pass D rules.
That said, I really like SF and think they continue to play well. I don't know what that means, since we all know they're winning the NFC West, anyway. But they remind me a lot of the 2009 Jets. Great running game and a great run D. They're the only team this year that hasn't allowed a rushing TD. They're the only team that hasn't allowed a RB to gain over 65 yards. And Gore/Hunter is an awesome 1/2 punch.
Moreover, I like the coach. If this was 1973, they'd be a legit SB contender. That said, I think they're still going to be a tough matchup for everyone, but could see Aaron Rodgers/Drew Brees torching them ala Manning against the Jets. But put them against any other team in the NFC (or any team in the AFC except NE) and I like their chances.
Harbaugh v. Harbaugh on Thanksgiving should be a real treat for fans who like that style of football.
True. 2008 Ravens or 2009 Jets.
With that kind of formula, they're going to live or die based on turnovers. It turns a lot of games into coin flips.
Interesting that 2008 Baltimore played so few coin flips. And, this style...if it holds up as an archetype, might struggle in coin flips.
2008 Baltimore (regular season)
11-2 in games decided by 7 or more
0-3 in games decided by 6 or less
2009 NY Jets (regular season)
8-2 in games decided by 7 or more
1-5 in games decided by 6 or less
Combined:
19-4 in games decided by 7 or more
1-8 in games decided by 6 or less
Still 5 divisional games ahead in a weak division for SF. A good test to see if the style creates coin flips...or lets you pull away from lesser lights.
I'm a Niner fan and have watched about every game and pretty much agree with the overall ranking.
The offense has had spurts of looking good but overall its been average at best. Alex Smith's low int rate might continue since he is throwing deep rarely and also throwing infrequently. Lots of short throws and screen passes.
The special teams have truly been outstanding this year and has a been a big reason why they are 5-1. Though probably not sustainable Ted Ginn might be having one of those magical returner seasons.
Also the biggest difference between this year and last is the turnover differential. They are +8 after 6 games. That always helps you win but again is not something you can count on.
However, one big and real improvement has been the pass defense. Rogers and Whiter have been excellent. The Niners pass SR% confirms this as it is near the top this season vs being near the bottom last season.
While their luck might regress some, their remaining schedule is pretty soft and their has been enough real improvement in the secondary that I expect them to keep winning at a pretty high rate.
Right now NFL forecaster has them and New Orleans with about the same probability of getting the 2 seed in the nfc. If they could get the 2 seed that would give them a punchers chance of making the superbowl with a bye and a home game.
-Dave
Also I think having the offensive and defensive gwp rather than the rank wold be more useful. Maybe have the rank in parenthesis?
-Dave
Brian -
Does the INT% you use in the power rankings include some regression towards the career INT% of the quarterbacks involved? It seems somewhat weird to me that the Eagles are being ranked at a 5.1% and the Colts at a 1% interception percentage while the career rates of Aaron Rodgers and Derek Anderson differ by less than 2% (1.9 to 3.8).
I wonder because the rankings are presented as a looking-forward projection system, yet it seems that an easy way to 'beat the system' would be simply to tweak the teams based on some sort of historical-to-current INT% coefficient.
start showing some respect to the raiders. beat houston, new york jets, cleveland, and denver. those aren't easy wins, three were tough excluding cleveland (denver is division, always tough games). their losses were to the unstoppable-at-the-time bills and the patriots who were coming off a loss. the lack of respect for the raiders is sickening. now with palmer at the helm, the next few weeks will tell the tale of the season. either way, your rankings should be:
green bay, san fran, oakland. duh.
This proves once again how stats show a small piece of the picture.
49ers... How they are not in the top 5 is mind boggling... when using common sense and thinking about all factors. When it comes down to it a win is a win and they win.
They have the first or second best defense in the league... the only competition is the Ravens. The Jets are not in the same class since they can be run on. The 49ers beat winning teams: wait for it... On the road. Their o-line is only going to get better. They have and will continue to have one of the top 5 special teams units.
Your metric should start including wins as a factor in rankings... If you think about it that helps incorporate coaching, mental strength and closing ability which your formulas do not properly account for.
Brian -
I'm also confused on how you determine the O-Rank and the D-Rank. Mainly, the Chargers are ranked behind the Raiders in O-Rank, D-Rank, and strength of opponents, yet are ranked better. Is this due, in its entirety, to the difference in penalty rates, or are O-Rank and D-Rank derived using a different method than the coefficients that go into the overall ranking?
