FO DVOA week to week playoff odds.

[Briz]

Hi everyone,
 
Football outsider’s DVOA playoff odds report is out.  A note on how it is compiled:
 
The playoff odds report plays out the season 50,000 times. A random draw assigns each team a win or loss for each game. The probability that a team will be given a win is based on an equation which considers the DAVE of the two teams as well as home-field advantage. (DVOA ratings are explained here: http://www.footballoutsiders.com/info/methods#dvoa.) Mean Wins is the average number of wins for this team over the 50,000 simulations. Odds are given for winning each playoff seed, as well as for winning the division (DIV), earning a bye week (BYE), or winning a wild card (WC). TOT represents odds of making the playoffs in any fashion.
 
Link to the full report:  http://www.footballoutsiders.com/stats/playoffodds
 
All of this information was compiled by Mike Harris at FO.  I do not take credit for it.
 
Through of Week 5 (note, this does not include the week 6 games yet):
 
AFC East:            
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
NE          3-2                          9.0                          40%                            10%                        8.5%                       49%
BUF        3-2                          9.0                          45%                           16%                        5.4%                        50%
MIA         2-2                          7.5                          14%                           2.1%                      5.6%                        19%
NYJ         1-4                          5.0                          1.3%                          0.0%                      0.2%                        1.5%
 
 AFC North:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
CIN         3-1                          10.2                        50%                            39%                        21%                        71%
BAL        3-2                          9.8                          29%                             24%                        33%                        62%
PIT          3-2                          9.0                          16%                            9.3%                      24%                        40%
CLE         2-2                          7.8                          4.8%                           3.3%                      12%                        17%
 
AFC South:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
IND        3-2                          8.9                          64%                              11%                        3.1%                      67%
HOU      3-2                          8.3                          33%                               5.4%                      6.2%                      39%
TEN        1-4                          5.9                          3.4%                            0.0%                      0.8%                       4.3%
JAC         0-5                          3.3                          0.1%                            0.0%                      0.0%                      0.1%
 
AFC West:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
DEN       3-1                          11.1                        61%                             53%                        27%                        88%
SD           4-1                          10.3                        36%                            25%                        39%                        75%
KC           2-3                          7.6                          3.8%                          1.9%                       13%                        17%
OAK       0-4                          3.6                          0.0%                             0.0%                      0.0%                      0.0%
 
NFC East:            
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
DAL        4-1                          9.7                          35%                             24%                        24%                        59%
PHI         4-1                          9.7                          38%                             22%                        22%                        60%
NYG       3-2                          9.0                          26%                             12%                        17%                        43%
WAS      1-4                          6.1                          0.8%                             0.1%                      1.9%                      2.6%
 
 NFC North:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
GB          3-2                          10.3                        67%                             42%                        14%                        80%
DET        3-2                          9.0                          25%                             13%                        24%                        49%
CHI         2-3                          8.0                          7.4%                           3.5%                      16%                        23%
MIN       2-3                          6.3                          1.3%                             0.4%                      2.5%                      3.9%
 
NFC South:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
ATL         2-3                          8.4                          51%                            11%                        3.0%                      54%
CAR        3-2                          7.7                          32%                            5.1%                      4.2%                      36%
NO         2-3                          6.9                          16%                             1.0%                      2.2%                      18%
TB           1-4                          4.7                          0.9%                           0.0%                      0.1%                      1.0%
NFC West:
Team     Record                  Mean Wins         Win Division                       Bye                        WC                         Playoff %
SEA        3-1                          11.2                        72%                             51%                        18%                        89%
ARI         3-1                          8.7                          14%                             10%                        28%                        42%
SF           3-2                          8.7                          14%                            5.8%                      24%                        38%
STL         1-3                          5.3                          0.4%                            0.0%                      0.9%                         1.3%
 
The odds for each playoff seed:
AFC:
1) Den (33%), Cin (22%), SD (13%)
2) Den (20%), Cin (17%), SD (12%), Bal (11%), Buf (11%)
3) IND (20%), Buf (18%, NE (14%), 
4) IND (34%), Hou (17%), NE (17%), Buf (10%)
5) SD (25%), Den (19%), Bal (17%), Cin (10%)
6) Bal (16%), Pit (15%), SD (14%), Cin (11%)
 
NFC:
1) Sea (31%), GB (21%), Dal (13%), Phi (11%)
2) GB (21%), Sea (20%), Dal (11%), Phi (11%)
3) GB (16%), Sea (15%), Atl (12%), Phi (11%)
4) Atl (28%), Car (20%), GB (9%)
5) Ari (16%), Det (12%), SF (12%), Dal (12%), Phi (11%)
6) Ari (12%), SF (12%), Det (12%), Dal (12%), Phi (11%)
 
The playoff seeding these data imply just based on simulated wins:
AFC:
1) Den (11.1 wins)
2) Cin (10.2 wins)
3) NE/BUF (both have 9.0 wins)
4) IND (8.9 wins)
5) SD (10.3 wins)
6) Balt (9.8 wins)
 
NFC
1) Sea (11.2 wins)
2) GB (10.3 wins)
3) Dal/Phi (both with 9.7 wins) 
4) Atl (8.4 wins)
5) Dal/Phi (both with 9.7 wins)
6) Det/NYG (both with 9.0 wins)
 
I will update once FO updates with the week 6 results.  Obviously, alot of this will change.  For example, NE/Buf will no longer be tied for that 3 seed.

 

[StupendousMan]

Updated odds for AFC East, including Week 6:
 
 

 

[snowmanny]

The Patriots are more likely than the Seahawks to make the playoffs.

 

[CaptainLaddie]

But the AFC East is up for grabs, right?

 

[Briz]

Updated with the results of Week 6.  Pats now win the Division in 2 out of 3 simulations.  They make the playoffs 70% of the time and gained a half a win in their projected win total.   
 
