Calculated Power Rankings (1 Viewer)

[Berger]

Had some interesting comments after posting this last week. Same formula, updated for this weekends results. Again, the number off to the right is the margin between the team and the team below them. Again, the Colts are so far above the next team, they could lose and remain #1. Comments welcome.

Rank Team Margin

1 Colts 5.61

2 Broncos 1.25

3 Bears 0.11

4 Ravens 0.47

5 Patriots 0.19

6 Giants 0.69

7 Saints 0.28

8 Chiefs 0.78

9 Panthers 0.30

10 Chargers 0.02

11 Jaguars 0.16

12 Seahawks 0.41

13 Vikings 0.25

14 Falcons 0.20

15 Bengals 0.50

16 Rams 0.44

17 Cowboys 0.05

18 Eagles 0.64

19 Redskins 0.17

20 49ers 0.31

21 Jets 0.33

22 Bills 0.39

23 Buccs 0.23

24 Lions 0.13

25 Dolphins 0.25

26 Steelers 0.16

27 Texans 0.33

28 Browns 0.14

29 Titans 0.17

30 Packers 0.91

31 Raiders 0.08

32 Cardinals

 

[phillzphan]

@ eagles # 18

 

[Limp Ditka]

@ eagles # 18

If they beat a team over .500 then they'll move up.

 

[Berger]

@ eagles # 18

Why? Because they were good last year? Or do you think they are too high?

 

[Jeff Pasquino]

Had some interesting comments after posting this last week. Same formula, updated for this weekends results. Again, the number off to the right is the margin between the team and the team below them. Again, the Colts are so far above the next team, they could lose and remain #1. Comments welcome.

Rank Team Margin

1 Colts 5.61

2 Broncos 1.25

3 Bears 0.11

4 Ravens 0.47

5 Patriots 0.19

6 Giants 0.69

7 Saints 0.28

8 Chiefs 0.78

9 Panthers 0.30

10 Chargers 0.02

11 Jaguars 0.16

12 Seahawks 0.41

13 Vikings 0.25

14 Falcons 0.20

15 Bengals 0.50

16 Rams 0.44

17 Cowboys 0.05

18 Eagles 0.64

19 Redskins 0.17

20 49ers 0.31

21 Jets 0.33

22 Bills 0.39

23 Buccs 0.23

24 Lions 0.13

25 Dolphins 0.25

26 Steelers 0.16

27 Texans 0.33

28 Browns 0.14

29 Titans 0.17

30 Packers 0.91

31 Raiders 0.08

32 Cardinals

I see SD, NYJ, Philly and GB as too low.Saints as well, by 1-2 spots.

KC, Wash and Cincy are too high.

Biggest issues are with GB, SD and KC.

My .

 

[Despyzer]

If I am understanding the formula correctly, it seems odd that the Chargers would be ranked as low as tenth because they have only beaten one quality team while losing to two quality teams, yet the Bears are 3rd for beating only one badly-bruised quality team while losing to one of the worst teams in the AFC. If number of wins and losses count so significantly to vault the Bears that much higher, then how do the Panthers and Chiefs rank above San Diego?

 

[BoltzNBrew]

Still a 4 loss team in the top 10?

 

[Aaronl5]

Colts are too high....somebody who actually has a chance to make the superbowl should be ahead of them.

 

[djcolts]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

In that system, the Colts are way ahead of everyone else (they wouldn't even be in 1st if margin of victory was a big factor), Dallas is much lower rated in that system than in a "Pure Points" system (#6 in Pure Points, #17 in ELO CHESS), and teams that just beat bad teams (like the Chargers and Eagles) are lower rated than expected. Oh - and in the ELO CHESS ratings, the Eagles exactly #18 (#11 in Pure Points).

SAGARIN RATINGS

 

[Despyzer]

and teams that just beat bad teams (like the Chargers and Eagles) are lower rated than expected.

Seventh seems about right for the Chargers. At least in that ranking system, there aren't any three or four loss teams in front of them.

 

[Snotbubbles]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.In that system, the Colts are way ahead of everyone else (they wouldn't even be in 1st if margin of victory was a big factor), Dallas is much lower rated in that system than in a "Pure Points" system (#6 in Pure Points, #17 in ELO CHESS), and teams that just beat bad teams (like the Chargers and Eagles) are lower rated than expected. Oh - and in the ELO CHESS ratings, the Eagles exactly #18 (#11 in Pure Points).

The Eagles are 4-4. The Cowboys are 4-4. The Eagles beat the Cowboys. How are the Cowboys ranked ahead of the Eagles?

