Man, there were a lot of upsets this week. We had a streak of some great pick records but this week we were 7-7. The good news is that that means a lot more drama in the graph – lots of changes.
The big news – it appears that the days of AFC dominance over the NFC are over. At least, for this week they are. San Diego defeating Indianapolis has an interesting effect on the graph.
Here’s the graph:
Yes, Minnesota has a beatpath to Indianapolis.
Here are the beatloops. For the first time, we have some season splits. Splits are interesting because they can cancel out beatpaths, or obliterate beatloops which actually add seemingly unrelated beatpaths back into the graph.
Splits: TEN< ->JAC, ATL< ->CAR, and PHI< ->WAS.
Loop: SEA=>TB=>NO=>SEA
Loop: ARI=>SEA=>TB=>ARI
Loop: ARI=>SEA=>SF=>ARI
Loop: SEA=>STL=>NO=>SEA
Loop: ARI=>PIT=>SF=>ARI
Loop: ARI=>PIT=>BAL=>ARI
Loop: DET=>CHI=>PHI=>DET
Loop: ARI=>DET=>TB=>ARI
Loop: HOU=>KC=>SD=>HOU
Loop: SF=>STL=>NO=>SF
Loop: NO=>JAC=>TB=>NO
Loop: CHI=>KC=>MIN=>CHI
Loop: CHI=>KC=>SD=>CHI
Loop: KC=>SD=>DEN=>KC
Loop: CHI=>GB=>SD=>CHI
Loop: CHI=>GB=>MIN=>CHI
Loop: CHI=>PHI=>MIN=>CHI
Loop: ATL=>HOU=>CAR=>NO=>ATL
And, here are the rankings:
| Rank | Team | Notes | Last Week | BeatPower |
1 |
 |
(Beat MIN) Since Green Bay’s win over Minnesota is beatlooped away, as is their win over San Diego. However, the vertical nature of the graph makes Green Bay’s remaining wins of higher strength. |
2 |
100.0(29/29 – 0/29) |
2 |
 |
(Bye) The win over Indianapolis doesn’t look as impressive as it did last week, so New England falls a slot. |
1 |
100.0(30/30 – 0/30) |
3 |
 |
(Beat NYG) Dallas climbs a spot due to Indianapolis falling in the rankings. |
4 |
96.7(29/30 – 1/30) |
4 |
 |
(Lost to DAL) The Giants, despite losing to Dallas, climb a spot due to Indianapolis falling in the rankings, and climb another spot due to Jacksonville also falling in the rankings (by being pushed down by Indianapolis). |
6 |
90.3(28/31 – 3/31) |
5 |
 |
(Lost to PHI) Washington splits the season series with Philadelphia, which means they no longer have a beatpath to Philadelphia. Washington isn’t otherwise hurt though, due to the long beatpath through Detroit to most of the rest of the league. |
7 |
85.5(26/31 – 4/31) |
6 |
 |
(Lost to ARI) Detroit’s loss to Arizona means they lose credit for their beatwin over Tampa Bay, but they still have an alternate beatpath to Tampa Bay through Indianapolis. |
8 |
82.3(25/31 – 5/31) |
7 |
 |
(Lost to GB) Despite losing to Green Bay, Minnesota climbs due to San Diego’s win over Indianapolis. |
10 |
77.4(23/31 – 6/31) |
8 |
 |
(Beat IND) A win is a win. San Diego wasn’t impressed with themselves after the win, but it was a win. A lot of people would want to put an asterisk to this win, but the fact that it’s not immediately beatlooped away does say something. And, New England, San Diego, and Indianapolis being the top three teams in the AFC was not seen as so insane just a few weeks ago. |
11 |
75.0(22/30 – 7/30) |
9 |
 |
(Lost to SD) Indianapolis drops quite a bit, and from what we’re hearing about their injury situation, it may be difficult for them to get back on top. |
3 |
72.4(21/29 – 8/29) |
10 |
 |
(Bye) For a while, Tampa Bay was kept from their beatwin over Tennessee, because Tennessee appeared to be better than Jacksonville, who appeared to be better than Tampa Bay. But Jacksonville defeated Tennessee this week. Tennessee can’t say they’re better than Jacksonville, they split the season series. Therefore, JAC->TB->TEN is reasonable, and Tampa Bay rises five slots on the bye. |
15 |
62.0(15/25 – 9/25) |
11 |
 |
(Beat TEN) Jacksonville is pushed down in the rankings due to still suffering under the beatloss to the weaker-looking Indianapolis team. |
5 |
