2009 NFL Week 6 Beatpath Power Rankings

Kenneth here! But first, here’s a note from our benevolent overlord:

Note from the editor! – We have a slightly new format of beatpath rankings this week, all in the final column. First, we have a new score that will be much more useful. Edgepower is a measure of each team’s Beatpath graph strength from week to week. Each arrow in a graph is a game. A team is connected to games if the games are in that team’s beatpaths. The tiebreaker is to take the games below a team, and subtract the games above a team. A team is considered as having maximum power if every game in the season is below them. So we normalize by taking the tiebreaker score and dividing it by the games played so far, and then normalizing it on a score of 0-100. Only one team can ever have 100, and it’s rare – I believe New England might have had 100 briefly a couple of years ago. I will have to check and see if Detroit had a score of 0 last year. Finally, the column has an indication of all the possible rankings a team can have each week if the Beatpath graph is respected. Some teams have more possible ranking slots than others, due to the nature of the teams they have played.

Got it? Good, let’s get to the inane commentary.

Rank Team Notes Last Week EdgePower

1

(Beat SD) Ho hum. Lots of interesting things happened underneath the Broncos, but it’s not of much interest to the kings on high.

1

83.78

(50 – 0) / 74

#1 – #7

2

(Beat TEN) I assure you this movement has little to do with the fact that they dropped almost 60 on the Titans. Pretty much, nothing, really. The big news for New England was that the Bills beat the Jets, which cut away a long beatloop of ATL->MIA->NYJ->NE->ATL and instead replaced it with BUF->NYJ->NE->BUF. So, the Patriots already looped away their loss to the Jets, but the new loop brings back their win over the Falcons. And that’s the reason for their big jump.

10

75.68

(39 – 1) / 74

#2 – #11

3

(Beat BAL) Meanwhile, the Vikings don’t get much from beating the Ravens–they already had paths to the teams that the Ravens have paths to.

2

72.97

(34 – 0) / 74

#1 – #14

4

(Beat CHI) Falcons beat the Bears, which boosts them up a bit. Being a Bears fan, that loss really hurt; I feel like the teams were mostly even but the Bears gave away their chances.

6

71.62

(34 – 2) / 74

#3 – #13

5

(Beat NYG) Life’s not fair, right? The Saints put a whupping on the Giants but don’t get anywhere in the graph because the Giants don’t bring in as many paths as the Bears (for ATL) or Falcons (for NE).

5

67.57

(26 – 0) / 74

#1 – #19

6

(Bye) One week off and you’re yesterday’s news. Why aren’t you putting up 50, Peyton?

7

66.89

(25 – 0) / 74

#1 – #18

7

(Lost to NO) The Giants do gain a bit this week due to the Raiders victory, but the real reason they go up two is because of two yet-to-be-mentioned teams that cratered this week. FORESHADOWING!

9

61.49

(18 – 1) / 74

#2 – #24

8

(Bye) The 49ers basically switch places with the Bears after the latter also loses to the Falcons. Albeit, not as harshly, which is a feature/limitation of beatpaths. I’ll say this–I feel better about the Bears going forward than the 49ers.

11

61.49

(21 – 4) / 74

#5 – #20

9

(Beat SEA) Another team helped by the Raiders (it boosts the Texans). Meanwhile, beating Seattle does little since…well, we’ll get to that.

17

58.11

(18 – 6) / 74

#7 – #21

10

(Lost to DEN) There’s no real shame in losing to the best team, right? Meanwhile, HOU->CIN created three big beatloops, 2 of which (HOU->CIN->PIT->SD->MIA->NYJ->HOU and HOU->CIN->BAL->SD->MIA->NYJ->HOU) removed 2 beatlosses (PIT and BAL) from the Charger’s ledger. Suddenly, they don’t look so bad! However, given the length of those beatloops, I’m not sure I’d expect that to last.

20

56.76

(11 – 1) / 74

#2 – #25

11

(Lost to ATL) The other big beatloop (HOU->CIN->GB->CHI->SEA->JAC->HOU) helps the Bears a great deal by shedding their beatloss to Green Bay. Of course, they promptly replace it with one below Atlanta, so they fall.

8

54.73

(10 – 3) / 74

#4 – #24

12

(Bye) So the rest of the NFC East lost–two of them to bad teams–and the Cowboys didn’t lose. As a friend of mine pointed out, this might have been the best week Dallas has played yet.

16

53.38

(8 – 3) / 74

#4 – #27

13

(Beat CIN) Beating the Bengals was huge for the Texans, as it let them shed a ton of teams from on top of them. They still lie under the Cardinals, but that’s much better than where they were.

25

53.38

(12 – 7) / 74

#8 – #24

14

(Beat DET) Did you figure it out? That this was one of the teams that dove off a cliff this week? It’s not really their fault–they got hurt by the fact that one of the long beatloops took away their win over Chicago, which was pretty much the only thing keeping them where they were.

4

52.70

(5 – 1) / 74

#2 – #28

15

(Lost to MIN) I really wanted them to win that game. Really, really did.

