Week 5 DVOA Ratings
by Aaron Schatz
The No. 1 team drops and the No. 3 team rises, but we still have the same top three teams in the same order atop the DVOA ratings after Week 5: Los Angeles Rams, Chicago Bears, and Kansas City Chiefs. Then there's a healthy gap between those three teams and the rest of the league.
Underneath the top three teams, the top ten has a number of teams that made healthy jumps this week. New Orleans and the Los Angeles Chargers climb together from No. 11 and No. 12 to No. 4 and No. 5 after two big wins. New England climbs from No 15 to No. 9, and Pittsburgh moves up all the way from No. 21 to No. 10. In the other direction, only one team can really compare with a similar-sized drop: Washington, which plummets from No. 9 to No. 22 after the Saints smashed them all over the Superdome. A week ago, Washington was in the top dozen on both offense and defense; now Washington ranks 20th or worse on both offense and defense.
Although these team jumps are impressive, what's historical about this year is that big gap between the top three teams and the rest of the league. We've seen similar gaps between the top of the league and everyone else, but it's usually one or two teams, not three. At 17.3% DVOA, New Orleans has the lowest rating ever for the No. 4 team after Week 5. The only other year where the No. 4 team at this point was below 20% was 1991, when the Chiefs were No. 4 behind 5-0 Washington and New Orleans plus 4-1 Denver. But that year was one team dominating, since Washington was over 70% and would eventually go on to be the best team DVOA has ever measured.
The unexpected team in the top ten is still the Miami Dolphins, whose DVOA stays roughly the same despite a 27-17 loss to the Cincinnati Bengals. Bear with me a moment, because I'm sort of obsessed with the Dolphins and why our ratings differ from conventional wisdom right now.
First, let's look at why our ratings thought the Dolphins had a better game than the Bengals. On offense/defense, the two teams came out basically the same. Cincinnati gained more yards per play, 5.7 to 5.0, but Miami had a better success rate on offense, 45 percent to 37 percent. That leaves special teams, where the Dolphins came out much better than the Bengals thanks to Jakeem Grant's 71-yard punt return, three Grant kick returns averaging 29.7 yards, and four Matt Haack punts that ended inside the 20. Miami's 24.1% special teams DVOA this week keeps the Dolphins stable in the DVOA ratings. The Dolphins lost the game in large part thanks to two turnovers returned for touchdowns, and the length of turnover returns is not really predictive for future performance.
That brings us back to why the Dolphins were so high going into this week. Most of the value is in their Week 2 and Week 3 wins, for completely different reasons. Week 2 is primarily about the defense, which held the Jets to a 33 percent success rate and forced two interceptions and three fumbles. Week 3 is about big plays on offense, as the Dolphins had just a 38 percent success rate but averaged 9.8 yards per play, the second-highest of any team in any game this year (behind the Rams against the Vikings in Week 4).
The Dolphins are doing different things well each week, except for Week 4 against the Patriots when they did nothing well. Somehow, the mix of plays has turned the Miami DVOA rating into something that's more than the sum of its parts.
Despite Miami's high rating, this week's strong New England performance helped sanity to reestablish itself. The Patriots are now tied with the Dolphins in the standings, and they once again have the higher DAVE rating. So it was a shock when I ran this week's playoff simulation and... we still came out with Miami as a slight favorite to win the AFC East. This time, the culprit is the remaining schedule for each team. Miami's remaining schedule ranks 28th with six games at home. New England's remaining schedule ranks 18th with five games at home.
Overall, I'm still left with a pretty complicated opinion on the Dolphins and how our ratings are treating them. I absolutely think that DVOA is overrating the Dolphins, and that in turn is leading to them being overrated in the playoff odds. (The playoff odds also don't know about New England's recent pattern of overcoming September struggles.) At the same time, I think DVOA is pointing something out that other people might be missing, that the Dolphins are not complete frauds. The defense, in particular, is better than we ever would have expected coming into this season, fifth in DVOA and sixth in success rate allowed.
OK, enough with my Dolphins obsession. Let's go to the top and bottom of the league and update how the Rams and Bills are doing among the best and worst offenses we've ever tracked.
|BEST OFFENSIVE DVOA
THROUGH 5 GAMES, 1986-2018
|WORST OFFENSIVE DVOA
THROUGH 5 GAMES, 1986-2018
One important thing about these historical tables is that DVOA each year is normalized so the average offense is at 0%. That's important for comparing teams in different years because the offensive environment in the NFL has changed so much between, say, the 2000 Rams and the 2018 Rams. You've seen a lot of stats quoted in a lot of places to show just how much offenses have dominated the season so far, but here's another table that demonstrates how extreme the 2018 season has been -- especially when compared to a one-year downturn in 2017 that now looks like a blip in offense's historic march upward.