61.1(15/27 – 9/27) |
12 |
 |
(Beat KC) Denver’s win enables them to shed their beatloss to San Diego, although San Diego found a new beatpath to them through Indianapolis. Denver is closely bunched with a few other teams here, but what pushes them ahead of the others is the quality of their win over Pittsburgh. |
14 |
52.3(11/22 – 10/22) |
13 |
 |
(Lost to JAC) Tennessee falls after their loss to Jacksonville, and their beatwin over Houston looks slightly better than Kansas City and Chicago’s win over Oakland. |
9 |
58.0(14/25 – 10/25) |
14 |
 |
(Beat OAK) Beating Oakland is actually a quality enough win to help Chicago quite a bit. Chicago didn’t have a beatwin last week. |
21 |
57.9(9/19 – 6/19) |
15 |
 |
(Lost to DEN) Kansas City is basically losing a bit of a tiebreaker here in the group of teams that are bunched up in the middle, but fall a couple of slots on the loss to Denver. |
12 |
47.7(9/22 – 10/22) |
16 |
 |
(Beat CLE) Pittsburgh reinforces an existing beatwin by defeating Cleveland. |
16 |
45.2(9/21 – 11/21) |
17 |
 |
(Beat WAS) Philadelphia splits the series with Washington, but it doesn’t make any difference – yet. |
17 |
40.0(2/10 – 4/10) |
18 |
 |
(Bye) Houston is pushed down in the rankings a little more this week, mostly because of TB->TEN re-emerging. |
13 |
44.0(9/25 – 12/25) |
19 |
 |
(Beat MIA) Buffalo defeats Miami as expected. |
19 |
33.3(6/18 – 12/18) |
20 |
 |
(Lost to CHI) Oakland isn’t particularly hurt by losing to Chicago, even though Chicago really helped themselves. |
20 |
33.3(8/24 – 16/24) |
21 |
 |
(Lost to ATL) Carolina loses a beatpath to Atlanta due to the split, and falls three. |
18 |
21.9(2/16 – 11/16) |
22 |
 |
(Lost to PIT) Cleveland loses to Pittsburgh as expected. |
22 |
28.0(7/25 – 18/25) |
23 |
 |
(Lost to STL) St. Louis has lost so much this year that this loss isn’t taken seriously, although it could emerge in the graph later this season if enough other results reinforce it. |
23 |
19.4(0/18 – 11/18) |
24 |
 |
(Beat SF) Seattle defeats San Francisco as expected. |
24 |
25.9(6/27 – 19/27) |
25 |
 |
(Beat BAL) Cincinnati defeats Baltimore as expected. |
25 |
19.2(5/26 – 21/26) |
26 |
 |
(Lost to CIN) Baltimore loses to Cincinnati as expected! |
26 |
16.7(4/27 – 22/27) |
27 |
 |
(Beat CAR) Atlanta isn’t at the bottom of an NFC South beatpath anymore, so they manage to climb a couple of spots. |
29 |
15.6(1/16 – 12/16) |
28 |
 |
(Beat DET) One of these weeks Arizona could start to rise. They’re improving a bit already in terms of background dynamics – their other victories aren’t seen as “flukey” anymore in the beatflukes variation. The Detroit victory, however, is. |
28 |
17.6(1/17 – 12/17) |
29 |
 |
(Bye) Other teams are looking slightly better, so the Jets slip a couple of notches. |
27 |
4.0(1/25 – 24/25) |
30 |
 |
(Lost to SEA) The last three teams are about tied – depending on graph dynamics, they can flip around. |
31 |
5.6(0/27 – 24/27) |
31 |
 |
(Lost to BUF) Since an occasional win would just be beatlooped away in most cases… |
30 |
0.0(0/25 – 25/25) |
32 |
 |
(Beat NO) … the relative rankings here are basically tied, and determined by the rankings from the previous week. But as soon as a team gets a beatwin, it should start affecting things. |
32 |
3.7(0/27 – 25/27) |
Yes – Chicago doesn’t look completely abysmal anymore!
Reposting my links here:
http://www.twomuffin.com/BeatPaths.htm
http://www.twomuffin.com/BeatPathsWeighted.htm
This week I fixed my Path Score system so that it is on a scale from -10 to 10. It isn’t perfect and probably will be tweaked, but I reran each week to see how the numbers look over the course of the season and I was pretty pleased. It also gives you a way to compare the standard rankings against the weighted rankings.