 

[Berger]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.In that system, the Colts are way ahead of everyone else (they wouldn't even be in 1st if margin of victory was a big factor), Dallas is much lower rated in that system than in a "Pure Points" system (#6 in Pure Points, #17 in ELO CHESS), and teams that just beat bad teams (like the Chargers and Eagles) are lower rated than expected. Oh - and in the ELO CHESS ratings, the Eagles exactly #18 (#11 in Pure Points).

The Eagles are 4-4. The Cowboys are 4-4. The Eagles beat the Cowboys. How are the Cowboys ranked ahead of the Eagles?

Because the Cowblys beat the Panthers, which carries more weight than beating the Cowboys.

 

[djcolts]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.In that system, the Colts are way ahead of everyone else (they wouldn't even be in 1st if margin of victory was a big factor), Dallas is much lower rated in that system than in a "Pure Points" system (#6 in Pure Points, #17 in ELO CHESS), and teams that just beat bad teams (like the Chargers and Eagles) are lower rated than expected. Oh - and in the ELO CHESS ratings, the Eagles exactly #18 (#11 in Pure Points).

The Eagles are 4-4. The Cowboys are 4-4. The Eagles beat the Cowboys. How are the Cowboys ranked ahead of the Eagles?

The margin looks very close. And, the Cowboys' biggest win was against the Panthers, who's in the top 10 in this ranking system. And both teams beat 3 other bad teams. I think that would explain it if I'm correct on how this ranking system works.

 

[Clayton Gray]

I'll throw mine in here. Rankings are based on wins/losses, SOS, points and yards for, and points and yards allowed.

1 - Indianapolis

2 - New England

3 - Denver

4 - NY Giants

5 - Baltimore

6 - Chicago

7 - San Diego

8 - Jacksonville

9 - New Orleans

10 - Dallas

11 - Kansas City

12 - Philadelphia

13 - Seattle

14 - Minnesota

15 - Atlanta

16 - Cincinnati

17 - Carolina

18 - Washington

19 - Pittsburgh

20 - St. Louis

21 - Buffalo

22 - NY Jets

23 - Tampa Bay

24 - Cleveland

25 - Green Bay

26 - Detroit

27 - San Francisco

28 - Houston

29 - Miami

30 - Tennessee

31 - Oakland

32 - Arizona

 

[Snotbubbles]

The margin looks very close. And, the Cowboys' biggest win was against the Panthers, who's in the top 10 in this ranking system. And both teams beat 3 other bad teams. I think that would explain it if I'm correct on how this ranking system works.

Ok. Then how are the Panthers at 4-4 ranked in the top 10 ahead of Seattle or Atlanta? Or how are the Giants ranked so high since their only quality win was against Atlanta who isn't even in the top 10.

 

[Snotbubbles]

How about the Lions ranked ahead of the Dolphins? Beating Chicago at Chicago holds less weight then beating Buffalo and Atlanta at home.

 

[Berger]

The margin looks very close. And, the Cowboys' biggest win was against the Panthers, who's in the top 10 in this ranking system. And both teams beat 3 other bad teams. I think that would explain it if I'm correct on how this ranking system works.

Ok. Then how are the Panthers at 4-4 ranked in the top 10 ahead of Seattle or Atlanta? Or how are the Giants ranked so high since their only quality win was against Atlanta who isn't even in the top 10.

For starters the Panthers beat the Ravens (#4), Saints (#7) and #'s 23 & 28 (worthless). Seattle has beaten #'s 32, 31, 24 (should these even count?), 16, and 6. Without their win vs the Giants (#6), they're not even in the picture.

 

[Ghost Rider]

@ eagles # 18

They are 4-4, and have lost three games in a row, meaning they are the downslide. 18 sounds about right.

 

[Berger]

How about the Lions ranked ahead of the Dolphins? Beating Chicago at Chicago holds less weight then beating Buffalo and Atlanta at home.

Yes But the margin is small.ETA: They did get the biggest boost this week though going up seven spots.

 

[Aaronl5]

Yes---18 seems a little high for the Eagles.....I can't remember when they last won a game....

 

[Snotbubbles]

Yes But the margin is small.ETA: They did get the biggest boost this week though going up seven spots.

The margin might be small but if you're using a rating system that gives weights the strength of a win then beating Detriot (however small) means more then beating Miami. So by making a mistake however small on the strength of a Miami then changes the value of a victory versus Miami, which in turn affects the overall value of the team that beat Miami, which in turn affects that value of a team that beat a team that beat Miami.

For starters the Panthers beat the Ravens (#4), Saints (#7) and #'s 23 & 28 (worthless). Seattle has beaten #'s 32, 31, 24 (should these even count?), 16, and 6. Without their win vs the Giants (#6), they're not even in the picture.

Are we throwing out wins now? Who did the Giants beat? Philly, Was, Dal, TB, Atl, Hou.