[tablegrid= Week 6 DVOA Playoff Odds ]AFC East             Team Mean Wins DIV BYE WC TOT CHANGE NE 9.6 65.80% 14.50% 3.80% 69.60% 20.70% BUF 8.1 20.30% 3.80% 5.50% 25.80% -24.20% MIA 7.3 13.00% 1.10% 3.50% 16.40% -2.80% NYJ 4.8 0.90% 0.00% 0.10% 1.00% -0.50%               AFC North             Team Mean Wins DIV BYE WC TOT CHANGE BAL 10.5 49.90% 42.10% 27.30% 77.20% 15.20% CIN 9.4 32.50% 24.40% 30.60% 63.10% -7.80% CLE 9 12.60% 10.00% 24.70% 37.40% 20.20% PIT 8 5.00% 2.30% 12.00% 17.00% -23.20%               AFC South             Team Mean Wins DIV BYE WC TOT CHANGE IND 9.6 85.40% 17.90% 0.90% 86.30% 19.20% HOU 7.5 11.70% 1.40% 5.20% 16.90% -21.80% TEN 6.1 2.90% 0.00% 0.90% 3.70% -0.50% JAC 3.2 0.00% 0.00% 0.00% 0.00% -0.10%               AFC West             Team Mean Wins DIV BYE WC TOT CHANGE DEN 11.4 61.80% 55.70% 28.30% 90.10% 2.00% SD 10.5 34.50% 25.10% 42.50% 77.00% 2.40% KC 7.8 3.70% 1.90% 14.80% 18.50% 1.30% OAK 3.3 0.00% 0.00% 0.00% 0.00% 0.00%               NFC East             Team Mean Wins DIV BYE WC TOT CHANGE DAL 10.8 50% 43% 28% 78% 19.40% PHI 10.5 45% 33% 27% 72% 12.20% NYG 7.8 4% 2% 7% 11% -31.40% WAS 5.7 0% 0% 1% 1% -2.00%               NFC North             Team Mean Wins DIV BYE WC TOT CHANGE GB 10.6 58.50% 43.50% 19.80% 78.30% -1.60% DET 9.9 31.60% 21.40% 30.20% 61.80% 12.80% CHI 8.8 9.80% 6.30% 23.30% 33.10% 10.10% MIN 5.6 0.10% 0.00% 0.40% 0.50% -3.40%               NFC South             Team Mean Wins DIV BYE WC TOT CHANGE CAR 7.7 54.00% 3.70% 0.80% 54.80% 18.70% ATL 7.2 25.70% 1.60% 0.90% 26.60% -27.40% NO 6.9 20.10% 0.70% 0.90% 21.00% 2.60% TB 3.8 0.20% 0.00% 0.00% 0.30% -0.70%               NFC West             Team Mean Wins DIV BYE WC TOT CHANGE SEA 10.2 48.80% 21.80% 20.70% 69.50% -19.70% SF 9.3 29.40% 9.50% 18.80% 48.20% 10.20% ARI 9 21.60% 13.40% 21.40% 43.00% 1.10% STL 4.9 0.20% 0.00% 0.20% 0.40% -0.90%  [/tablegrid] 
 
Based on wins, the simulations predict this playoff scenario after 6 weeks of play:
 
AFC:
1) Den 
2) Balt
Tie for 3)/4) NE and Indy both at 9.6 wins
5) SD
6) Cin
 
NFC:
1) Dal
2) GB
3) Sea
4) Car
5) Phi
6) Det
 
 
Full report here at FO:  http://www.footballoutsiders.com/stats/playoffodds

 

[SeoulSoxFan]

great work Briz - really appreciate the updates!

 

[tims4wins]

It's hard for me to buy the Ravens as getting a bye 42% of the time. I don't know that they win the division 42% of the time.

 

[Briz]

As you pointed out in your thread, DVOA really likes Balt.  They are #2 right now (table below).  I think FO's results are the product of that.  However, I think I agree with you. 
 
[tablegrid= DVOA summary week 6 ] TEAM TOTAL OFFENSE DEFENSE     DVOA DVOA DVOA 1 DEN 38.7% 21.5% -16.5% 2 BAL 28.2% 18.2% -7.0% 3 SEA 26.8% 12.1% -9.3% 4 GB 23.8% 18.6% -3.1% 5 CIN 18.0% 14.3% -2.6% 6 PHI 13.9% -4.0% -6.4% 7 DET 13.0% -5.7% -28.6% 8 SD 12.9% 14.1% 2.6% 9 CLE 11.7% 20.7% 10.4% 10 DAL 11.6% 13.2% 0.1% 11 NE 9.4% -1.3% -6.9%  [/tablegrid] 

 

[Briz]

Updated after Week 7:  http://www.footballoutsiders.com/stats/playoffodds
 
Pats are projected to have 9.5 wins (down 0.1 from last week).  FO's simulations predict they will get the 4 seed. They win the division in 56% of the simulations and make the playoffs 62% of the time.
 
[tablegrid= FO's DVOA playoff odds Report through week 7 summary ]AFC East             Team Mean Wins DIV BYE WC TOT CHANGE NE 9.5 55.5% 8.3% 6.8% 62.3% -7.3% MIA 8.3 24.8% 2.2% 8.2% 32.9% 16.5% BUF 8.2 19.3% 2.8% 7.4% 26.7% 0.9% NYJ 4.6 0.4% 0.0% 0.1% 0.5% -0.5%               AFC North             Team Mean Wins DIV BYE WC TOT CHANGE BAL 11 74.4% 55.1% 13.3% 87.7% 10.5% PIT 8.6 10.0% 3.2% 22.3% 32.2% 15.3% CIN 8.2 11.4% 5.0% 23.5% 34.9% -28.2% CLE 8 4.3% 2.3% 14.5% 18.8% -18.6%               AFC South             Team Mean Wins DIV BYE WC TOT CHANGE IND 10.7 96.8% 35.0% 0.4% 97.2% 10.9% HOU 7.2 2.8% 0.4% 4.9% 7.7% -9.2% TEN 5.9 0.5% 0.0% 0.9% 1.4% -2.3% JAC 4 0.0% 0.0% 0.0% 0.0% 0.0%               AFC West             Team Mean Wins DIV BYE WC TOT CHANGE DEN 12.2 81.8% 75.2% 15.2% 97.0% 6.9% SD 9.7 13.1% 7.9% 49.2% 62.3% -14.7% KC 8.6 5.1% 2.6% 33.3% 38.4% 19.9% OAK 2.9 0.0% 0.0% 0.0% 0.0% 0.0%               NFC East             Team Mean Wins DIV BYE WC TOT CHANGE DAL 11.3 59.5% 52.6% 30.6% 90.1% 11.8% PHI 10.5 39.3% 30.8% 37.0% 76.4% 4.0% NYG 7.3 1.1% 0.5% 4.4% 5.5% -5.8% WAS 5.9 0.1% 0.0% 0.6% 0.8% 0.1%               NFC North             Team Mean Wins DIV BYE WC TOT CHANGE GB 11.1 69.6% 55.2% 19.4% 89.0% 10.7% DET 10 28.0% 20.0% 40.5% 68.5% 6.7% CHI 7.7 2.3% 1.3% 11.9% 14.2% -18.9% MIN 5.5 0.1% 0.0% 0.4% 0.4% -0.1%               NFC South             Team Mean Wins DIV BYE WC TOT CHANGE CAR 7.1 48.0% 1.2% 0.4% 48.4% -6.5% NO 7.1 31.3% 0.5% 0.4% 31.7% 10.8% ATL 6.5 20.0% 0.4% 0.4% 20.4% -6.2% TB 4.1 0.7% 0.0% 0.0% 0.7% 0.4%               NFC West             Team Mean Wins DIV BYE WC TOT CHANGE ARI 9.7 43.3% 25.1% 20.9% 64.2% 21.2% SEA 9.5 39.7% 10.0% 20.5% 60.3% -9.2% SF 8.4 16.2% 2.4% 11.9% 28.1% -20.1% STL 5.7 0.7% 0.0% 0.7% 1.4% 1.1%  [/tablegrid]
 