Indianapolis’ comeback allowed them to barely get rid of the loss to San Diego on the weighted chart, keeping them near the top.
I’m a little confused how the Packers passed the Patriots (and yes I’m a Pats fan). Each have no beat losses because of the beat loop the Packers are involved in. The Patriots have a beat path to every team except the Packers. The Packers have a beat path to every team except Dallas and New England. Each has 100.00 beat power. The Patriots formula is 30/30-0/30 and the Pack have 29/29 – 0/0. Do the Packers move up in the formula because of the beat loop and thus they have more relationships? So effectively the Packers move up because they’ve been beaten?
It looks like there is a lot of instability in the graph, especially in that WAS -> DET -> MIN -> SD -> IND chain. There look to be several upcoming games that could result in a beatloop wiping out portions of that path. A game next week that looks like it could do some damage is SD @ JAC. I’m not going to look at all the paths, but from the graph above it appears that a JAC win would really shake things up again.
Or maybe the NFC East and North really are as good as they look on this graph.
Having watched Washington play this year, I’m amazed they’ve managed to hold on to their position in the beatpath rankings as well as they have. They just don’t look that good. But they’ve only lost to GB, NE, NYG, and split the season series with PHI. It’s really that win over DET that’s hold them up, if they loose that through a beatloop without another quality win they’ll tumble. The most obvious place they could loose it is the week 16 game at Minnesota, but who knows what the graph will look like by then.
Jake, it has to do with the teams that the Packers and Patriots have beaten, and where those teams are in the graph. Essentially it’s saying that Green Bay’s competition has been a little bit tougher than New England’s.
Keep in mind that as the teams they beat move up and down, the “strength” of the victory moves up and down as well. Right now the NFC East and North as divisions look quite strong, so they’re pushing GB up. Both NE and GB have direct wins over WAS and SD, so if those teams improve in the rankings they’ll push both teams up (or drop of they fall). But GB is helped by the apparent strength of wins over NYG, PHI, and MIN. NE has wins over DAL and IND, but Indianapolis suddenly doesn’t look as good as they did a few weeks ago.
Here’s a summary of iterative beatwin removal results. Maintained beatins that change the graph are as follows:
PHI->DET (removes need for PHI->NYJ)
HOU->KC (removes need for JAC->KC and HOU->OAK)
DEN->KC (removes need for DEN->OAK)
NO->SF
SF->ARI (removes need for BAL->STL)
So, same total number of arrows in the graph, but the graph becomes two levels taller. The Eagles would be the only big mover in the power rankings as a result of all this.
A few other wins are “maintained” but removed when 4-team loops are considered:
NO->JAC->TB, gets removed when TB->TEN-NO gets considered
ARI->DET->TB, gets removed when TB->CAR->ARI is considered
KC->MIN, gets removed when MIN->SD->HOU->KC and MIN->SD->DEN->KC are considered. MIN->SD->CHI->KC is gone before four team beatloops enter the picture.
If Denver had beaten Green Bay, would the chart look substantially different? Very interesting as it is!
See, now this is fun! Beatpaths thinks the NFC North is a powerhouse! The Bears are the lowest ranked team at 14, the other 3 are at 7 or above!
Of course, total nonsense, as anyone who has watched the division will tell you.
One of the things that strikes me about this system is that it treats every victory as absolutely equal, even though I think most people would argue that’s not the case. (This might have been addressed before, I don’t recall). Consider this: next week, OAK is going to play MIN, who will not have Adrian “Purple Jesus” Peterson. If OAK wins, beatpaths will consider that as impressive as GB’s or PHI’s win over MIN, even though most people would probably argue that they aren’t really beating the same team. Similar issues would probably happen with Indy in the coming weeks, and whoever beat Philly without Westbrook (was that the Giants?).
Basically, the point being that beatpaths views teams as stable units that don’t really change over a season, which isn’t always the case. I don’t know how you would even begin to try to compensate for that, though.
It seems the NFC North has drawn the AFC West as the Interconference opponents this season, and the NFC East as their Intraconference opponents. (See http://en.wikipedia.org/wiki/Regular_Season_%28NFL%29#Formula)
They’re beating up on the AFC West thus far, so far NFC North teams are 8-2 vs AFC West teams (SD beat CHI week 1, KC beat MIN week 3). In contrast, NFC North teams are 4-5 vs NFC East teams. They’re also 1-1 vs the NFC South and 1-0 vs the NFC West in their portion of the “strength of schedule” games played this year.