Every run of DVOA starts with "raw VOA," simply comparing success on each play to the baselines based on multiple years of data. After that, we normalize each year to zero, with different down-and-distance combinations being normalized separately. But take a look at what has happened to that raw VOA over time. With a couple of exceptions, primarily 2013 and 2017, offensive levels and passing levels keep going up and up. Rushing levels, on the other hand, bounce around a bit without any kind of heavy trend. As a result, the passing premium -- the dotted purple line, the difference between passing efficiency and rushing efficiency -- has also grown over time.
Note that this graphic shows defensive VOA rather than offensive VOA, so that only passes and runs are included without having to consider aborted snaps and certain penalties such as false starts.
* * * * *
A little bit of news as far as how we're computing our various formulas. Based on some preliminary research on how long to consider preseason projections in rating teams into the regular season, I'm changing the percentages we use for DAVE over the next few weeks. This isn't a complete, permanent change, just something to give DAVE more accuracy until I'm able to do full research on changing the percentages. Here's how we'll be changing the percentage of DAVE based on our preseason forecast, compared to what we used to do:
After 5 games: Now 40 percent instead of 27 percent (or in recent years, 30 percent)
After 6 games: Now 30 percent instead of 19 percent
After 7 games: Now 20 percent instead of 8.5 percent
After 8 games: Now 10 percent instead of fully replacing DAVE with weighted DVOA
* * * * *
Stats pages should now be updated through Week 5, including playoff odds, the FO Premium DVOA database and snap counts.
* * * * *
These are the Football Outsiders team efficiency ratings through five weeks of 2018, measured by our proprietary Defense-adjusted Value Over Average (DVOA) system that breaks down every single play and compares a team's performance to the league average based on situation in order to determine value over average. (Explained further here.)
OFFENSE and DEFENSE DVOA are adjusted to consider all fumbles, kept or lost, as equal value. SPECIAL TEAMS DVOA is adjusted for type of stadium (warm, cold, dome, Denver) and week of season. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.
Because it is early in the season, opponent adjustments are only at 50 percent strength; they will increase 10 percent every week through Week 10. As always, positive numbers represent more points so DEFENSE is better when it is NEGATIVE.
DAVE is a formula which combines our preseason projection with current DVOA to get a more accurate forecast of how a team will play the rest of the season. Right now, the preseason projection makes up 40 percent of DAVE for most teams (50 percent for teams with just four games played).
To save people some time, please use the following format for all complaints:
<team> is clearly ranked <too high/too low> because <reason unrelated to DVOA>. <subjective ranking system> is way better than this. <unrelated team-supporting or -denigrating comment, preferably with poor spelling and/or chat-acceptable spelling>
- NON-ADJUSTED TOTAL DVOA does not include the adjustments for opponent strength or the adjustments for weather and altitude in special teams, and only penalizes offenses for lost fumbles rather than all fumbles.
- ESTIMATED WINS uses a statistic known as "Forest Index" that emphasizes consistency as well as DVOA in the most important specific situations: red zone defense, first quarter offense, and performance in the second half when the score is close. It then projects a number of wins adjusted to a league-average schedule and a league-average rate of recovering fumbles. Teams that have had their bye week are projected as if they had played one game per week.
- PAST SCHEDULE lists average DVOA of opponents played this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
- FUTURE SCHEDULE lists average DVOA of opponents still left to play this season, ranked from hardest schedule (#1, most positive) to easiest schedule (#32, most negative). It is not adjusted for which games are home or road.
- VARIANCE measures the statistical variance of the team's weekly DVOA performance. Teams are ranked from most consistent (#1, lowest variance) to least consistent (#32, highest variance).
79 comments, Last at 15 Oct 2018, 3:59pm
#1 by poplar cove // Oct 09, 2018 - 7:48pm
Frustrating a bit for us Detroit Lions DVOA fans after watching the Lions get a comfortable win over the Packers on Sunday. I know this has been discussed many times before on here but the DVOA game numbers in that Green Bay/Detroit game feel wrong. Green Bay was down by two possessions or more to Detroit for the entire final 46 minutes of the contest (until kicking a meaningless FG with 2 seconds left to cut the lead to 8 points). Before the 4th quarter started the Lions defense held the Packers to 5.5 yards per play after 3rd quarters, allowed just 3 of 10 on 3rd down conversions, had four sacks (two strip sack fumble recoveries) and allowed a 55% completion pct to Rodgers. They went into the standard prevent (which worked) defense and ended up closing out the football game. In the 4th the Packers ran 19 offensive plays all while being down either two or three possessions in the game versus a soft Detroit zone defense and the Packers end up putting up 200+ yards (nearly 11 yards per play in the 4th quarter) on offense and overall it looks like they dominated he game when the reality is that wasn't the case. Not saying the Lions were lights out but they played a solid game when it mattered most.