Seedings if the playoffs started today based on projected wins:
 
AFC:
1) Den (12.2 wins)
2) Balt (11.0 wins)
3) Ind (10.7 wins)
4) NE (9.5 wins)
5) SD (9.7 wins)
6) Pit or KC (8.6 wins)
 
NFC
1) Dal (11.3 wins)
2) GB (11.1 wins)
3) Ari (9.7 wins)
4) Car or NO (7.1 wins)
5) Phi (10.5 wins)
6) Det (10.0 wins)
 
Some observations:
 
1)  Indy jumped 1.1 wins and is now contending with Balt for the #2/3 seed instead of fighting with the Pats for the 3/4 seed.
2)  Jets playoff odds are now 0.5%.  Jac and Oak remain the only teams with a 0.0% chance to make the playoffs.
3)  Ari is now projected to win the NFC west.  In fact, the simulations have Seattle on the outside looking in based on mean wins.
4)  8.6 wins makes the playoffs in the AFC while 10 wins will be required for the NFC.  Well, unless you are in the NFC south where a 7-9 record might get you in.
 

 

[Silverdude2167]

They love Baltimore even though they lost to the two "good" teams they faced and they should have lost to Cleveland if not for missed field goals.

 

[coremiller]

They love Baltimore even though they lost to the two "good" teams they faced and they should have lost to Cleveland if not for missed field goals.

 
DVOA loves teams that blow out bad teams, and Baltimore has four wins by 20+ points.  Also, opponent adjustments are not yet at full strength, so Baltimore isn't being fully penalized for the fact that their four blowout wins came against Pittsburgh, Carolina, Tampa, and Atlanta.  

 

[MainerInExile]

They project the Pats at 4.5 more wins, on average, so a 50% chance they go 4-5 or worse over their last 9 games?  I don't get that at all.

 

[EricFeczko]

I really appreciate the work put into these posts, but I think this is a good example of why I don't trust DVOA as a predictive metric at all. While it is very good at explaining past history, I suspect that DVOA over-fits the data, leading to the inclusion of factors that are not reliable, and therefore may not be predictive.

 

[dcmissle]

I can get to 4 losses pretty easily -- Denver, at Colts, at Chargers, at Packers.

That would leave them winning Bears, Lions, at Jets, Miami and Bills.

So they have split the difference at 4.5, which does not strike me as nuts. Any of the first four could go the Pats way with varying degrees of probability. But the last five could go the other way too,

EDIT

Entirely agree with DVOA not being a predictive machine, especially at this stage. They can't possibly account for probably the most important factor, the significant injuries yet to come.

 

[tims4wins]

What is weird to me is that they went down by 0.1 despite the win. I get that their DVOA got worse due to the nature of the win, but does that really offset the fact that going into the game there was a non-insignificant chance the Pats would lose the game? Seems to me that actually achieving the win would offset the drop in DVOA. But maybe I'm wrong.

 

[Briz]

I really appreciate the work put into these posts, but I think this is a good example of why I don't trust DVOA as a predictive metric at all. While it is very good at explaining past history, I suspect that DVOA over-fits the data, leading to the inclusion of factors that are not reliable, and therefore may not be predictive.

 
 

EDIT

Entirely agree with DVOA not being a predictive machine, especially at this stage. They can't possibly account for probably the most important factor, the significant injuries yet to come.

 
I agree with both of you that DVOA is not really a predictive machine.  It only really speaks to how a team is doing right now in comparison to the competition.  FO is using that data to project the remainder of the season.  As you pointed out dc, intangibles such as injuries do not get factored in.  I do agree with you that there is a danger of making too big of a conclusion from what is actually a limited dataset.
 
However, I enjoy following these predictions because it adds a data driven perspective on how the season is going for the Pats on a week to week basis.  I am compiling and posting it because I figured many of you would as well.  My intent is to only show the data and let people draw conclusions.  If the thought is these summaries are misleading until a more complete dataset is available (say week 10 or so), I can wait (or not do it all).  Besides, I do have an issue with the numbers this early because they use DAVE (which factors in preseason projections).  However, I would prefer to continue.
 
That being said, this week will be the last week that FO will use DAVE.  Their projections will switch solely to DVOA and weighted DVOA.  In FO's words:
 
"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 19 percent of DAVE for teams with six games played, and 8.5 percent of DAVE for teams with seven games played.  This will be the last week we consider DAVE; starting after Week 8, we will switch solely to using DVOA and WEIGHTED DVOA."
 
http://www.footballoutsiders.com/dvoa-ratings/2014/week-7-dvoa-ratings  This is stated at the end of the article for this week.  
 
 
 
 

What is weird to me is that they went down by 0.1 despite the win. I get that their DVOA got worse due to the nature of the win, but does that really offset the fact that going into the game there was a non-insignificant chance the Pats would lose the game? Seems to me that actually achieving the win would offset the drop in DVOA. But maybe I'm wrong.