If the former, perhaps it could be wise to start separating the penalty rates into offensive and defensive components so that the offensive and defensive ranks can be more definitive.
- Marver
"Your metric should start including wins as a factor in rankings... If you think about it that helps incorporate coaching, mental strength and closing ability which your formulas do not properly account for."
What is the point of coaching, mental strenght and closing ability if they do not improve your efficiency?
Marver - Without being certain, I'd say the relatively small difference in GWP (.53 - .51) is due to the Raiders league high penalty rate to the Chargers near leage low rate.
Also, as Brian has mentioned here before, penalties are reported together and therefore it's difficult for him to parse which penalties were committed by the offense and defense.
"Fair question. It's due to how the 49ers won the game, which was basically in the least repeatable way possible.
-a couple long breakaway runs"
Heh. They ran that Wham play 3 times, and each time gained 10 yards or more. This is also the second week in a row they've gained more than 200 yards on the ground, and the week before that it was 165.
I know, "repeatable" means something distinct to a statistician. Still, it's humorous when its meaning so immediately contradicts its standard English meaning.
I'm a bit surprised that Baltimore ranks so high because of defense's relative lack of "predictivity" as compared to offense.
I was playing around with the other numbers a bit, and I think if you used passing success rate, rather than AY/A, which has about the same predictivity according to that article( and perhaps a bit more early in the season to account for some randomness in big plays?) Baltimore would fall a bit, and San Diego would climb a bit, I think.
In terms of YPC, SF is 10th in the league at 4.5. 4.2 is average. For comparison, PHI is #1 with 5.7.
In terms of SR, SF is on the flip side, at about 20th in the league with 39%.
The crack about standard English meaning is fair. But people abuse the language when talking about stats all the time, myself included. Instead saying "SF is averaging 4.5 YPC," we should be saying "SF has averaged 4.5 YPC." There's a difference, just like there's a difference between "was repeated" and "is repeatable".
"Is repeatable" to me just means "is likely to be repeated in the future." It's not a statement about the possibility that it could be repeated. Of course it's possible.
Bottom line on SF is that their offensive-defensive YPA differential is negative. (Oops, that is, has been negative!)
5.8 YPA thrown - 6.0 YPA allowed
They've managed to win largely due to a low int rate, which is likely to regress.
A successful run-heavy style can throw monkey wrenches into the process. And, having Michael Vick break some scrambles against you can create temporary illusions in your run defense stats.
Non QB Carries:
S.F.: 4.65 ypc (744 on 160)
Opp: 3.13 ypc (344 on 110)
Facing Vick is non-repeatable until the playoffs, so that data probably gives a better sense of the SF rushing advantages moving forward (as long as Gore stays healthy, which is always a big if).
Owning such a big edge with your RB's allows you to play a less risky offensive style. So, while Smith is unlikely to maintain a pace of 2 INT's for every 150 or so passes...he will be throwing conservative passes for an offense that's doing fairly well already playing it safe.
And, the fact that their opponents can't run as successfully will create an edge in the risk/reward relationship.
RB Runs/QB Passes
S.F.: 160 and 171
Opp: 110 and 241
That creates the potential for a TO edge to continue, even if it's not to the degree of 6-14 that we've seen so far.
Also...that makes it easy to lose YPP and YPA because opposing teams are throwing a lot more and are throwing further downfield.
Add it up...and you get a team that may not look very good in overall raw stats...but is doing a very efficient job with the risk/reward challenge in a way that is potentially repeatable.
The simplest of efficiency measures, borrowing from Dean Oliver in hoops, is points per possession. SF is #10 on offense and #2 on defense according to Jim Armstrong's numbers at FO.
2009 NYJ: 18 on offense, 1 on defense
2008 Balt: 19 on offense, 1 on defense
Can see Smith's probable INT regression dropping SF down into the class that's already been discussed in this thread...which suggests a current rating of 17th might be too harsh. If Gore gets hurt though...we're looking at a lot of games where both teams may be in the 200's in yardage (SF was 209-206-226 before Gore ignited).
Mix of a rushing edge with a strong defense may be underappreciated in projection methodology in terms of game-to-game stuff...then in terms of playoff potential given the postseason success of those 2008 and 2009 examples.