 

[Snotbubbles]

Also, why is Jacksonville ranked ahead of Seattle?

Jax wins: Dal, Pit, NYJ, Phi, Ten (best win #17 Dallas) They don't even have a game where as you said, "if they didn't beat..."

 

[Berger]

Yes

But the margin is small.

ETA: They did get the biggest boost this week though going up seven spots.

The margin might be small but if you're using a rating system that gives weights the strength of a win then beating Detriot (however small) means more then beating Miami. Correct

So by making a mistake however small on the strength of a Miami then changes the value of a victory versus Miami, which in turn affects the overall value of the team that beat Miami, which in turn affects that value of a team that beat a team that beat Miami.

Why are you assuming there is a mistake in the calculations?

For starters the Panthers beat the Ravens (#4), Saints (#7) and #'s 23 & 28 (worthless). Seattle has beaten #'s 32, 31, 24 (should these even count?), 16, and 6. Without their win vs the Giants (#6), they're not even in the picture.

Are we throwing out wins now? Who did the Giants beat? Philly, Was, Dal, TB, Atl, Hou.I did not mean to imply that I was throwing out wins. Just that the "points" for beating the bottom two teams and one other low team wont add up to much. The Giants have beaten alot of mid ranked teams. This adds up to more than beating alot of low ranked teams.

 

[Berger]

Also, why is Jacksonville ranked ahead of Seattle?Jax wins: Dal, Pit, NYJ, Phi, Ten (best win #17 Dallas) They don't even have a game where as you said, "if they didn't beat..."

I think we've already done this.

 

[MLBrandow]

Had some interesting comments after posting this last week. Same formula, updated for this weekends results. Again, the number off to the right is the margin between the team and the team below them. Again, the Colts are so far above the next team, they could lose and remain #1. Comments welcome.

Rank Team Margin

1 Colts 5.61

2 Broncos 1.25

3 Bears 0.11

4 Ravens 0.47

5 Patriots 0.19

6 Giants 0.69

7 Saints 0.28

8 Chiefs 0.78

9 Panthers 0.30

10 Chargers 0.02

11 Jaguars 0.16

12 Seahawks 0.41

13 Vikings 0.25

14 Falcons 0.20

15 Bengals 0.50

16 Rams 0.44

17 Cowboys 0.05

18 Eagles 0.64

19 Redskins 0.17

20 49ers 0.31

21 Jets 0.33

22 Bills 0.39

23 Buccs 0.23

24 Lions 0.13

25 Dolphins 0.25

26 Steelers 0.16

27 Texans 0.33

28 Browns 0.14

29 Titans 0.17

30 Packers 0.91

31 Raiders 0.08

32 Cardinals

If I might make a suggestion in calculating your rankings.While it's important for a team to beat good teams, it should also be factored in those victories against crippled teams, i.e. teams without their star player(s).

For example, The Vikings are in my opinion one of the worst teams in the league. None of their victories were against quality teams with their pieces in tact. They've beaten up on hobbled teams (plus the Lions) all season for their wins.

They get exposed by real teams and even bad teams that just happen to show up.

Should the Vikings be rated 13th?

Who have they beaten?

W @WAS 19-16 (19th) - hobbled Clinton Portis, no Shawn Springs, injuries have derailed their season so far.

W CAR 16-13 (9th) - No Steve Smith

L CHI 16-19 (3rd) - this is their most impressive game IMO.

L @BUF 12-17 (22nd) - beaten by a bad team

W DET 26-17 (24th) - beat a bad team.

W @SEA 31-13 (12th) - No Shaun Alexander, injured Matt Hasselbeck in the first half (they were losing at the time)

L NWE 7-31 (5th) - Whooped by a real team with no key injuries.

L @SNF (20th) - Beaten by a bad team.

If you take it more empirically, let's quantify these wins and losses as Quality (beating a good team), and then the same extension with losses (going toe to toe with a good team).

Quality Victories: 0

Quality Losses: 1 (CHI Wk3)

Perhaps one might not completely agree with my definitions of "quality", but with the numbers there, you can draw your own conclusions.

My point is that clearly this is not the 13th best team in the league when you factor in the state of the teams they've beaten against the quality of teams they've been losing against.

I'd really, really like to see the equations you use to derive these rankings, because they seem far more accurate than the rankings I see elsewhere (probably more accurate than I could come up with on my own as well), and those numbers just tease me into wondering what exactly goes into the calculations and with what weight.

The only flaw I see is that I don't believe you've factored in key injuries to certain wins/losses. If they face a healthy WAS, CAR and SEA, I'm almost certain we'd be looking at a 1-7 team instead of a 4-4 team, and personally I don't see the Vikings beating the Lions in their rematch in Week 14.