12

52.70

(7 – 3) / 74

#4 – #26

16

(Beat CLE) Honestly, I thought the Steelers had beaten the Browns once already.

13

52.70

(8 – 4) / 74

#5 – #25

17

(Lost to HOU) Yes, the other massive drop this week. That will happen when you lose beatloops to 3 teams with one loss–one of whom was in the top 5. On the field, I wonder if losing Odom is going to sink them for the rest of the year.

3

51.35

(3 – 1) / 74

#2 – #29

18

(Lost to BUF) Jets lose a beatwin over Houston, so that hurts them. I am curious how you keep the Patriots to 9 points but let the Bills score 16.

15

50.00

(1 – 1) / 74

#2 – #31

19

(Bye) The shorter BUF-NYJ-NE beatloop, along with the massive HOU-CIN ones, means the Dolphins exchange a beatloss to the Chargers for one for the Falcons. A fair trade, if you ask me, and beatloops agrees.

24

48.65

(2 – 4) / 74

#5 – #30

20

(Beat PHI) See, I am convinced that it’s fools gold games like this that convince Al Davis that he’s on to something.

29

47.97

(10 – 13) / 74

#12 – #26

21

(Beat STL) Well, I hope the Jaguars are going to be for real and compete this year, because if they aren’t they really should have let the Rams have that one, just to be nice.

23

45.95

(2 – 8) / 74

#8 – #30

22

(Lost to OAK) Seriously, guys, what happened? Don’t you at some point look at the scoreboard and say, “Hey we’re losing to the RAIDERS”? And the worst part is that they’re probably to beat the Giants at some point this season, creating a beatloop that will make my head asplode.

14

45.27

(8 – 15) / 74

#13 – #27

23

(Lost to GB) It is really quite surprising how much cachet a win over the Redskins gives them. Well, not really; truly it’s the fact that their opposition has been pretty good so far that leaves them this high.

22

45.27

(3 – 10) / 74

#9 – #29

24

(Lost to ARI) The Seahawks are involved in another blowout, but on the other side. This seems to happen a lot. I have a theory that the Seahawks are like that guy who used to play as Scorpion in Mortal Kombat all the time and just do “get over here”+uppercut the entire time. If you didn’t know what to do, you got smoked, and if you did, he had nothing else. I don’t know what the Seahawks’ harpoon-thingy is, but it seems like some teams figure it out and some don’t.

18

44.59

(1 – 9) / 74

#8 – #31

25

(Lost to PIT) We understand. That win over the Bills took a lot out of you. No one expected you to win this week.

21

43.24

(2 – 12) / 74

#9 – #30

26

(Beat TB) Carolina beat the Bucs, which means little, and sees Philly sink, which means a lot. Do you see what happens, Eagles? When you lose to the Raiders you’re not just hurting yourself; other people get hurt too.

19

39.19

(4 – 20) / 74

#15 – #29

27

(Beat NYJ) I guess winning ugly is better than losing ugly, but at some point it would be nice if they just got rid of the ugly part.

27

39.19

(1 – 17) / 74

#13 – #31

28

(Lost to NE) At one point I’m pretty sure the Patriots were just calling plays out of Tecmo Super Bowl. This leads me to believe the Titans problem was that they just kept being unlucky and not picking the exact same play for defense.

28

37.84

(0 – 18) / 74

#14 – #32

29

(Beat WAS) I’m not sure if “cost another coach his job” is really the kind of thing you want in your postseason accomplishments list, but that thing is looking kind of bare right now.

31

35.81

(3 – 24) / 74

#16 – #29

30

(Lost to KC) Losing to the Chiefs just puts them under a whole other heap of teams. We all remember how Zorn wasn’t hired originally as the head coach, right? How he was hired as OC with no coach, and then they didn’t find a coach so they gave it to him? How did we expect this one to turn out?

26

25.00

(2 – 39) / 74

#22 – #30

31

(Lost to JAC) I’ll say this: that Leonard Little pick-6 was fantastic.

30

16.22

(0 – 50) / 74

#25 – #32

32

(Lost to CAR) Meanwhile, no one’s calling for a coaching change in Tampa because…why?

32

12.16

(0 – 56) / 74

#27 – #32

13 Responses to 2009 NFL Week 6 Beatpath Power Rankings

  1. Tom says:

    I like the smiley face in Jacksonville’s EdgePower the best 😉

  2. ThunderThumbs says:

    I know I should fix that, but it’s kind of charming.

    As a historical comparison, DEN’s 83.78 is pretty high for this point in the season. 2008 didn’t have any team get that high until Week 10, when Carolina (!) reached 84.71. They cratered after that, and that score was surpassed by NYJ, TEN (who was propping up NYJ), and IND.

    PIT ended up at 81.68, and ARI ended up at 67.18.

    Detroit ended last year at 17.18 . In 2007, New England was at 100.0 from Week 14 through the end of the season, when they were defeated by the 71.85 Giants.