 
I scratched my head at this as well.  Where DVOA seems to love Baltimore right now, it seems to not like the Pats. I am still not believing Baltimore is a legitimate contender.  They have feasted on what is by far the worst division in football, the NFC south.  
 
Another example of DVOA potentially liking a team where it should also benefit the Pats: Indy.  I know Indy looks great right now, but was that decisive win over the Bengals really worth a 1.1 game jump?  This is even more puzzling when you consider the Pats also defeated the Bengals by almost the same margin. At that time, Cinci was considered one of the best teams in football.  But now that Cinci has played a couple of non-NFC south teams, they no longer look like one of the best teams in the NFL.  I have trouble rationalizing that jump for indy while NE stayed the same.  I guess not destroying a bad Jets team (by DVOA) at home offset it.
 
My thoughts go to that the AFC north is weaker than FO's math suggests.   Cinci started the season with a win in division against Balt and then 2 decisive wins against the NFC south.  As I stated above, Balt has done well against the NFC south.  However, they lost to Indy by 7 and struggled against Cleveland (won by 2).  Basically, both teams are beating up the NFC south, and it appears to me their DVOA is over-benefiting from it.  But let's see how things play out.
 
If anyone is curious, here total DVOA rankings for the top half of the NFL through week 7.  I included offense, defense and special teams numbers as well.
 
[tablegrid= DVOA through week 7 for the top 16 teams ] TEAM TOTAL LAST TOTAL RANK OFFENSE OFF. DEFENSE DEF. S.T. S.T.     DVOA WEEK DAVE   DVOA RANK DVOA RANK DVOA RANK 1 DEN 48.7% 1 44.2% 1 31.4% 1 -18.2% 2 -0.9% 20 2 BAL 28.5% 2 26.1% 2 16.2% 4 -8.8% 6 3.5% 8 3 GB 26.1% 4 25.6% 3 19.9% 2 -4.4% 10 1.8% 14 4 SEA 22.7% 3 21.0% 4 15.5% 6 -8.1% 7 -0.9% 21 5 IND 19.4% 13 18.1% 5 8.7% 8 -5.6% 9 5.1% 4 6 PHI 14.5% 6 12.6% 7 -2.9% 18 -6.3% 8 11.1% 1 7 DAL 13.7% 10 12.2% 8 16.5% 3 1.3% 19 -1.6% 22 8 SD 13.2% 8 13.2% 6 15.7% 5 4.7% 23 2.1% 11 9 KC 9.9% 14 7.2% 9 6.6% 10 -1.4% 14 2.0% 13 10 MIA 6.8% 17 4.8% 11 5.6% 11 -11.1% 3 -9.9% 32 11 NE 5.7% 11 6.2% 10 1.8% 16 0.1% 16 3.9% 6 12 DET 5.2% 7 4.0% 12 -11.3% 26 -24.9% 1 -8.4% 31 13 CIN 2.7% 5 2.2% 13 1.9% 15 1.7% 20 2.5% 10 14 PIT 2.2% 20 2.0% 14 5.6% 12 7.4% 26 4.0% 5 15 ARI 0.6% 19 -0.5% 15 -11.2% 25 -10.0% 5 1.8% 15 16 BUF -0.4% 18 -1.0% 16 -16.5% 29 -10.5% 4 5.5% 3  [/tablegrid]

 

[mauf]

What is weird to me is that they went down by 0.1 despite the win. I get that their DVOA got worse due to the nature of the win, but does that really offset the fact that going into the game there was a non-insignificant chance the Pats would lose the game? Seems to me that actually achieving the win would offset the drop in DVOA. But maybe I'm wrong.

I can't imagine DVOA gave the Jets more than a 20% chance of pulling the upset. So DVOA marked them up 0.2 wins for avoiding the upset, but marked them down 0.3 wins for the way they won. That sounds about right to me.

 

[Briz]

I think I will start this weeks update with a quote from George E. P. Box:  "Essentially, all models are wrong, but some are useful."  
 
The teams ranked 2, 3, 5, 6, 7 and 8 by DVOA through week 7 all lost in week 8.  Was week 8 an anomaly or were these teams over valued by the model?  Anyway, here is the results:  http://www.footballoutsiders.com/stats/playoffodds
 
The Pats are now projected to have 9.9 wins and win the division is 50% of the simulations.  They make the playoffs 60% of the time.  In 16% of the simulations, they earned a bye.
 
[tablegrid= Week 8 DVOA playoff odds ]AFC East               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE NE 7.30% 9.9 50.40% 16.10% 10.00% 60.40% -1.90% MIA 9.00% 9 26.50% 6.40% 13.00% 39.50% 6.60% BUF 3.10% 8.9 23.10% 7.40% 10.90% 34.00% 7.30% NYJ -19.20% 3.7 0.00% 0.00% 0.00% 0.00% -0.50%                 AFC North               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE BAL 26.10% 10.2 48.80% 34.70% 21.30% 70.00% -17.70% PIT 10.20% 9.4 24.20% 8.40% 21.80% 46.10% 13.90% CIN 6.00% 8.7 19.70% 10.30% 20.60% 40.30% 5.40% CLE -0.50% 8.5 7.20% 4.50% 13.70% 20.90% 2.10%                 AFC South               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE IND 11.00% 9.8 88.10% 18.50% 0.80% 88.80% -8.40% HOU -4.80% 7.9 11.50% 1.50% 5.10% 16.60% 8.90% TEN -10.20% 5.4 0.40% 0.00% 0.10% 0.50% -0.90% JAC -25.50% 3.6 0.00% 0.00% 0.00% 0.00% 0.00%                 AFC West               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE DEN 47.70% 12.6 89.50% 85.30% 8.40% 97.90% 0.90% KC 15.20% 9.3 7.00% 4.90% 40.50% 47.50% 9.10% SD 8.80% 9.1 3.50% 2.10% 33.80% 37.40% -24.90% OAK -19.00% 2.6 0.00% 0.00% 0.00% 0.00% 0.00%                 NFC East               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE DAL 8.50% 10.4 50.50% 39.50% 27.60% 78.10% -12.00% PHI 12.40% 10.1 45.20% 27.00% 23.30% 68.50% -7.90% NYG -4.90% 7.4 2.70% 0.80% 4.80% 7.60% 2.10% WAS -6.50% 7 1.60% 0.10% 2.80% 4.40% 3.60%                 NFC North               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE DET 6.10% 10.7 56.20% 42.20% 29.10% 85.30% 16.80% GB 14.80% 10.2 42.40% 29.80% 31.40% 73.70% -15.30% CHI -9.00% 7 1.30% 0.50% 4.40% 5.70% -8.50% MIN -24.50% 6.3 0.20% 0.10% 0.80% 1.00% 0.60%                 NFC South               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE NO 5.90% 8.2 63.80% 2.40% 0.40% 64.20% 32.50% CAR -14.70% 6.5 25.50% 0.30% 0.40% 25.90% -22.50% ATL -2.80% 6 10.70% 0.10% 0.20% 10.80% -9.60% TB -43.70% 3.2 0.10% 0.00% 0.00% 0.10% -0.60%                 NFC West               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE ARI -1.20% 10.2 49.90% 38.10% 28.30% 78.20% 14.00% SEA 18.20% 9.8 37.40% 15.90% 30.50% 67.90% 7.60% SF -3.70% 8.4 12.50% 3.10% 15.90% 28.50% 0.40% STL -24.30% 4.9 0.10% 0.00% 0.10% 0.20% -1.20% [/tablegrid]
 