 

[Maurile Tremblay]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

 

[Berger]

Had some interesting comments after posting this last week. Same formula, updated for this weekends results. Again, the number off to the right is the margin between the team and the team below them. Again, the Colts are so far above the next team, they could lose and remain #1. Comments welcome.

Rank Team Margin

1 Colts 5.61

2 Broncos 1.25

3 Bears 0.11

4 Ravens 0.47

5 Patriots 0.19

6 Giants 0.69

7 Saints 0.28

8 Chiefs 0.78

9 Panthers 0.30

10 Chargers 0.02

11 Jaguars 0.16

12 Seahawks 0.41

13 Vikings 0.25

14 Falcons 0.20

15 Bengals 0.50

16 Rams 0.44

17 Cowboys 0.05

18 Eagles 0.64

19 Redskins 0.17

20 49ers 0.31

21 Jets 0.33

22 Bills 0.39

23 Buccs 0.23

24 Lions 0.13

25 Dolphins 0.25

26 Steelers 0.16

27 Texans 0.33

28 Browns 0.14

29 Titans 0.17

30 Packers 0.91

31 Raiders 0.08

32 Cardinals

If I might make a suggestion in calculating your rankings.While it's important for a team to beat good teams, it should also be factored in those victories against crippled teams, i.e. teams without their star player(s).

For example, The Vikings are in my opinion one of the worst teams in the league. None of their victories were against quality teams with their pieces in tact. They've beaten up on hobbled teams (plus the Lions) all season for their wins.

They get exposed by real teams and even bad teams that just happen to show up.

Should the Vikings be rated 13th?

Who have they beaten?

W @WAS 19-16 (19th) - hobbled Clinton Portis, no Shawn Springs, injuries have derailed their season so far.

W CAR 16-13 (9th) - No Steve Smith

L CHI 16-19 (3rd) - this is their most impressive game IMO.

L @BUF 12-17 (22nd) - beaten by a bad team

W DET 26-17 (24th) - beat a bad team.

W @SEA 31-13 (12th) - No Shaun Alexander, injured Matt Hasselbeck in the first half (they were losing at the time)

L NWE 7-31 (5th) - Whooped by a real team with no key injuries.

L @SNF (20th) - Beaten by a bad team.

If you take it more empirically, let's quantify these wins and losses as Quality (beating a good team), and then the same extension with losses (going toe to toe with a good team).

Quality Victories: 0

Quality Losses: 1 (CHI Wk3)

Perhaps one might not completely agree with my definitions of "quality", but with the numbers there, you can draw your own conclusions.

My point is that clearly this is not the 13th best team in the league when you factor in the state of the teams they've beaten against the quality of teams they've been losing against.

I'd really, really like to see the equations you use to derive these rankings, because they seem far more accurate than the rankings I see elsewhere (probably more accurate than I could come up with on my own as well), and those numbers just tease me into wondering what exactly goes into the calculations and with what weight.

The only flaw I see is that I don't believe you've factored in key injuries to certain wins/losses. If they face a healthy WAS, CAR and SEA, I'm almost certain we'd be looking at a 1-7 team instead of a 4-4 team, and personally I don't see the Vikings beating the Lions in their rematch in Week 14.

Thank you for your comments and constructive critisizm. My goal is to do this without any opinions, homerisms, or qualifications going into the rankings. You are 100% correct that injuries are not factored in. Should I say that without Steve Smith, Carolina is worth 50% of what they are worth with him? How do I arrive at 50% (or whatever % I would use)? And how do I know what full strength is? Are the Jags 90%, 100%, or 110% with Leftwich hurt and Gerrard in?

Unfortunately I cant quantify injuries. Just teams as a whole. Injuries are random (for the most part) and a good team will have a viable backup.

 

[MLBrandow]

Thank you for your comments and constructive critisizm. My goal is to do this without any opinions, homerisms, or qualifications going into the rankings. You are 100% correct that injuries are not factored in. Should I say that without Steve Smith, Carolina is worth 50% of what they are worth with him? How do I arrive at 50% (or whatever % I would use)? And how do I know what full strength is? Are the Jags 90%, 100%, or 110% with Leftwich hurt and Gerrard in?Unfortunately I cant quantify injuries. Just teams as a whole. Injuries are random (for the most part) and a good team will have a viable backup.