#9 by dank067 // Oct 09, 2018 - 9:45pm
DVOA probably doesn't give Detroit proportionally as much credit for the three turnovers as much as it punishes Green Bay given the nature of fumbles, and I would bet that's especially the case on the punt deflection. Scoring 17 points from short fields is also very good but there's probably a cap on how much credit they can ultimately get from DVOA, particularly when scoring on a drive starting from the 1 yard line.
It's probably fair to feel like the Lions were a little hard done by DVOA - they also had a few big plays on offense taken off the board via penalty, which DVOA is not going to count - but FO's response to complaints about prevent/4th quarter defense has always been that leaving it in is ultimately more predictive. Keep in mind the Packers also never punted - the Lions got the sack fumbles and held them to FG attempts in the 1st half, but with a lot of decent gains mixed in.
#27 by LyleNM // Oct 10, 2018 - 11:32am
Lions also get no credit for Mason Crosby missing FGs.
#10 by Cythammer // Oct 09, 2018 - 10:10pm
It is an oddity to see a team with two easy, dominant wins in their first five games ranked as the next to worst team in the league. Even more so since those wins were over the Patriots and Packers.
#17 by JoeyHarringtonsPiano // Oct 10, 2018 - 4:51am
Not surprised at all the Lions took a DVOA hit after that 4th quarter, especially on defense. The defense was conceding consistent gains and first downs (probably more than they actually wanted to) in order to avoid giving up a big play, and DVOA loves consistent gains and first downs. Also keep in mind that the Lions were bit lucky to recover all three Packers fumbles. The second half also confirmed that this team still has a lot of problems.
If you told me before the season that the Lions would be 2-3 heading into their bye week with this early schedule, I would have taken it. They obviously did it in a different way than I expected. Their DVOA is much lower than I would have liked, but that Jets game (which is looking more and more like an anomaly, like the Bills and Vikings game) is 20% of their season at this point. Remember last year, the Lions were at the top of the DVOA tables early in the season, then settled towards the middle of the pack. I suspect the opposite will happen this year.
#61 by Chappy // Oct 11, 2018 - 10:17am
Right. The easiest way to see this is by looking at STDEV this year. The Lions are 32nd, while they were at the top (ultimately 3rd) last year. I suppose a speculative (and charitable) reading of this flip in STDEV rank is that this is exactly what Bob Quinn wanted. The 2017 Lions were consistently mediocre. I think Quinn said he made a change because he felt that the Lions did poorly against top teams over the last few years. The flip side of this is that when you attempt to smash your ceiling you blow out your floor.
You get blown out by the Jets and beat the Patriots. I suppose some of this is inevitable pain of switching regimes, but only time will tell if the Lions can truly build a higher ceiling and reach exit velocity on mediocrity.
#2 by Laufy // Oct 09, 2018 - 8:14pm
Having watched every play of every Dolphins game, and followed DVOA for years, I think you're right that Miami is overrated here, and it's definitely on the offense. The big plays they've managed are great, but there's little to no sustenance, especially the last 3 games. I wouldn't be surprised to see their rating continue to slide each week unless something drastic changes, but with the OL in shambles, and Parker unable to stay on the field, the strength of their speed has no offsetting threat so teams are able to key in on that aspect and shut it down. Hope I'm wrong, and the defense has vastly overachieved thus far, with a hat tip to Fitzpatrick's excellent play, so there's at least something fun to watch.
#3 by anonymousse // Oct 09, 2018 - 8:32pm
This week's playoff odds made me think that an interesting stat to see would be how many teams each division is expected to send to the playoffs. There were pretty big shifts in both conferences this week—from the NFC East to the NFC South, and from the AFC South pretty evenly to the other three AFC divisions.
Something like "expected playoff teams" would only require a bit of addition of what's already there—assuming that (e.g.) the NFC East's sum total playoff odds percentage (~120%) means that the division is expected to send 1.2 teams to the playoffs.
#4 by smarmyplankton // Oct 09, 2018 - 9:07pm
Do all 3 components (OFF, DEF, ST) weigh equally? Does ST value drop in recent years due to less returns with new rules? From observation it feels like ST is less than 33% relevant.
#13 by Sixknots // Oct 10, 2018 - 12:25am
I believe that ST DVOA is calculated at 1/2 the value of OFF DVOA. So, a ST DVOA that would calculate to 12% is entered in the table as 6% and added to total DVOA as 6%. A -4% would be -2%. Is this correct?
#16 by Mountain Time … // Oct 10, 2018 - 3:30am
8 parts total: 4 offense, 3 defense, 1 ST, I believe
#19 by InTheBoilerRoom // Oct 10, 2018 - 8:24am
No specific weighting is applied to the different components. They combine together straight up for total DVOA. Aaron has stated in the past the the average values for each component in the current NFL landscape tend to naturally come out to a 4:3:1 ratio, roughly, but no weighting is actually applied.