Playoff seeds:
 
AFC
1) Den 12.6 wins
2) Balt 10.2 wins
3) NE 9.9 wins
4) Indy 9.8 wins
5) Pit 9.4 wins
6) KC 9.3 wins
 
NFC
1) Det 10.7 wins
2) Dal 10.4 wins
3) Ari 10.2 wins
4) NO 8.2 wins
5) GB 10.2 wins
6) Phi 10.1 wins
 
A couple notes:
 
1) The model continues to like the AFC north.  Every team is projected to have 8+ wins.  Is this because they are matched up against the NFC south?
2) Seattle is still on the outside looking in even though they have the 3rd best weighted DVOA.
3) NYJ join Jac and Oak at 0.00% for making the playoffs.
4) The rest of the AFC East simulated out to be Pats 10-6, Mia 9-7 and Buf 9-7.  So 3 of 4 teams above .500
5) The NFC west is projected to have 3 teams with 8+ wins.  But only one team is projected to make the playoffs from that division right now, Ari.
6) A team will need 10 wins to make the NFC playoffs unless they are from the NFC south.  
7) The tiebreakers might get interesting in the NFC as there is only a 0.6 win differential between the #1 seed and the #6 seed (excluding NO).

 

[DJnVa]

So, the Patriots going 2-0 the last 2 weeks has lowered their playoff odds (and moved AFC East chances from 65% to 50%) and taken them from 9.6 to 9.9 expected wins?

 

[Briz]

Somehow, yes.
 
Edit:  Hence the quote at the top of my update.  Their model really likes the AFC north.  Again, that might just be because they are matched up with the NFC south.  But Balt has the second highest weighted DVOA. It leaves me scratching my head a bit.   
 
DVOA rankings:
 
[tablegrid= Week 8 DVOA rankings (top 16 teams only) ] TEAM TOTAL LAST WEIGHTED RANK OFFENSE OFF. DEFENSE DEF. S.T. S.T.     DVOA WEEK DVOA   DVOA RANK DVOA RANK DVOA RANK 1 DEN 47.60% 1 47.70% 1 31.80% 1 -17.50% 2 -1.70% 22 2 BAL 26.00% 2 26.10% 2 11.30% 7 -10.70% 5 4.00% 6 3 SEA 18.80% 4 18.20% 3 8.30% 10 -10.20% 6 0.30% 17 4 KC 15.10% 9 15.20% 4 5.10% 13 -5.20% 9 4.80% 5 5 GB 15.00% 3 14.80% 5 14.90% 5 1.80% 18 1.90% 13 6 PHI 12.40% 6 12.40% 6 -3.70% 23 -6.00% 8 10.10% 1 7 IND 11.00% 5 11.00% 7 9.70% 8 3.70% 21 4.90% 3 8 PIT 10.00% 14 10.20% 8 15.90% 4 8.40% 26 2.50% 11 9 SD 8.80% 8 8.80% 10 16.60% 3 9.20% 27 1.40% 14 10 MIA 8.80% 10 9.00% 9 5.20% 12 -11.90% 4 -8.40% 32 11 DAL 8.50% 7 8.50% 11 12.90% 6 2.90% 19 -1.50% 21 12 NE 7.00% 11 7.30% 12 7.50% 11 5.40% 22 4.90% 4 13 CIN 6.70% 13 6.00% 14 0.30% 16 -3.10% 12 3.30% 9 14 DET 6.20% 12 6.10% 13 -10.80% 26 -23.60% 1 -6.60% 31 15 NO 5.50% 21 5.90% 15 18.00% 2 13.70% 30 1.20% 16 16 BUF 3.20% 16 3.10% 16 -16.60% 28 -13.70% 3 6.10% 2  [/tablegrid] 
 
Also, I'm curious how many AFC coaches/teams would say they are more concerned with Balt, KC, Indy, Pitt, SD and MIA (bolded for effect) than the healthy-Gronk-fueled Patriots they saw on Sunday.  All of those AFC teams are ranked above NE by DVOA.

 

[pokey_reese]

For all those "WTF?"ers out there, this page has a good explanation of why DVOA doesn't like the Pats, but also why that could rapidly change:
 
http://www.footballoutsiders.com/dvoa-ratings/2014/week-8-dvoa-ratings
 
TLDR: The Pats schedule has been ridiculously easy thus far, so now that the opponent adjustments are pretty set, we have basically beaten one (slightly) above-average team (Cincy), lost badly to the one good team we faced (KC), lost to MIA who is about the equivalent of the Pats by DVOA, and the rest of the wins have been against some of the worst offenses in the league. 
 
However, much like college football rankings, now the Pats are entering a really difficult part of the schedule.  Wins against any of the next 4 opponents (esp. DEN) could rapidly move us up in the rankings.  Basically, we have been lucky to skate to a 6-2 record, but now we have to actually prove something.

 

[dcmissle]

The Bears and Jets are dreadful.  Miami and Buffalo have not fallen off the earth as some imagined.  So I'm not surprised by the results in terms of playoff odds and projected wins.