You can easily in a few ways.1) Compare the % of team offense without him versus the percentage of team offense with him.2) If you want to be more technical, since offense can be misleading (e.g. - garbage time or something), take the % of wins with him versus the % of wins without him, and then multiply your strength of victory constant by that %. For example, if Carolina only wins 80% as many games without Steve Smith as they do with him, then you multiply the SVC (let's call it K), by 80%. If K were 20 per say, then it would now be 16.If you have a QB Controversy, as with Leftwich v. Garrard, then there is no handicap. The Jags are not down a skill player, because we're not (objectively speaking) even sure that QB1 > QB2.In the case of a pro bowler like Steve Smith, Clinton Portis, Shaun Alexander or Matt Hasselbeck, it is very obvious that the replacement handicaps that team.

 

[Berger]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

 

[djcolts]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

You're correct. That is why Sagarin says to use Pure Points and not ELO Chess when determining point spreads. ELO Chess is more of a "NFL BCS" system.

 

[MLBrandow]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

I agree. Factoring in point spreads are silly. It can be relevant, but there is too much uncertainty.For example, if Team A loses to Team B by 30 points, they definitely got whooped.But if Team A loses to Team B by 3 points, they may have been dominated the whole game or played toe to toe.

 

[Maurile Tremblay]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .

2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

They don't have to be. Trying to take them into account in an objective manner makes things pretty complicated, but FootballOutsiders' DVOA ratings do it.In any event, I'm not saying that any particular system has to do this. I'm just saying that any system that doesn't do it will be pretty easy to outperform against the spread when used as a predictive tool.

 

[Berger]

Thank you for your comments and constructive critisizm. My goal is to do this without any opinions, homerisms, or qualifications going into the rankings. You are 100% correct that injuries are not factored in. Should I say that without Steve Smith, Carolina is worth 50% of what they are worth with him? How do I arrive at 50% (or whatever % I would use)? And how do I know what full strength is? Are the Jags 90%, 100%, or 110% with Leftwich hurt and Gerrard in?Unfortunately I cant quantify injuries. Just teams as a whole. Injuries are random (for the most part) and a good team will have a viable backup.

You can easily in a few ways.1) Compare the % of team offense without him versus the percentage of team offense with him.2) If you want to be more technical, since offense can be misleading (e.g. - garbage time or something), take the % of wins with him versus the % of wins without him, and then multiply your strength of victory constant by that %. For example, if Carolina only wins 80% as many games without Steve Smith as they do with him, then you multiply the SVC (let's call it K), by 80%. If K were 20 per say, then it would now be 16.If you have a QB Controversy, as with Leftwich v. Garrard, then there is no handicap. The Jags are not down a skill player, because we're not (objectively speaking) even sure that QB1 > QB2.In the case of a pro bowler like Steve Smith, Clinton Portis, Shaun Alexander or Matt Hasselbeck, it is very obvious that the replacement handicaps that team.

I'm not totally disagreeing but this would completely ignore who the team is playing against. Car may not have had SS in wk x and lost but they might have lost anyway because they were playing the Colts. Or they may have won in wk x but they may have won anyway because they were playing Arizona.

 

[Berger]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .

2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

They don't have to be. Trying to take them into account in an objective manner makes things pretty complicated, but FootballOutsiders' DVOA ratings do it.In any event, I'm not saying that any particular system has to do this. I'm just saying that any system that doesn't do it will be pretty easy to outperform against the spread when used as a predictive tool.

Exactly. They have Philly at #2, implying they have the worst luck on the planet.

 

[Maurile Tremblay]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .

2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

They don't have to be. Trying to take them into account in an objective manner makes things pretty complicated, but FootballOutsiders' DVOA ratings do it.In any event, I'm not saying that any particular system has to do this. I'm just saying that any system that doesn't do it will be pretty easy to outperform against the spread when used as a predictive tool.

Exactly. They have Philly at #2, implying they have the worst luck on the planet.

They've been pretty unlucky.From the FO commentary:

Let's say Matt Bryant's 62-yard field goal had gone 61 yards and fallen just short. Would that make the Eagles a better team? No. Let's say Plaxico Burress's fumble had rolled five more yards and gone out the end zone for a touchback. Would that make the Eagles a better team? No. Let's say Sean Payton had chosen a couple weeks earlier to unveil his brilliant "Let's have Reggie Bush throw into double coverage" play. Would that make the Eagles a better team? No. But if those things had happened, the Eagles would be 7-1, and no one would give a moment's thought to ranking them third.

That's the strength of DVOA — unlike the won-loss record, DVOA knows that the Eagles are a few funny plays away from being near the top in the standings, rather than in the middle. The point of a game is to win, but whether the team won or lost isn't the most accurate reflection of whether a team played well and how good they are for the season as a whole.

If the question is whether the Eagles or the Chiefs deserve more accolades for their performance this year, maybe your list has the correct answer: the Chiefs.But if the question is which team should be favored on a neutral field if they were to play next week . . . I think the DVOA ratings give a more profitable answer.