#5 by smarmyplankton // Oct 09, 2018 - 9:09pm
The Bengals collapse in week 5 is interesting. They seemed to dominate both sides of the ball and gave up a few ST mistakes (punt return and blocked kick). But they seemed from play to play naked eye dominant over Miami yet the results say the opposite. Miami had some 3rd and 25 screen passes for 19 but I assumed that was DVOA useless.
#12 by Laufy // Oct 09, 2018 - 11:39pm
they weren't dominant at all - first half was pretty even until the end when miami jumped up 14-0 on a long drive and the punt return. 3rd quarter was more of the same until cincy was able to put together a FG and TD then Tunsil got hurt and Miami disintegrated. 4th quarter was Cincy's D dominating Miami's O, but Miami's defense was ok-ish in the 4th and much better than Cincy's O first 3 quarters.
#6 by dank067 // Oct 09, 2018 - 9:31pm
Packers special teams DVOA drops from 9th to 30th over the course of a single game. It's still only a 5 game sample size, but lol
#7 by Cythammer // Oct 09, 2018 - 9:33pm
Another way DVOA's view of the NFL differs greatly from popular opinion is in having the Giants as the best team in the NFC East. Of course, they are only very slightly ahead of the other three teams, but still. The NFC East, whether by the eye test or by DVOA certainly seems to have no good teams right now. I imagine the Eagles have by far the best chance of eventually working their way to being a decent team.
#8 by Bobman // Oct 09, 2018 - 9:36pm
This week's Jets/Colts game should be a bit snoozy (ignoring, for the moment, all Indy's injuries). Both ranked 25-28 on O, and then kinda close on D and ST. With pretty good Ds facing pretty inefficient Os, it might end up 6-6 until a great ST play breaks it wide open. I'm thinking a punt downed at the 1. That's a game-winner if ever I heard of one.
You know it's sad when your 13 year-old makes a comment about their OL sucking and you say, "no, this year they're actually doing okay. But everybody's injured and last game a rookie they drafted as a G played RT, then another G just went on IR." And his comment is "It's been that way the last five years." Yes, indeed. They're getting predictable. Kids these days....
Then I told him about the Packers, leading the league in games lost to injuries, won the SB a few years back.
#11 by thok // Oct 09, 2018 - 11:02pm
The Jets and Ravens are rated too high because how exactly do you lose to Cleveland? Making fun of teams that lost to Cleveland is a better system than this. Cleveland!!!
(Yes, I basically cut and pasted my comment from week 1. And I'm quite aware that Cleveland seems to have improved to a cromulent level of play.)
#15 by Mountain Time … // Oct 10, 2018 - 3:29am
+1 for avoiding the "perfectly" that seems to be a prefix for cromulent sometimes
#21 by mathesond // Oct 10, 2018 - 9:08am
Perfectly is a perfectly cromulent prefix!
#58 by Mountain Time … // Oct 11, 2018 - 12:01am
Of course, but it doesn't have to be there, and people seem to forget that. I get it, I forget that too sometimes.
#66 by dryheat // Oct 11, 2018 - 12:40pm
You're embiggened in my eyes for having the courage to admit that.
#67 by Mountain Time … // Oct 11, 2018 - 1:05pm
D'oh, nice one!
I didn't realize that Principal Skinner actually says "he's embiggened that role with his cromulent performance." Or that embiggen actually dates back to 1884!
#72 by deus01 // Oct 11, 2018 - 2:34pm
The original comment was also a missed opportunity to mention how the Browns embiggened their performance to that of a real NFL team.
#20 by Boots Day // Oct 10, 2018 - 9:01am
I was stunned to see the Browns ranking second in defense, ahead of even the Jaguars.
#43 by OldFox // Oct 10, 2018 - 4:47pm
Yes, good observation. I’m a Browns fan, and I was shocked by that rating as well. On the one hand it’s encouraging, but I’m a bit skeptical. The Browns’ D did look great against Baltimore, but it also gave up 45 points to Oakland. I guess you’d have to agree that the defense hasn’t been consistent. I’ll be interested to see how it ranks after a few more weeks.
#47 by JoeyHarringtonsPiano // Oct 10, 2018 - 5:34pm
The 45 points scored by the Raiders is a bit misleading, as 7 was on a pick-six, and a couple of other Raiders scoring drives were of the short fields-after Browns turnovers variety.
#51 by OldFox // Oct 10, 2018 - 9:07pm
True. And I do agree that the Browns' defense has in general played well. Denzel Ward has been a huge addition; Myles Garrett is healthy now and playing well. The linebacking corps looks good. I'm still a little stunned to see them rated as high as #2 in the NFL, but I like it.
#14 by GoDog // Oct 10, 2018 - 1:46am
What is going on with the NFC teams this year? After LAR, CHI, and NO at 1, 2, and 4, you have to wade all the way down to CAR, GB, and SEA(?!?) at 12, 13, and 14. Six in the top 18. You know it's bad when the NYG are seventh best in the NFC at #19. How long has it been since the NFC as a group did so poorly?