  3. ThunderThumbs says:

    By the way, one note of explanation. There have actually been 90 games so far this season, but edgepower is based off of the number of games that are in the graph (and haven’t been beatlooped away). Otherwise, you’d have the 2007 NE Patriots get penalized by beatloops that other teams fell into, and all the scores would be compressed towards the middle. Having it based off of the non-beatlooped games (74 this week) is more informative.

  4. JT says:

    The discussion in the previous post about how to resolve beatloops gave me another idea. I’m not sure if it would work, or if it would be a good idea, but what about using the (Beat/Edge)Power numbers to help resolve loops?

    As we’ve seen with some of the other analysis, we’ve made a confidence number, based on the difference in the Power numbers between two opponents. Would it be possible to resolve a loop by removing the link that went most against the Confidence number? The thinking is that whatever game in a loop had the highest Confidence number is most likely the “fluke” game of the loop.

  5. The MOOSE says:

    My understanding is that the confidence number is one teams score minus another’s. And these scores are determined by how the loops are resolved. Using different methods results in different numbers. Therefore you can’t have the scores dependent on the loops and the loops dependent on the scores at the same time.

    Also, regarding DEN’s historical comparison: Using my scoring method DEN is currently at 6.27 (on a scale from -10 to 10). Going back to 2001, as far back as I have processed so far, DEN’s score is slightly below the average for the #1 team through 6 weeks (6.43). The best team in the last 9 years was STL of 2001 who were far and away the best at 7.88 while the second best was CAR of 2003 at 6.99.

    Where DEN is doing well is margin between #1 and #2. NE has a score of 4.55, 1.72 behind. This is the 3rd best rank which suggests DEN is well ahead of the pack. The best was the same STL-01 who were an incredible 3.37 ahead of second place SF. Second is PHI-04 who had a 2.08 lead over DET.

    While the obvious point of comparison would be NE-07, by week 6 NE had just risen to first, and only by a 0.06 margin over GB. Even #3 IND was only 0.41 behind. The top teams were considered equally strong until NE beat IND head-to-head in week 9 and ran away from the pack.

  6. ThunderThumbs says:

    Also remember that for undefeated teams, the difference between scores is really just the difference in their schedules – the strength of other teams throughout the league. The undefeated teams have zero control over this. What’s interesting is putting an undefeated team against an undefeated team. As for the team with the higher score, you can feel more secure that they are good, and it’s true that they have been more tested, and because of them being more prepared, might have a better chance of beating the other team, but really beyond that there is no basis to believe they are better than the other team.

  7. JT says:

    I had a feeling that the idea wouldn’t really work, but just in case I wanted to throw it out there.

  8. […] The Winning Ways of Winners « 2009 NFL Week 6 Beatpath Rankings October 21st, 2009 2008 NFL Picks, […]

  9. ThunderThumbs says:

    Finding an alternate way to break loops is an entire different family of tinkering that what I’ve focused on up to now – better performing tiebreakers. JT, I’m pretty curious about your idea, actually. Once I get more into alternate ways to break loops, I’ll look at that and a few other methods.

  10. Thurhame says:

    I still don’t like your method of breaking loops. If A->B->C->A, in your method, adding the additional games B->D->A changes nothing. Why should B get no benefit from defeating D? There is twice as much evidence that B is stronger than A as there is evidence that A is stronger than B. You say you do this because it is impossible to decide which beatloop to remove from an objective standpoint. The “iterative” method Moose and I prefer solves this by taking away half of each beatloop.

    Feel free to ignore this post if you think my persistence is annoying. You have your opinion; I have mine; we disagree; that’s fine. However, if you like a good argument please reply.

  11. ThunderThumbs says:

    At some point I’ll backtest the iterative approach. I don’t think your persistence is annoying, but no one has been able to explain to me *why* your scenario above means there is twice as much evidence that B is stronger than A. I mean, Houston beat Cincinnati pretty conclusively. Before Houston won, you could argue that CIN really should have beaten them. But they didn’t, and that has to be taken into account. The loop breaking is agnostic at that point and doesn’t make judgments on which of those games are flukey. Just because removing HOU->CIN would be the easiest way to solve it doesn’t mean it’s the way that is most representative of the teams’ quality.

  12. Thurhame says:

    The ‘why’ is because B has two beatpaths to A, while A has only one beatpath to B. Alternatively, B has beatpaths to two games that A lost, while A only owns one game that B lost.

    Interestingly, if those two beatpaths were not of the same length, your method would remove only one of them when it removes A->B. Thus, B would still have one beatpath to A. I agree with this reasoning, which is why I disagree with removing both beatpaths when they are the same length.

    I agree that removing only A->B is the wrong thing to do. So, I don’t do that. The iterative approach removes one A to B beatpath (A->B), and one B to A beatpath (half of B->C->A plus half of B->D->A).

  13. […] effect of the new tiebreaker method seems to have changed the dynamic of the stability chart. Last year, the rankings became more […]

Leave a Reply

Your email address will not be published. Required fields are marked *