 

[Berger]

This looks like a ratings system based more on winning and losing and who they have beaten more than anything else. It looks a little (but not exactly) like the Sagarin's ELO CHESS rating - which only looks at wins and losses and not margin of victory.

That's as good a system as any for handing out awards, but when used to try to predict future results it suffers from at least two major deficiencies.1. Margin of victory does matter. It's more impressive to win by three touchdowns than it is to win by a last-second field goal. And more importantly . . .

2. It doesn't take luck into account. For example, if there are six fumbles in a certain game and Team A recovers five of them but manages only to eek out a two-point win over Team B, it is very likely that Team B is the better team and should be favored in their next matchup. (Recovering fumbles is pretty much all "luck" in the sense that it does not carry over from one game to the next. Teams that recover a high percentage of fumbles in the first half of the season do not recover an above-average percentage of fumbles in the second half of the season, on average.) Also, if Team A wins by two points despite the fact that Team B missed four field goals, Team A should not get much credit (as a predictive tool) for the win, since having the opposing team miss a bunch of field goals against you also does not carry over from one game to the next.

The problem with factoring these things in is that they are too subjective.

They don't have to be. Trying to take them into account in an objective manner makes things pretty complicated, but FootballOutsiders' DVOA ratings do it.In any event, I'm not saying that any particular system has to do this. I'm just saying that any system that doesn't do it will be pretty easy to outperform against the spread when used as a predictive tool.

Exactly. They have Philly at #2, implying they have the worst luck on the planet.

They've been pretty unlucky.From the FO commentary:

Let's say Matt Bryant's 62-yard field goal had gone 61 yards and fallen just short. Would that make the Eagles a better team? No. Let's say Plaxico Burress's fumble had rolled five more yards and gone out the end zone for a touchback. Would that make the Eagles a better team? No. Let's say Sean Payton had chosen a couple weeks earlier to unveil his brilliant "Let's have Reggie Bush throw into double coverage" play. Would that make the Eagles a better team? No. But if those things had happened, the Eagles would be 7-1, and no one would give a moment's thought to ranking them third.

That's the strength of DVOA — unlike the won-loss record, DVOA knows that the Eagles are a few funny plays away from being near the top in the standings, rather than in the middle. The point of a game is to win, but whether the team won or lost isn't the most accurate reflection of whether a team played well and how good they are for the season as a whole.

If the question is whether the Eagles or the Chiefs deserve more accolades for their performance this year, maybe your list has the correct answer: the Chiefs.But if the question is which team should be favored on a neutral field if they were to play next week . . . I think the DVOA ratings give a more profitable answer.

So they have had no calls/plays go their way all season? I'm going to say that if they were the second best team in the NFL these "unlucky" plays wouldn't have mattered, they would be winning reguardless of a last minute fg or fumble.

 

[Maurile Tremblay]

So they have had no calls/plays go their way all season? I'm going to say that if they were the second best team in the NFL these "unlucky" plays wouldn't have mattered, they would be winning reguardless of a last minute fg or fumble.

If I point out that Allen Cunningham was unlucky at the final table this year, would you ask if that meant he never got a good hand? He got some good hands and some bad hands, but they didn't even out over such a small sample of hands.The NFL season is only nine weeks old. Good luck and bad luck have not evened out for every team. The Eagles have had more than their fair share of game-changing bad luck, which means that just using their W-L record will not be a great predictive tool to handicap future games (just like Jamie Gold should not be favored over Allen Cunningham in future poker tournaments).DVOA tries to grade teams on plays that are the result of skill, and to ignore plays that are the result of luck. Some things that are mostly luck include: fumble recovery percentage, field goal percentage by opposing teams, significantly higher or lower conversion rates on third downs versus other downs, etc. (Where "luck" means that there's no game-to-game correlation in these things.)

 

[Berger]

But if the question is which team should be favored on a neutral field if they were to play next week . . . I think the DVOA ratings give a more profitable answer.

Lets look back at last week (I know the sample size makes this meaningless, but I have nothing else to compare this to). Anyway, lets assume we take the team ranked higher as our pick. The results would have been

won				  lost								 Berger	DOVADetroit 30		   Atlanta 14 					 L	LBaltimore 26	 Cincinnati 20 					W	WWashington 22   Dallas 19 					 L	LBuffalo 24			Green Bay 10 					W	LN.Y. Giants 14	Houston 10 					W	WKansas City 31  St. Louis 17 					W	WMiami 31		   Chicago 13 					 L	LNew Orleans 31  Tampa Bay 14 					 W	WJacksonville 37   Tennessee 7 					 W	WSan Francisco 9  Minnesota 3 					  L	LSan Diego 32	 Cleveland 25 					 W	WDenver 31			 Pittsburgh 20 					 W	LIndianapolis 27   New England 20				  W	LSeattle 16		  Oakland 0 					 W	W

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

 

[Maurile Tremblay]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

 

[Berger]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

I'm up for it. I can go back and run my calculations for the earlier weeks. I admittly dont account for bye weeks (although I can think of a quick fix for that) but aside from that I should be able to do this backwards. What week do they start?Or we could use last year.