#18 by Will Allen // Oct 10, 2018 - 6:56am
It's weird to look at the Vikings past strength of schedule ranked 25 through 5 games, because road games (and I know hfa isn't part of the calculation) at Lambeau, L.A., and Philly through 5 doesn't seem soft.
If the Bears (two games against the Vikings) truly are the 2nd best team , New Orleans the 4th best, and Miami the 7th, that's a hard road for Minnesota the rest of the way, especially given they have a game at Foxboro as well.
Getting stomped by Buffalo is going to keep the Vikings DVOA down all year.
#22 by BJR // Oct 10, 2018 - 9:51am
Vikings also had the misfortune to play the 49ers before Garoppolo's injury, so won't get full credit for that performance.
#24 by jmusilli // Oct 10, 2018 - 9:55am
I just recently discovered these rankings, and I was wondering how you would effectively compare teams in weekly matchups using these rankings going forward. Is it as simple as comparing their DAVE rankings and choosing the higher? Or is it a more complicated process of comparing both offensive and defensive DVOA ratings to determine how teams will do on both sides of the ball?
Thanks for the great work, I'm loving the articles and all of the data!
#32 by Richie // Oct 10, 2018 - 2:04pm
I assume you are talking about it from a gambling aspect?
I've been using the DVOA ratings to pick games in a pool for a few years.
Early in the season I use the DAVE ratings, until they are removed, then I use straight DVOA. I add 3 points to the DAVE/DVOA of the home team and compare the numbers. Picking the team with the higher rating correctly picked the winner 65% of the time last season. So far this season, DVOA is only at 58%.
Last year through 5 weeks, DVOA was at 61%, but 67% for weeks 6-17. So it's possible that the ratings get a little more useful as DAVE reduces value.
I also find that DVOA is even more efficient if you avoid the games where the ratings between two teams are closer. (Not surprising info.)
But, if you are looking to use it against the point spread, I don't have info. But Football Outsiders Premium has some "against the spread" tools: https://www.footballoutsiders.com/premium-info
#33 by Eddo // Oct 10, 2018 - 2:25pm
In the past, people here have used 17% for home field advantage, not 3% (3 points is usually assumed to be what bookies use for setting spreads). I wonder if you'd get better or worse results if you used 17%.
#34 by ChrisS // Oct 10, 2018 - 2:57pm
I have a vague recollection of Aaron stating that 1 point is about 5% DVOA , so 3 points for home field is about right.
#49 by Richie // Oct 10, 2018 - 7:15pm
I just changed the formula to 17% increase (not 17 percentage points), and the straight up results for last year dropped from 65% to 64%.
It really doesn't seem to make much difference what multiplier I use. Instead of 17%, I plugged in 300%, and it still had 64% accuracy. I even tried the other way and used a multiplier of zero. (Which means that each home team's DVOA was changed to 0.), and that increased the accuracy to 66%!!
I had to switch to a 700% multiplier (58% accurate) or a minus-100% multiplier (53%) to really get some different results.
I don't know what this means.
#63 by Eddo // Oct 11, 2018 - 10:58am
You're using it as a multiplier? That's not what I've seen people do in the past. Usually they add or subtract the 17% based on home or road status, not multiply by 1.17. So a 10% DVOA team would be looked at as a 27% DVOA team if they are playing at home, and a -7% team on the road (or maybe you only add to the home team, and don't subtract from the road team?).
Or is that what you're doing, and just used the term "multiplier" instead?
#68 by Richie // Oct 11, 2018 - 1:39pm
When you mentioned the 17%, I thought you meant that as a 17% increase, not a 17-percentage-point increase. So yeah, I was trying the multiplier effect.
If I add 17 points instead of the 3 points I've been doing, it does improve the results:
2016: 61% correct from 59%
2017: 66% from 65%
2018: 64% from 18%
I guess I'll switch to +17% for home teams going forward.
#50 by jmusilli // Oct 10, 2018 - 7:24pm
I'm having a little trouble following the home team advantage thing you've got going on here. Is the 3 points 3%? Or is it like the other replies to this comment that infer that you should add roughly 15%?
Additionally, you're saying that in games where the DAVE is generally close (5 spots? 7?) it's more or less a toss-up? (Maybe toss-up is the wrong phrasing, but these matchups are obviously more volatile than say, the Pats against the Cardinals.)
Thanks for the info!
Edit: I think I've completely read it wrong, and that the number would be 15%, but it would be adding 15% of the current DAVE? I'm afraid I'm lost.
#54 by ChrisS // Oct 10, 2018 - 10:51pm
Yes hfa is about 3 points or about 15% of dvoa. So add 15-17% to the home teams dvoa or Dave.