So I examined Jeff's implict claims using 2010 and 2009 correlations. There is basically no correlation between run efficiency and interception rate, nor is there any correlation between defensive running efficiency and interception rate.
There is a relatively strong correlation between passing efficiency and interception rates over those 64 observations. However, they run the other way. This is obviously, to a large extent, a measure of quarterback skill.
There is almost no relationship between ORE-DRE and OINT%.
San Francisco's D has been quite good, but I have no reason to think it's better than the 7 teams ranked ahead of them. I also don't think the data supports your theories, which have a nice intuitive appeal. Alex Smith remains Alex Smith, and we see it in the data, no matter what 150 passes have shown us.
Funny you put it that way. Alex Smith is not in fact Alex Smith this year. Before he was 11-A.Smith according to the NFL. Now he's 11-Alex Smith, because there is another A.Smith I suppose.
Almost every week I have to go back into the database to fix, and I still need to do it for last weekend.
Did it.
"Bottom line on SF is that their offensive-defensive YPA differential is negative. (Oops, that is, has been negative!)
5.8 YPA thrown - 6.0 YPA allowed
They've managed to win largely due to a low int rate, which is likely to regress."
Yes, this is a good point. Someone on a different message board had an interesting rebuttal to this negative differential -- that SF's special teams have gained them large chunks of real estate, making up for the offense's defficiency. They've had the best special teams in the league by many measures...what's that worth in raw yardage?
Also, thank you for the kind response to my "repeatable" crack. I realized it might come off meaner than I intended, and appreciate that you took it in the best spirit.
Appreciate the research Joel W. During your studies were you able to find any recent teams that were a reasonably good match for SF in terms of:
Strong Rush Defense
Edge of at least 1.0 ypc over opponents
Completion PCT of say, 62% or better
I'm trying to pick some thresholds there. SF presents an interesting cocktail at the moment with those point of attack edges and what looks to be a conservative passing approach from Harbaugh (SF is 9th in completion percentage at 64% currently). Assuming they don't run into another QB who scrambles for a bunch of yards, plus 1.0 ypc differential may be feasible going forward (if Gore stays healthy).
Cocktails are known to spit out extremes that surprise projections(high K pitchers in a getaway game in a pitcher's park will post stats that are a bit off the charts, but are also repeatable when that combo happens again...poor NBA defenses when playing their 4th game in five nights will get run over by an up tempo opponent in a way that busts up models a bit..but is repeatable the next time that combination happens)
I think we all agree that 2 interceptions for every 150 passes isn't sustainable. Are there recent teams similar to SF in those mentioned areas that might give us a sense of where he regresses to? So that, maybe, SF isn't as good as a 5-1 record would suggest...but is better than 17th in the Power Ratings would suggest because that combo works well in terms of risk-reward impact?
"Your metric should start including wins as a factor in rankings... If you think about it that helps incorporate coaching, mental strength and closing ability"
Past wins is a demonstratably poorer predictor of future wins than several other metrics. This is especially so if one thinks they demonstrate "mental strength and closing ability" in terms of "clutch play that wins close games".
Last year the public and pop-pundits were all hailing the Falcons as the team to beat in he NFC because they were at the top of the league in close-game and come-from-behind wins, proving their clutch character and closing ability with that 13-3 records But the stat mavens who know close games are determined overwhelmingly by random chance, and one-sided game results much better predict the future, favored the Packers, even though they were only 10-6 and just squeaked into the playoffs. How'd that work out? And how is it still working out for the Falcs?
"What is the point of coaching, mental strenght and closing ability if they do not improve your efficiency?"
Coaching definitely is reflected in efficiency. "Clutch play" is a myth, so it's not surprising that it isn't. As to "mental strength", well, when you can quantify it and rank teams by it, we'll be able to calculate how it relates to efficiency ratings.
Jeff I haven't looked at the interaction effects in any meaningful way. The 10 best teams and the 10 worst teams at run differential over the past two years had exactly the same interception rates (2.6 v. 2.5) so that didn't suggest anything meaningful. The data on here don't have completion percentage and OINT in the same place so I'll get to it perhaps when not at work, but my initial instinct is not to think it exists.
"Coaching definitely is reflected in efficiency."