 

[Berger]

wk 7

1 Colts 0.79

2 Bears 0.59

3 Saints 0.19

4 Broncos 0.38

5 Patriots 0.09

6 Giants 0.08

7 Vikings 0.07

8 Panthers 0.62

9 Bengals 0.19

10 Falcons 0.04

11 Seahawks 0.42

12 Rams 0.53

13 Eagles 0.23

14 Jaguars 0.14

15 Ravens 0.10

16 Buccaneers 0.83

17 Chiefs 0.02

18 Chargers 0.23

19 Jets 0.09

20 Cowboys 0.11

21 Redskins 0.17

22 Bills 0.27

23 49ers 0.13

24 Texans 0.04

25 Steelers 0.71

26 Lions 0.06

27 Titans 0.10

28 Packers 0.11

29 Cardinals 0.32

30 Raiders 0.06

31 Dolphins 0.05

32 Browns

 

[Maurile Tremblay]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

I'm up for it. I can go back and run my calculations for the earlier weeks. I admittly dont account for bye weeks (although I can think of a quick fix for that) but aside from that I should be able to do this backwards. What week do they start?Or we could use last year.

They start in week one. Week one's ratings are based strictly on last year's results. Week two's ratings are based a little bit on week one's results, but mostly on last year's results. Each week the shift is toward emphasizing the current year's results more and the previous year's results less. By week 8, the previous year's results are completely ignored. Week five is the first time strength of schedule is considered (since before that there isn't enough to go on).Link to index of DVOA ratings for each week.

 

[Berger]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

I'm up for it. I can go back and run my calculations for the earlier weeks. I admittly dont account for bye weeks (although I can think of a quick fix for that) but aside from that I should be able to do this backwards. What week do they start?Or we could use last year.

They start in week one. Week one's ratings are based strictly on last year's results. Week two's ratings are based a little bit on week one's results, but mostly on last year's results. Each week the shift is toward emphasizing the current year's results more and the previous year's results less. By week 8, the previous year's results are completely ignored. Week five is the first time strength of schedule is considered (since before that there isn't enough to go on).Link to index of DVOA ratings for each week.

The way mine are calculated would be pretty meaningless early in the nfl season, although I suppose I could splash in last years results to come up with something (I really hate this idea though). Anyway wk 6

Rank Team Margin

1 Bears 0.3345

2 Colts 0.9107

3 Saints 0.0477

4 Panthers 0.4258

5 Seahawks 0.2657

6 Broncos 0.2743

7 Giants 0.1709

8 Patriots 0.1241

9 Rams 0.3489

10 Vikings 0.0012

11 Jaguars 0.1932

12 Falcons 0.1499

13 Eagles 0.7543

14 Ravens 0.0754

15 Bengals 0.0058

16 Chargers 0.2522

17 Redskins 0.2042

18 Jets 0.0138

19 Cowboys 0.0905

20 Bills 0.3182

21 49ers 0.0303

22 Chiefs 0.2626

23 Steelers 0.0828

24 Buccs 0.1412

25 Lions 0.1497

26 Cardinals 0.0059

27 Titans 0.2972

28 Packers 0.0606

29 Dolphins 0.1515

30 Texans 0.3939

31 Browns -

32 Raiders -

 

[Maurile Tremblay]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

I'm up for it. I can go back and run my calculations for the earlier weeks. I admittly dont account for bye weeks (although I can think of a quick fix for that) but aside from that I should be able to do this backwards. What week do they start?Or we could use last year.

They start in week one. Week one's ratings are based strictly on last year's results. Week two's ratings are based a little bit on week one's results, but mostly on last year's results. Each week the shift is toward emphasizing the current year's results more and the previous year's results less. By week 8, the previous year's results are completely ignored. Week five is the first time strength of schedule is considered (since before that there isn't enough to go on).Link to index of DVOA ratings for each week.