#57 by Mountain Time … // Oct 10, 2018 - 11:55pm
I think the confusion is over what "15%" means, since DVOA is also expressed as a percentage. Take the Rams at 39.1% DVOA. Does adding 15% come out to 54.1% (39.1 + 15) or 44.9% (39.1 x 1.15)? I'd actually guess the former is better than the latter, but that's based on nothing but feelings.
Or for a negative team like Dallas at -12.9% DVOA, is it 2.1%, or -11.0%?
#62 by ChrisS // Oct 11, 2018 - 10:22am
My interpretation, a DVOA of 0% means an average team, a 15% DVOA team would outscore an average (DVOA=0%) team by 3 points. So I think adding makes more sense. Multiplying as you say would have more of an effect for higher value (absolute) DVOA teams and I think HFA is somewhat consistent across teams and should not depend on how good/bad teams are.
#64 by Eddo // Oct 11, 2018 - 11:00am
What people have done - to mimic the point spread three point thing - is add the 17%, not multiply by 1.17. So the Rams would come out to 54.1% if you used that technique. (Dallas would be +2.1% in your example.)
#69 by Richie // Oct 11, 2018 - 1:53pm
See my post #69.
I was originally talking about multiplying, but adding is the way to go. (Though the differences still aren't huge.)
"Additionally, you're saying that in games where the DAVE is generally close (5 spots? 7?) it's more or less a toss-up?"
Here's what I'm saying. I like to use DAVE/DVOA for a confidence pool. We pick the games straight up, but we assign a weight of 1 to 16 for each game (only using each weight once per week). If you pick every game correctly, you would score 136 points (in a week without byes).
Looking at 2017 results, if I give them home team an additional 17 percentage points in every game, DVOA correctly predicts the outcome 66% of the time.
But if I apply the weighting system noted, where the matchup with the largest differential between the two teams is assigned 16 points. Next-largest is 15 points, all the way down to the closest matchup being worth 1 point, then for the whole season DVOA will score 72% of the possible points.
So, the weighted system outperforms the unweighted system by 72%-66%. This tells me that generally, DVOA is a bit more effective at predicting outcomes when you are using it on the games where the two teams have larger differences.
#71 by Mountain Time … // Oct 11, 2018 - 2:25pm
Great info, thanks for sharing!
FWIW, IIRC, Aaron has said the Premium picks against the spread are north of 80% correct in most seasons, but I think that might be a selection of the most sure games. I've never bought FO Premium, but I intend to as soon as gambling comes to Colorado.
#23 by Aaron Brooks G… // Oct 10, 2018 - 9:54am
I didn't realize DVOA was zeroed out for each year. I thought it was zeroed across a multiple-year era.
Certainly colors the historical analyses, that the 1986 Bears were only a really great defense relative to 1986 and that can't be directly compared to other defenses.
#35 by ChrisS // Oct 10, 2018 - 2:59pm
Not sure I get this. To me, it seems like resetting each year normalizes the results so that no era adjustment is needed.
#25 by smarmyplankton // Oct 10, 2018 - 10:02am
Listing Miami's kick returns as a positive attribute is interesting. They had 3 kick returns for 29.7 average. But all three started 7 yards deep and ended up with possession starting at the 18, 19, and 25. The 18 was due to a hold enforced at the 28. So even if you credit returner getting to the 28 - there was no benefit to those kick returns
#26 by nat // Oct 10, 2018 - 10:57am
...we still came out with Miami as a slight favorite to win the AFC East.
The playoff simulations are pointing to an absolutely brutal season. There are just 5 teams with a mean simulated win total of 9.0 or higher. (CHI, NO, LAR, KC and LAC). Only KC and LAR project at 10.0 wins or higher. Those two both project at between 11 and 12 wins on average. In other words, their average simulated records in their last 11 games are between 6-5 and 7-4.
Those are just averages, of course. Somebody is going to outperform expectations. But who the heck will it be?
#28 by dmstorm22 // Oct 10, 2018 - 11:38am
I wonder how different it would be if it was based on DVOA to project playoff odds instead of DAVE as it does at this point in the season.
I think LA, KC & CHI have fairly high DVOAs for #1, #2, #3 at this point, and good records, so my guess is it would look a bit more pointed if it was based on DVOA.
#29 by Will Allen // Oct 10, 2018 - 11:46am
So much revolves around the unknowable of Trubisky's performance over the next 11 games.
#37 by Chip // Oct 10, 2018 - 3:30pm
My guess is that he finishes the year slightly below avg. around QB20.
He looked awful the first three weeks b/c he faced three top 10 pass defenses (GB, SEA, ARI). WRs simply weren’t getting seperation. GB has invested a 1st and three 2nds in the secondary the last two years. ARI has pro bowlers Patrick Peterson and Buddha Baker. SEA has Earl Thomas and Shaquill Griffin is a good player. Trubisky got happy feet when his first and second reads weren’t open. He’s still young (27 total starts since HS), learning a new offense and how to read his progressions. He was highly accurate in college so the accuracy should be there as we saw against TB. Nagy is an above avg coach and does a nice job tailoring scheme to talent (it helps that the FO also got him talent that fits his scheme). This is most obvious in how the weekly game plan changes with certain players favored based on matchups (Cohen vs Howard).