 

[Silverdude2167]

I am surprised how high they still are on Baltimore. They still have yet to beat a good team.

 

[MainerInExile]

So, the Patriots going 2-0 the last 2 weeks has lowered their playoff odds (and moved AFC East chances from 65% to 50%) and taken them from 9.6 to 9.9 expected wins?

I don't get it at all.  In the past one week, they won an actual game, and picked up exactly 0.4 expected wins.  How does that work?

 

[epraz]

If the projection had them beating the Bears 60% of the time, then it already credited them with .6 of a win.  Getting the win bumps that up to 1 and adds ,4 to the projection.  

 

[MainerInExile]

If the projection had them beating the Bears 60% of the time, then it already credited them with .6 of a win.  Getting the win bumps that up to 1 and adds ,4 to the projection.  

Oh well, sure, if you're just going to like, answer my question.  I'm going to slink away into a corner now.

 

[Stitch01]

Pats are going to be in a weird spot with DVOA because 1) half the sample is the first four games, where the Pats were mediocre to bad and 2) the weighted version is going to put extra weight on what DVOA is likely going to grade as a terrible performance against the Jets given opponent adjustments and the performance against Chicago that's likely not as good as the scoreboard looks, particularly on defense, given Chicago moved the ball and scored in garbage time.

 

[DanoooME]

 
 
A couple notes:
 
1) The model continues to like the AFC north.  Every team is projected to have 8+ wins.  Is this because they are matched up against the NFC south?
2) Seattle is still on the outside looking in even though they have the 3rd best weighted DVOA.
3) NYJ join Jac and Oak at 0.00% for making the playoffs.
4) The rest of the AFC East simulated out to be Pats 10-6, Mia 9-7 and Buf 9-7.  So 3 of 4 teams above .500
5) The NFC west is projected to have 3 teams with 8+ wins.  But only one team is projected to make the playoffs from that division right now, Ari.
6) A team will need 10 wins to make the NFC playoffs unless they are from the NFC south.  
7) The tiebreakers might get interesting in the NFC as there is only a 0.6 win differential between the #1 seed and the #6 seed (excluding NO).

 
 

For all those "WTF?"ers out there, this page has a good explanation of why DVOA doesn't like the Pats, but also why that could rapidly change:
 
http://www.footballoutsiders.com/dvoa-ratings/2014/week-8-dvoa-ratings
 
TLDR: The Pats schedule has been ridiculously easy thus far, so now that the opponent adjustments are pretty set, we have basically beaten one (slightly) above-average team (Cincy), lost badly to the one good team we faced (KC), lost to MIA who is about the equivalent of the Pats by DVOA, and the rest of the wins have been against some of the worst offenses in the league. 
 
However, much like college football rankings, now the Pats are entering a really difficult part of the schedule.  Wins against any of the next 4 opponents (esp. DEN) could rapidly move us up in the rankings.  Basically, we have been lucky to skate to a 6-2 record, but now we have to actually prove something.

 
This is the reverse issue for Seattle.  They've had one of the toughest schedules so far (6th), so if they win the next two games at home against Oakland and the Giants, their DVOA is probably going to go down despite ending up 6-3.  Then they have the schedule gauntlet to run of @KC, Ariz, @SF, @Phi, SF, @Ariz before ending at home against St. Louis.  FO says this is the 23rd toughest schedule the rest of the way, but that's because Ariz and SF are ranked so low (18 and 20 respectively).

 

[lexrageorge]

Pats are going to be in a weird spot with DVOA because 1) half the sample is the first four games, where the Pats were mediocre to bad and 2) the weighted version is going to put extra weight on what DVOA is likely going to grade as a terrible performance against the Jets given opponent adjustments and the performance against Chicago that's likely not as good as the scoreboard looks, particularly on defense, given Chicago moved the ball and scored in garbage time.

 
That is clearly an issue with the methodology being used, as those drives were completely meaningless when it comes to judging the effectiveness of the NE defense.  However, I will agree over a 16 game season those blips tend to be smoothed out, and that by themselves the Pats victories against Buffalo, the Yets, and the Bears are not necessarily any more predictive than the Miami game when it comes to determining how well the Pats will do when they play Denver and Indy. 

 

[coremiller]

 
That is clearly an issue with the methodology being used, as those drives were completely meaningless when it comes to judging the effectiveness of the NE defense. 

 
This has long been a bone of contention for those who have followed FO/DVOA closely for a while.  FO claims that garbage time performance still has some predictive value (they say they have run the model both with and without this data and the model is more predictive with it), just not as much as regular performance, and so they include it in the model but give it less weight (I think Aaron said it's worth about half of regular performance).
 
The issue many people complain about is not using garbage time data at all, but where the garbage time baselines are set.  The way DVOA works is by comparing a team's performance (based on the success-rate concept) to a baseline, which is how an average team performs in the identical down/distance (and sometimes score?) situation.  But the garbage time sample is skewed because good teams are going to be ahead by a lot much more often than bad teams, and bad teams will be behind by a lot much more often than good teams.  So the actual data for "average performance on 3rd and 6 from midfield when trailing by 21+ points" does not reflect how an average-quality team would perform if you created an artificial scenario in which they had the ball on 3rd and 6 at midfield down 21+ points, because the teams that tend to be behind by 21+ are generally much worse than average.
 
Nobody at FO has ever indicated (that I'm aware of, at least) that they correct or adjust for this issue.  Since DVOA is a black box, maybe they do, but nobody knows.  If they don't, the result is that the baselines for garbage time performance are likely too high for teams that are ahead and too low for teams that are behind.  

 

[tims4wins]

Very curious to see how this changes the Pats mean wins and odds to make the playoffs. Those Miami and KC losses - while not pretty to watch - are less and less bad on paper as each week goes by.

 

[tims4wins]

Pats gained nearly a full win and 18% increase in playoffs. If they go into Indy and win I would imagine they would gain close to another win and get around 90% to make it.

 

[Ed Hillel]

Pats gained nearly a full win and 18% increase in playoffs. If they go into Indy and win I would imagine they would gain close to another win and get around 90% to make it.

 
So they had the odds of Denver winning in NE at nearly/over 90%?

 

[tims4wins]

 
So they had the odds of Denver winning in NE at nearly/over 90%?