The way mine are calculated would be pretty meaningless early in the nfl season, although I suppose I could splash in last years results to come up with something (I really hate this idea though). Anyway wk 6

Rank Team Margin

1 Bears 0.3345

2 Colts 0.9107

3 Saints 0.0477

4 Panthers 0.4258

5 Seahawks 0.2657

6 Broncos 0.2743

7 Giants 0.1709

8 Patriots 0.1241

9 Rams 0.3489

10 Vikings 0.0012

11 Jaguars 0.1932

12 Falcons 0.1499

13 Eagles 0.7543

14 Ravens 0.0754

15 Bengals 0.0058

16 Chargers 0.2522

17 Redskins 0.2042

18 Jets 0.0138

19 Cowboys 0.0905

20 Bills 0.3182

21 49ers 0.0303

22 Chiefs 0.2626

23 Steelers 0.0828

24 Buccs 0.1412

25 Lions 0.1497

26 Cardinals 0.0059

27 Titans 0.2972

28 Packers 0.0606

29 Dolphins 0.1515

30 Texans 0.3939

31 Browns -

32 Raiders -

Let's start with week nine and keep track going forward. You're up 10 to 7.

 

[Berger]

Obviously I cant use the code thing but I would have been 10-4 and DOVA would have been 7-7.

Yep, that is the right way to evaluate the two systems. I submit that if you did that over a full season, DVOA would be the favorite to come out ahead.

I'm up for it. I can go back and run my calculations for the earlier weeks. I admittly dont account for bye weeks (although I can think of a quick fix for that) but aside from that I should be able to do this backwards. What week do they start?Or we could use last year.

They start in week one. Week one's ratings are based strictly on last year's results. Week two's ratings are based a little bit on week one's results, but mostly on last year's results. Each week the shift is toward emphasizing the current year's results more and the previous year's results less. By week 8, the previous year's results are completely ignored. Week five is the first time strength of schedule is considered (since before that there isn't enough to go on).Link to index of DVOA ratings for each week.

The way mine are calculated would be pretty meaningless early in the nfl season, although I suppose I could splash in last years results to come up with something (I really hate this idea though). Anyway wk 6

Rank Team Margin

1 Bears 0.3345

2 Colts 0.9107

3 Saints 0.0477

4 Panthers 0.4258

5 Seahawks 0.2657

6 Broncos 0.2743

7 Giants 0.1709

8 Patriots 0.1241

9 Rams 0.3489

10 Vikings 0.0012

11 Jaguars 0.1932

12 Falcons 0.1499

13 Eagles 0.7543

14 Ravens 0.0754

15 Bengals 0.0058

16 Chargers 0.2522

17 Redskins 0.2042

18 Jets 0.0138

19 Cowboys 0.0905

20 Bills 0.3182

21 49ers 0.0303

22 Chiefs 0.2626

23 Steelers 0.0828

24 Buccs 0.1412

25 Lions 0.1497

26 Cardinals 0.0059

27 Titans 0.2972

28 Packers 0.0606

29 Dolphins 0.1515

30 Texans 0.3939

31 Browns -

32 Raiders -

Let's start with week nine and keep track going forward. You're up 10 to 7.

Ok, when do they update?

 

[MLBrandow]

I like how you try to keep the rankings as objective as possible, but if you don't include some subjectivity, your rankings will always be fundamentally flawed in some way (e.g. - the Vikings I can see clearly in this).

If there were a way to properly rank teams via statistical analysis alone, there'd be no reason to debate because there would be definitive answers.

I do surmise that your rankings are probably the most accurate I've read thusfar, but I think they could be improved.

One question I do have for you and anyone doing rankings.... do you consider injuries in your rankings? Do you rate the Seahawks a lot lower without SA and Hass or keep them high because you know when they are in there, that the Hawks are among the best teams in the league?

Interested in that answer.

 

[Maurile Tremblay]

Let's start with week nine and keep track going forward. You're up 10 to 7.

Ok, when do they update?

Tuesday afternoons. The Week 10 DVOA ratings are up now.

 

[Berger]

Let's start with week nine and keep track going forward. You're up 10 to 7.

Ok, when do they update?

Tuesday afternoons. The Week 10 DVOA ratings are up now.

Ok Games that we differ on:SF@Det

NO@Pit

StL@Sea

and lets just score the games we differ on, mkay?

3-0.

 

[Berger]

Let's start with week nine and keep track going forward. You're up 10 to 7.

Ok, when do they update?

Tuesday afternoons. The Week 10 DVOA ratings are up now.

Ok Games that we differ on:SF@Det

NO@Pit

StL@Sea

and lets just score the games we differ on, mkay?

3-0.

Gameday.

 

[Maurile Tremblay]

Let's start with week nine and keep track going forward. You're up 10 to 7.

Ok, when do they update?

Tuesday afternoons. The Week 10 DVOA ratings are up now.

Ok Games that we differ on:SF@Det

NO@Pit

StL@Sea

and lets just score the games we differ on, mkay?

3-0.

Gameday.

SF, PIT, and SEA won, so you are up 5-1.