Coaching may be reflected in efficiency, but it certainly isn't completely captured by it. Things like 4th down decisions, run/pass selection, use of timeouts, 2 pt conversion decisions, challenges etc do not necessarily show up in efficiency stats.
If a team adopted an aggressive 4th down strategy for example you would expect it to out perform Brian's model.
Re: Penalty rates
Do we know anything about the types of teams that have higher penalty rates? What I mean is: if you were to remove the penalty rates as input to the team rankings, what correlates best with penalty rates? In particular, I want to see if better teams have lower penalty rates because of some inherent quality about them or if penalty rate is a decoupled skill; this would also help me grasp/defeat the notion that teams who have penalties called against them are more likely to have future penalties called against them due to the first batch of penalties (in other words, the snowballing human effect to calling penalties that you hear announcers talk about).
Thanks very much JoelW...I'll try to play around with some numbers tonight too...
Did a little digging at Football-Reference for the last three full seasons. Not many samples for teams that had good defenses and plus at least 1.0 in runs-per-carry differential. So, I lowered it to 0.7 to increase the sample size some. Mostly good news for SF I think, but a bad counter-example.
So, we're looking at:
Quality Run Defenses (4.0 allowed or better)
Owning Point of Attack (+0.7 or better)
Conservative Passing (62% or better)
Then we'll see what happens with interceptions. Granting...that can be a bit loaded because a decent completion percentage can imply you're not throwing many picks. Or...an outlier with somebody being over their head in avoiding interceptions may also be over their head in completion percentage.
2010
*Pittsburgh (4.1 to 3.0 ypc, 62% completion)
Only 9 picks for the season, 1.9% pick rate, 5th best in the league.
*Houston (4.8 to 4.0 ypc, 64% completion)
Only 12 picks for the season, 2.1% pick rate, 8th best in the league
2009
*Baltimore (4.7 to 3.4 ypc, 63% completion)
13 picks, for a 2.5% pick rate, 10th best in the league
*Dallas (4.8 to 4.0 ypc, 63% completion)
Only 9 picks, for a 1.6% pick rate, 3rd best in the league (and the highlight of Romo's career in terms of avoiding picks)
*Green Bay (4.3 to 3.6 ypc, 65% completion)
Only 8 picks, for a 1.4% pick rate, 2nd best in the league
2008
*NY Jets (4.7 to 3.7 ypc, 66% completion)
This was the Favre saga...and he was very INT prone anyway, with 23 picks and a 4.3% pick rate
So, that's five of six positive examples for the model at least in terms of suggesting that the cocktail of rushing edges and conservative passing may lead to less of a concern with interceptions. Far from definitive. It's a rare cocktail. Will be interesting to monitor the storyline through the season...
Having watched the Ravens every year since their start, it's my opinion that this team is the best yet.
The 2000 Ravens defense was amazing, but their offense was the Achilles heal. Baltimore got hot at the right time after a 5-4 start where the NFL record was set for longest streak without a TD (still standing).
2006 was in my opinion a better team, as the Ed Reed led a defense achieving multiple turnovers every game, many of which went for scores. Steve McNair was not a bad QB, and although the running game was dead and there was little talent at WR, this was a more rounded team. After a 13-3 season the Ravens had a disappointing playoff loss, which is the only reason most people don't rank it with the 2000 team.
This year the defense is once again very solid (number 1 so far in your rankings), as many good RBs have been limited (Chris Johnson, Arian Foster), and teams are having a hard time passing against what "experts" keep calling a banged up secondary. The main reason for this is the dominance shown by the defensive line as they are getting to the QB with ridiculous ease, like two ends of a magnet.
Leading the offense is Ray Rice - easily one of the top 3 RBs in the league right now (if not the best), and Flacco is a very capable QB with weapons everywhere.
The only other teams at this level right now are the Packers and Patriots, only because of the importance of the QB position in the NFL. Although superb at QB, both teams are having defensive troubles and Baltimore seems to the most rounded.
I know it's early, but how much fun would a Ravens/Packers Super Bowl be??
Posted something a couple of hours ago…and it showed up after I posted but is no longer here. Not sure what happened. Will try again. Went through the numbers at football reference to find teams from the past three years who had quality run defenses (4.0 allowed per carry or better), a clear point of attack edge (plus 0.7 in ypc differential between offense and defense…started at 1.0 but there were so few samples I moved back to 0.7), and a conservative passing attack (at least 62% completed passes). Those may provide some guidance about the INT regression that’s on the horizon for Smith.