Trubisky’s ceiling is unknown, but I’d guess that he tops out after a few years as a rich man’s Jay Cutler, a Stafford / Dalton type, maybe a poor man’s Philip Rivers. A guy that can maybe carry the team for a game or two (like against TB), but who can’t win consistently on his own without good talent / coaching around him.
#40 by nat // Oct 10, 2018 - 4:15pm
I'm not sure about your idea that this year's top three DVOAs are unusually high.
Top three DVOA after week 5
2018: 39.1% 34.0% 33.9%
2017: 37.0% 24.1% 23.7%
2016: 36.0% 33.5% 32.6%
2015: 55.9% 51.5% 41.2%
2014: 41.3% 38.5% 29.3%
2013: 51.3% 34.9% 29.4%
That's 4th, 4th, and 2nd when compared to the same point of the previous five seasons. Pretty typical, if you ask me.
 They are higher than the previous two seasons. So I can see why you had that impression.
#42 by dmstorm22 // Oct 10, 2018 - 4:37pm
Thanks for digging it out.
RealIze I wrote fairly high but not sure why, as what I more was thinking was not that low.
Anyway, I think DAVE is still flattening the league more than it is
#46 by nat // Oct 10, 2018 - 5:17pm
Maybe so. Maybe not.
Last year the top three DVOAs at the end of the season where 30.7% 27.7% 27.1%
DAVE in week five can't tell us who will outperform expectations later in the season. But it seems to get the scale about right.
I can see what you're thinking, though. If the top three teams' DVOAs are "for real" then they should do better than 6-5 or 7-4 for the rest of the season.
It will be fun to find out.
#30 by big10freak // Oct 10, 2018 - 11:52am
Has any special teams unit value dropped more from game to game than Green Bay's thanks to the Crosbypalooza? And that penalty that negated a good kick return. And the I guess was a fumble or maybe not said the ref ok whatever Lions ball in the first quarter? The GB special teams was so awful on Sunday I thought maybe Zook had a bet on the Lions.
#31 by jgibson_hmc95 // Oct 10, 2018 - 12:16pm
I'm totally shocked to see Denver as good as they are in these rankings. I guess I'm probably suffering from recency bias, watching that pathetic performance against the Jets.
#56 by mehllageman56 // Oct 10, 2018 - 11:29pm
I'm thinking the garbage time stats are raising Denver's DVOA up. a little, and knocking down the DVOA for the Jets, especially on defense. Keenum did throw for 370 yards, and that last play interception gets discounted as a Hail Mary, even though it wasn't.
#60 by Mountain Time … // Oct 11, 2018 - 12:10am
As a Broncos fan so am i. They are nowhere near 10th best on offense, and they are way better than 22nd on defense. Actually if you swapped those ranks I'd be okay with them. DVOA has been way off on this team for multiple reasons
#36 by theslothook // Oct 10, 2018 - 3:16pm
I've been thinking about the offensive trend in the NFL for a long time - namely that it has gone on for so damn long. Just when you think the 2004 season or the 2011 season were extremes, you realize they are just part of the same secular trend.
What's interesting is - the same is true for the nba. Offense has been surging every year. The halcyon days of the 90s Bulls would be considered slightly above average offenses in today's nba.
Are these two trends in two vastly different sports completely coincidental? You can certainly make an argument they are(and I don't follow other sports closely enough to know if the same trend exists everywhere else).
I'll haphazard a guess and say - as the two sports have grown, smart people have joined these leagues and have embraced the innovation. Its not just fancy analytics like go for it on 4th down and shoot more three's - scouting, film work, and other traditional coaching methods have all gotten better. And since defense is always a reactive enterprise than offense - the innovations in offense are just larger and more magnified than defense.
#38 by Will Allen // Oct 10, 2018 - 3:44pm
I don't disagree much, but who knows what the Bulls are like offensively, in a league where hand checking is not allowed? I can't decide if that change in the NBA is akin to the NFL's 5 yard rule instituted after 77, or the more like the decision to more tightly enforce it after 03. I am enough of a curmudgeon to at least suspect that basketball was better without the 3 point shot. I really find the fact that NBA offenses revolve around getting open 3s to be kind of boring.
#39 by theslothook // Oct 10, 2018 - 3:46pm
I am a golden state warrior fan - so I have to disagree!
#45 by Aaron Brooks G… // Oct 10, 2018 - 5:10pm
The difference with the NBA is that numbers today are still down relative to the 1980s, when no one (except Detroit, late in the decade) played defense.
What's weird about the NFL is that scoring isn't up all that much; just yardage.