 
Well I think the 0.9 increase is a combination of beating Denver, plus the increase in the Pats DVOA projected over the rest of the schedule. So if they had the game as 50-50, the Pats picked up a half a win from the game and another 0.4 for the rest of the year. Something like that.

 

[Ed Hillel]

I'm an idiot, I misread a smililar post above. I apologize.

 

[Briz]

Updated through week 9:  http://www.footballoutsiders.com/stats/playoffodds
 
Pats now are projected to have 10.8 wins (+0.9 as Tims4wins stated above), win the division 60% of the time and make the playoffs 78% of the time (18% increase from last week).  They earn a bye in 33% of the simulations.
 
[tablegrid= FO DVOA Playoff odds through week 9 ]AFC East               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE NE 14.8% 10.8 59.8% 32.6% 18.5% 78.2% 17.9% MIA 19.6% 9.8 29.1% 12.4% 26.7% 55.8% 16.3% BUF 3.9% 8.9 11.1% 6.1% 17.2% 28.4% -5.6% NYJ -17.8% 3.4 0.0% 0.0% 0.0% 0.0% 0.0%                 AFC North               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE PIT 13.9% 10 44.6% 12.5% 19.8% 64.3% 18.3% BAL 24.2% 9.8 26.0% 14.2% 27.4% 53.5% -16.6% CIN 4.4% 8.8 23.5% 9.9% 17.4% 40.9% 0.6% CLE -5.1% 8.4 5.9% 3.2% 11.2% 17.1% -3.8%                 AFC South               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE IND 14.1% 10.5 97.7% 34.4% 0.1% 97.9% 9.0% HOU -12.2% 7.2 2.1% 0.2% 2.3% 4.5% -12.1% TEN -14.9% 5.2 0.1% 0.0% 0.1% 0.2% -0.3% JAC -23.0% 3.6 0.0% 0.0% 0.0% 0.0% 0.0%                 AFC West               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE DEN 36.4% 11.5 77.1% 63.4% 14.6% 91.7% -6.2% KC 15.3% 9.6 20.3% 10.8% 34.1% 54.4% 6.9% SD -1.8% 8.3 2.5% 0.5% 10.6% 13.2% -24.2% OAK -20.3% 2.6 0.0% 0.0% 0.0% 0.0% 0.0%                 NFC East               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE PHI 15.3% 10.8 72.5% 41.7% 13.9% 86.4% 18.0% DAL 4.7% 9.5 25.7% 16.2% 32.7% 58.4% -19.7% NYG -7.9% 6.8 1.2% 0.1% 2.4% 3.6% -4.0% WAS -8.6% 6.4 0.6% 0.0% 1.3% 1.9% -2.5%                 NFC North               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE DET 4.8% 10.4 50.8% 32.8% 31.2% 82.1% -3.2% GB 17.3% 10.2 47.3% 29.5% 29.7% 77.0% 3.3% CHI -7.7% 7 1.5% 0.4% 5.5% 7.0% 1.3% MIN -17.2% 6.9 0.4% 0.1% 2.3% 2.7% 1.7%                 NFC South               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE NO 10.2% 8.9 82.7% 4.6% 0.3% 83.0% 18.8% CAR -12.5% 6.1 10.8% 0.1% 0.5% 11.3% -14.6% ATL -5.1% 5.8 6.4% 0.0% 0.1% 6.6% -4.3% TB -35.9% 3.5 0.1% 0.0% 0.0% 0.1% 0.0%                 NFC West               Team WEI DVOA Mean Wins DIV BYE WC TOT CHANGE ARI 1.3% 11.1 69.3% 60.1% 24.7% 93.9% 15.7% SEA 15.1% 9.8 27.6% 13.8% 43.1% 70.7% 2.8% SF -7.2% 7.6 2.8% 0.5% 11.2% 14.0% -14.5% STL -19.4% 6 0.3% 0.0% 1.0% 1.3% 1.0%  [/tablegrid] 
 
Playoffs seeds based on simulated wins:
 
AFC 
1) Den @ 11.5 wins
2) NE @ 10.8 wins
3) Indy @ 10.5 wins
4) Pit @ 10 wins
5 and 6) Mia and Balt @ 9.8 wins
 
NFC
1)  Ari @ 11.1 wins
2) Phi @ 10.8 wins
3) Det @ 10.4 wins
4) NO @ 8.9 wins
5) GB @ 10.2 wins
6) SEA @ 9.8 wins
 
 

 

[Briz]

DVOA of the top 16 teams according to FO:  http://www.footballoutsiders.com/dvoa-ratings/2014/week-9-dvoa-ratings
 
[tablegrid= DVOA through Week 9 ] TEAM TOTAL LAST WEIGHTED RANK OFFENSE OFF. DEFENSE DEF. S.T. S.T.     DVOA WEEK DVOA   DVOA RANK DVOA RANK DVOA RANK 1 DEN 36.9% 1 36.4% 1 26.6% 1 -15.6% 3 -5.3% 26 2 BAL 23.0% 2 24.2% 2 8.8% 10 -6.5% 9 7.7% 2 3 MIA 19.8% 10 19.6% 3 6.7% 13 -19.3% 2 -6.2% 31 4 SEA 17.5% 3 15.1% 7 8.1% 12 -12.2% 5 -2.8% 23 5 GB 16.7% 5 17.3% 4 16.9% 3 2.1% 16 1.9% 12 6 PHI 14.0% 6 15.3% 5 -2.6% 18 -7.1% 8 9.6% 1 7 NE 14.0% 12 14.8% 8 9.1% 9 2.3% 17 7.3% 3 8 IND 13.8% 7 14.1% 9 9.9% 5 1.9% 15 5.8% 5 9 PIT 13.5% 8 13.9% 10 20.6% 2 6.8% 27 -0.3% 17 10 KC 12.8% 4 15.3% 6 8.4% 11 -0.2% 12 4.3% 6 11 NO 7.9% 15 10.2% 11 15.5% 4 9.4% 29 1.9% 13 12 DET 6.2% 14 4.8% 12 -10.0% 25 -22.8% 1 -6.6% 32 13 CIN 6.1% 13 4.4% 14 3.6% 15 1.1% 14 3.5% 8 14 BUF 4.1% 16 3.9% 15 -15.4% 27 -13.3% 4 6.1% 4 15 DAL 3.1% 11 4.7% 13 9.2% 8 4.6% 22 -1.5% 22 16 ARI 1.2% 18 1.3% 16 -9.5% 24 -10.1% 6 0.6% 15  [/tablegrid] 
 