2010
*Pittsburgh (4.1 to 3.0 in ypc, 62% passing)
Only 9 interceptions for the year, with a 1.9% rate…which ranked 5th in the league.
*Houston (4.8 to 4.0 in ypc, 64% passing)
Only 12 interceptions for the year, with a 2.1% rate…which ranked 8th in the league.
2009
*Baltimore (4.7 to 3.4 in ypc, 63% passing)
13 interceptions for the year, with a 2.5% rate…which ranked 10th in the league.
*Dallas (4.8 to 4.0 in ypc, 63% passing)
Only 9 interceptions for the year, with a 1.6% rate…which ranked 3rd in the league (this was Romo’s best year in terms of picks by a good bit).
*Green Bay (4.3 to 3.7 in ypc, 65% passing)
Only 8 interceptions for the year with a 1.4% rate…which ranked 2nd in the league.
2008
*NY Jets (4.7 to 3.7 in ypc, 66% passing)
23 interceptions for the year, with a poor 4.3% rate. This was the Favre saga, and he had trouble with picks despite having a productive run game backing him up. Tough to call him “conservative” though. He was very accurate when not throwing to the other team.
So, with six cocktail qualifiers over the past three seasons, we have five success stories in terms of keeping INT rate low, but one turnover disaster. Too small a sampling for anything to be rock solid…but this cocktail isn’t going to have many samples. Mentioned the first time that the stat combo might be a bit loaded, as an outlier having a great year in accuracy would affect both completion percentage and INT rate. Maybe Tony Romo had a great year because the point of attack edges allowed him to play the percentages better. Maybe it was just a fluky year.
Will be interesting to see if Smith slides into the 9-12 range with picks…or blows up well beyond that. Some hope for Niners fans in these numbers I think. Some successful teams managed the risk/reward ratio well thanks to their edge at the point of attack.
Coaching may be reflected in efficiency, but it certainly isn't completely captured by it. Things like 4th down decisions, run/pass selection, use of timeouts, 2 pt conversion decisions, challenges etc do not necessarily show up in efficiency stats.
I'd think they would -- better 4th down decisions and run-pass ratios and such would result in correspondlingly better results per play, so why wouldn't they they be reflected in correspondingly better efficiency numbers?
More significantly maybe, 4th-down decisions and such -- as much as we stat mavens love them -- really have only a quite small impact in total play and game results. Football even in our day is 65% blocking, shedding blocks and tackling. Knock the other team on its collective butt and while going 24 points up one single decision on a 4th and short isn't going to matter much.
Coaching blocking and tackling, and coaching the basic tactical mismatches that make plays work, and coaching 50+ extremely competitive, aggressive and very disparate personalities into one postively motivated coherent team (plus the level of talent provided to the coach by the GM) is I think what produces 95% of the difference in efficiency numbers. Statistically insightful playcalling is the cherry on top, making a difference just at the margin.
Not that the margin isn't important. If it turns just one close game a year it can be the difference between making the playoffs or not, and if the game turned is in the playoffs it can be the difference between going all the way or not. So it can matter a lot -- but I wouldn't expect it to make much of a visible impact in overall team statistics.
Mastery of the statistics of the sport doesn't make one a coach or GM. My perfect coach would teach blocking and tackling and basic tactics and motivate like Lombardi, and have the statistical insights of the Bill James of football. But if he could have only one skill set and have to be deficient in the other, I know the combination I'd choose.
"I'd think they would -- better 4th down decisions and run-pass ratios and such would result in correspondlingly better results per play, so why wouldn't they they be reflected in correspondingly better efficiency numbers?"
Brian has often said that teams run too much on 1st down and pass too much in short yardage.
So imagine two teams with identical run and pass efficiencies. One of them has an NFL average coach. The other one passes more on 1st down and runs more in short yardage. You would expect the 2nd team to score more points, despite the fact that they have identical efficiencies.
I can't see how better 4th down decisions would have any impact on efficiency ratings. If a coach goes for it on 4th and 8 at the 40, from an efficiency standpoint it's probably just another passing play.