#48 by Richie // Oct 10, 2018 - 7:05pm
This post is a few years old, but Chase Stuart showed that the majority of the increase in scoring is due to field goals: http://www.footballperspective.com/scoring-distribution-since-1940/
#53 by sbond101 // Oct 10, 2018 - 9:36pm
This really bears discussion; It's certainly true that from 1940-2000 the kickers just plain got a lot better and more consistent, which led directly to a rise in FG scoring. What I think is more interesting is that FG scoring has continued to rise in the last two decades more than all other scoring. This coincides with my subjective observation that football in the red zone has changed a little bit in the last 20 years, but not really very much (running is still very important, zone defense is effective, the sideline is the "safe" place to throw the ball, etc...) - however football between the 20's is like a totally different sport. It should be possible to study this a bit more closely using drive-situation data.
#59 by Mountain Time … // Oct 11, 2018 - 12:07am
538 cites that Stuart piece in an article that comes to the same conclusion via a different route… https://fivethirtyeight.com/features/kickers-are-forever
Edit: I HATE google amp. Is it possible to turn that crap off?
#65 by Eddo // Oct 11, 2018 - 11:02am
That's really weird, I wonder what a cause for yardage increasing, but scoring not, would be.
My gut is to say "teams must be starting farther from the end zone, so there's more yardage available". Is that true? Does this mean special teams coverage and kicking are up?
Is red zone efficiency down, perhaps? Are teams getting a little more yardage on drives, but still stalling short of the end zone, so those drives are still just producing three points?
#70 by Richie // Oct 11, 2018 - 1:56pm
Maybe fewer turnovers means that offenses don't have as many short fields, so they gain more yards before scoring or punting?
#41 by OutToLunch // Oct 10, 2018 - 4:19pm
May only interest me, but...
How many times does a team have an estimated wins of x.5 which actually matches their actual wins total?
(Steelers, Est Wins = 2.5, Record 2-2-1)
#44 by theslothook // Oct 10, 2018 - 4:57pm
I went to check the Colts Pass Run offensive dvoa split and I saw that Buffalo pass offense is an unconscionable -72.9% with the next worst being -30%. In case you are bad at math, thats more than twice as bad as the second to worst pass offense and somehow miles behind the Jeff fisher - Jared Goff led rams. I don't remember all of the worst pass offenses in DVOA history, but I remember Alex Smith's 2005 SF year was pretty awful and Buffalo is still almost 25% worse than that unit.
That is mind blowing. I know some of those games involved Peterman, but this is a pretty stunning and dire outlook for the long term success of Josh Allen. I think, for his benefit, the best thing to do would be to fire Sean McDermot and hire an offensive minded coach asap.
#52 by Vincent Verhei // Oct 10, 2018 - 9:15pm
Worst pass offense in DVOA history is the 1992 Seahawks at -65.3%.
#55 by ChrisS // Oct 10, 2018 - 11:01pm
That is some ugly QB play. Stan Gelbaugh, Kelly Stoufer were the main QB's with Dan McGuire playing some as well. They had a 12.3% sack rate and an any/a of 1.7. John Williams, a full back, was the leading receiver. UGLY.
#73 by tictoc // Oct 11, 2018 - 5:56pm
John *L* Williams, please. ;)
He was always referred to with the 'L'. Just like '80s RB Dan Doornink was always referred to as 'Dr. Dan Doornink'.
-----------> to exist is to comply<-----------
#74 by Mountain Time … // Oct 11, 2018 - 6:01pm
To exist is to be Hydra?
#75 by Richie // Oct 12, 2018 - 1:04pm
I see that the Dolphins moved up to 5th on the playoff odds report. I don't think they were that high when this was published on Tuesday.
Did something change since Tuesday, or did I just not notice the Dolphins on there?
(I'm a Dolphins fan who can't possibly imagine the Dolphins making the playoffs.)
#76 by Aaron Schatz // Oct 12, 2018 - 6:23pm
They moved up between Monday and Tuesday, and it's weird, and there's some kind of error in the playoff odds report that didn't count Week 5 in the mean wins computations, and I'm not sure if that affected the playoff odds themselves. We're working on fixing everything before next week.
Also, one reason they moved up is that Washington was fifth before Monday Night Football and their huge loss.
#77 by Richie // Oct 12, 2018 - 6:44pm
I'll go buy my playoff tickets now.
#78 by Richie // Oct 15, 2018 - 3:27pm
Aaron, not sure if you are still monitoring this. But how did Miami fall out of the top 5 in playoff odds despite winning? I assume the updates today don't include updated DVOA - since you haven't calculated DVOA for this week yet - but are only updated based on game results?
#79 by Richie // Oct 15, 2018 - 3:59pm
Oh, I guess it isn't that the Dolphins odds have decreased, but the Ravens' situation improved quite a bit due to their win and Cincinnati's loss.