 

 

[Briz]

A few observations/comments:
 
AFC side:
1)  Baltimore, though last place in their division, still makes the playoffs according to the sims.  
2)  Mia is actually the biggest winner of the week jumping up to #3 in the DVOA rankings.  FO states their game against SD was the current best single game of the year with a 117% DVOA. (article above)
3)  All the teams in the AFC North are projected to be 8-8 or better
4)  Pit is now just ahead of Balt for the NFC north.
5) NYJ, Jac and Oak are basically eliminated from the playoffs
 
NFC Side:
1)  Sea is back to being in the playoffs
2)  The NFC north is looking interesting between Det and GB according to how the simulations played out.
3)  For the NFC, 10 wins still seems necessary to make the playoffs unless you are in the NFC south.
4)  NO appears to be taking control of the NFC south
5) TB is the only NFC team with a 0.0% for playoff odds

 

[Silverdude2167]

I am not loving DVOA this year.
 
I do not understand the love of Baltimore. They have only beaten bad teams. Miami's blow out SD (#9 team last week) is some how more impressive than a blowout out of Denver (#1 team) in terms of DVOA gained.
It even graded Denver's Defense as a ~-10 for the Pats game, a defense that gave up 36 points.

 

[coremiller]

I am not loving DVOA this year.
 
I do not understand the love of Baltimore. They have only beaten bad teams. Miami's blow out SD (#9 team last week) is some how more impressive than a blowout out of Denver (#1 team) in terms of DVOA gained.
It even graded Denver's Defense as a ~-10 for the Pats game, a defense that gave up 36 points.

 
Are you really surprised that a 37-0 victory over a good team with a 441-178 yardage disparity and 4-0 turnover margin came out as more impressive than a 43-21 victory over a great team where the winning team got outgained 472-398 and had a 2-1 turnover margin?
 
Denver's defense gave up 36 points, but b/c of turnovers and special teams New England had great field position all game.  New England's starting field position was their 37-yard line, and their scoring drives started on the Den 39 (held to FG), NE 22, Den 34, NE 43, NE 20, Den 10, NE 20.  The points allowed don't really show how well the Denver defense played, which was neither great nor terrible.  Meanwhile, Denver had drives of 52, 72, and 59 yards that resulted in zero points (missed FG and two turnovers on downs) -- DVOA still gives Denver credit for their positive plays on those drives, even though the scoreboard doesn't.

 

[Silverdude2167]

I am surprised, but I just realized that I am assuming that the opponent is taken into account and Denver was rated so much higher than SD that Miami would not receive larger boost than NE. And even with those factors saying Denver's D played well is incorrect.

 

[dcmissle]

Pats have been DVOA darlings for years, to the point where folks outside here accused Schatz of putting his thumb on the scale, which would be stupid, of course.
 
Halfway through a meh year (that kicked off miserably) and DVOA suddenly not so good.
 
Adorable.

 

[Stitch01]

Which thread are you reading where folks here are down on DVOA because of where the Pats are ranked?  I think I missed that thread, can you link it?

 

[dcmissle]

"Why aren't the Pats rated higher? Why are the Ravens rates so highly? ..."

I am just flummoxed why folks get exercised by these FO's rankings.

Do they offer useful insights? Yes. Are the Baltimore Ravens the second best team in the League half way through the season. Nope. And there is no sensible argument that they are.

 

[Silverdude2167]

It is a good model that in my opinion has had some weird results this season.

Those questions are fair to ask. And it was not why are the Pats not rated higher. It was why did Miami gain more than the Pats for their respective wins.

 

[coremiller]

Baltimore's ranking is mostly due to their excellent special teams: DVOA ranks their ST 2nd, with 7.7%.  With average special teams they would be about 6th or so in DVOA, which is similar to where Simple Rating System has them (7th).  
 
FWIW, maybe it's because John Harbaugh used to be a special teams coach, but Baltimore under him generally have very good special teams.  They were 3rd last year, and 1st the year before that.  

 

[Stitch01]

I see a pretty even-handed and useful discussion about DVOA, but what do I know.
 
Im sort of surprised the weighted DVOA figure is still so high on Seattle, particularly on offense.

 

[lexrageorge]

Miami's defense does get a lot of love from DVOA.  Interestingly, they've also allowed the 3rd fewest points in the NFL, so there's something to it.  
 
The Pats defense is not getting enough love from DVOA.  The Bears drives were meaningless, and the Denver turnovers (both the INT's and the turnovers on downs) were more meaningful the total yards given up during garbage time.  However, the Pats D did have two horrendous games at the beginning of the year, and those do count.  A win over Indy would probably help a whole lot. 

 

[snowmanny]

"Why aren't the Pats rated higher? Why are the Ravens rates so highly? ..."

I am just flummoxed why folks get exercised by these FO's rankings.

Do they offer useful insights? Yes. Are the Baltimore Ravens the second best team in the League half way through the season. Nope. And there is no sensible argument that they are.

You seem to be responding to a post from October 24th.

 

[TomTerrific]

Based on comments Schatz et al. have made, and based on what they cite as the initial source of their thinking, my impression is that FO treat football as a large Markov state space and they're just trying to estimate the transition probabilities (which then allows them to make estimates of how "good" a team is, etc., etc.).
 
I think the fundamental problem is that, while this extended Palmer-and-Thorn type of method works for baseball, it's just not a great model for football. Especially for a sport where game plan and adjustments to the opponents tactics have such a dramatic impact. Think of a Markov process where the probabilities keep changing as coaches react and adjust--that strikes me as a system so complex and opaque that traditional estimation methods with the limited amount of play data we have are totally overmatched. (Never mind the fact that FO like to give individual player contribution scores, even when they themselves admit that football analysis just isn't reductive in that kind of way--this is the same point others have noted earlier when talking individual vs. team.)
 
I respect the fact that they have at least done some predictive comparisons against other metrics, such as point differential, etc. But that's such a low bar in the first place that it doesn't really mean that much IMHO. It shows they have gleaned some  additional information from their method beyond those other metrics, but I'm increasingly convinced that that amount of additional info is slight.
 
That being said, I do enjoy reading their site and find their articles entertaining.