"More significantly maybe, 4th-down decisions and such -- as much as we stat mavens love them -- really have only a quite small impact in total play and game results. Football even in our day is 65% blocking, shedding blocks and tackling. Knock the other team on its collective butt and while going 24 points up one single decision on a 4th and short isn't going to matter much. "
Obviously 4th down decisions don't matter much in a 24 point blowout. Nothing you do matters much when you're up by 24 points (aside from turning the ball over).
But it could matter a lot in close games. Accuscore has done NFL season simulations that suggest following a more aggressive 4th down policy could result in as much as a 5% difference in win probability.
In other words if Brians model pegged them at a 0.50 GWP a team with an aggressive 4th down policy would actually be at around 0.55 GWP. That would move them from 18th to 12th.
If the Niners aren't winning repeatedly, can someone tell me why they keep winning?
"better 4th down decisions and run-pass ratios and such would result in correspondlingly better results per play, so why wouldn't they they be reflected in correspondingly better efficiency numbers?"
Brian has often said that teams run too much on 1st down and pass too much in short yardage.
So imagine two teams with identical run and pass efficiencies. One of them has an NFL average coach. The other one passes more on 1st down and runs more in short yardage. You would expect the 2nd team to score more points, despite the fact that they have identical efficiencies.
Does the second coach improve his team's efficiency numbers with his superior pass-run decisions on first down and in short yardage, compared to if he'd called plays by the inferior norm? I say he does. Then his better play calling *is* reflected in the team's efficiency numbers. That's how he's gotten his reputation for getting better performance out of lesser talent.
BTW, when two teams have equal offensive efficiency numbers, how can one have a systematic edge over the other in scoring? Your logic seems to confuse "efficiency" with "talent". If two teams have equal talent, the team with superior playcalling will have superior efficiency and superior scoring ability. If two teams have different levels of talent, the team with lesser talent may use superior playcalling to equalize efficiency and scoring ability between them.
As to the degree that playcalling on 4th downs, 2-point conversions and the like affect efficiency numbers and game outcomes, you seem to agree with what I wrote. I described their impact as...
"Marginal ... Not that the margin isn't important. If it turns just one close game a year it can be the difference between making the playoffs or not, and if the game turned is in the playoffs it can be the difference between going all the way or not. So it can matter a lot -- but I wouldn't expect it to make much of a visible impact in overall team statistics."
Your "a more aggressive 4th down policy could result in as much as a 5% difference in win probability" equals "as much as 0.8 wins a season". Does that really differ from my "if it turns just one close game a year..."?
Less than one win per season seems a pretty good example of "marginal" -- having as much as 5% of win probability attributable to this leaves 95% or more attributable to other things, such as team talent level, the coaching of blocking and tackling and the basic tactics that all teams use, and so on.
Not that the margin isn't important. One win per season can be the difference between winning the SB or not, making the playoffs or not, the coach getting fired or not. But it's still just the margin.
"Does the second coach improve his team's efficiency numbers with his superior pass-run decisions on first down and in short yardage, compared to if he'd called plays by the inferior norm? I say he does."
Probably, but that's not the point.
The point is that despite the two teams having equal efficiency ratings they have different probabilities to score.
"BTW, when two teams have equal offensive efficiency numbers, how can one have a systematic edge over the other in scoring?"
Because one of the two is passing when it should pass and running when it should run and the other is not.
The coach who passes on 1st down will tend to get more yards on first down than the coach who runs on first down. The coach who runs in short yardage will get more first downs than the one who passes.
All this is true regardless of their equal efficiency ratings. If coach A and coach B both get 7 yards per pass and 4 yards per run but coach A passes on 1st and coach B runs on first then coach A gets more yards on 1st down. The coach who better uses his run/pass selection will score more points.
"Your logic seems to confuse "efficiency" with "talent". If two teams have equal talent, the team with superior playcalling will have superior efficiency and superior scoring ability. If two teams have different levels of talent, the team with lesser talent may use superior playcalling to equalize efficiency and scoring ability between them."
It does not. I believe I cover this adequately above.
"Less than one win per season seems a pretty good example of "marginal" -- having as much as 5% of win probability attributable to this leaves 95% or more attributable to other things, such as team talent level, the coaching of blocking and tackling and the basic tactics that all teams use, and so on."
It depends how you define marginal. If you consider roughly 1 win per season to be marginal than yes, but the 5% difference is actually quite substantial in terms of the over all rankings. The #1 team only has a GWP of 67%. That's only 17% above 50%. 5% in a fairly large deal in that context.