Since the AFC West is a big part of that clogged up middle of the graph, they’re holding the NFC North above that, and, in some ways, propping the NFC East up as well.
I’m still confused after JT’s response to Jake. The beat power explanation (http://beatpaths.com/?p=66) doesn’t say how teams that are still tied after (wins/total – losses/total) are decided. I’d think that it would go first to total — having the same percentage on more relationships is harder than less relationships. If they are still tied, well, I don’t know what to do…
It just seems wrong to have an undefeated team under a defeated team unless the undefeated team has much fewer beatpaths (that may be the wrong term).
miles, beatpower actually isn’t used in the power rankings. Here’s a quick way to look at it: Both teams have no beatlosses because the CHI->GB loss is looped away. So it’s simply a strength of schedule comparison.
The remainder of Green Bay’s wins (i.e. not counting the SD and MIN wins that they are stripped of) are, in total, more impressive than New England’s wins. New England has beaten the AFC East teams, which have won very few games outside of the division. The (unbeatlooped) teams Green Bay has beaten look more impressive overall.
See this comment from a post earlier in the season: http://beatpaths.com/?p=205#comments (2nd comment, by ThunderThumbs). Basically, the beatpower formula isn’t being used right now, but it’s a holdover from previous seasons and attempts to make sense of these graphs. I don’t remember the specifics of the current tiebreaker, and can’t find it right now, but at it’s heart it takes into account the current position of the opponents faced by the teams, and how they rank in the ratings.
Hypothetically, let us assume that for the rest of the season, every game goes the way the current graph and rankings “predicts” it is going to. That would mean that both NE and GB would win the rest of their games, as neither team plays another team ranked higher than them. In that case, it’s possible that NE could move head of GB in the rankings. Both GB and NE play all 4 NFC East teams this season, so the strength of those 4 hypothetical victories would be the same, and it would come down to the uncommon opponents. Unfortunately for NE, they play 6 games against the rest of the AFC East, which isn’t looking so good this season. GB has 6 games this season vs the NFC North, which (as noted in a previous post) is ranking somewhat high this season.
Jake, since you’re a New England fan, you’ll probably like my rankings better (links in post #2). They’re calculated by summing the total number of paths beneath a team.
In short, if B => A, B gets 1 point and A gets 0. If then C => B, C gets all of B’s points for having defeated them, and in addition gets 2 more for the paths C => B and C => B => A. If then B => D the point totals would be:
A = 0
B = 2 (B => A / B => D)
C = 5 (B’s points / C => B, C => B => A, C => B => D)
D = 0
Loss points are calculated the same way. Take the same scenario.
A = 2 (B’s losses / B => A)
B = 1 (C => B)
C = 0
D = 2 (B’s losses / B => D)
After getting these two numbers, each team’s losing points is subtracted from their winning points. This number determines the order of the teams.
RJ: If you switch it so that Denver beats Green Bay, that win gets beatlooped out by DEN => GB => SD => DEN. Since GB => SD is lost by CHI => GB => SD => CHI anyway, the only effect it appears to have on the graph is Denver shedding its beatloss to San Diego. San Diego would maintain their beatpath to Denver in SD => IND => JAX => DEN, and therefore there would be no visual change in the graph at all. SD would lose some points in my ranking system for it, but not in TT’s.
JT–You’re absolutely right, and I understand the reasons WHY the beatpath system is ranking the NFC North so highly. I’m just saying that I think the truth is much different. Right now, there are a few wins (mostly MIN => SD => IND) that are keeping the division’s opponents ranked highly in the graph, which I don’t think is truly respective of the team’s abilities.
But then, that’s what beatpaths is–an unopinionated statistical analysis. I’m allowed to disagree with it for my own unscientific reasons.
I routinely find beatpaths entertaining for the same reasons. However, I gotta share something – the “picks” this year, which I’m used to being rather inaccurate in past years, are one game out of first place in King Kaufman’s “Panel O’ Experts”… I’m thinking that going back to the vanilla mode, plus finding a less complicated way to handle the first couple weeks of the season, must have helped the overall “picks” performance.
But on the other hand, beatpaths went 7-7 last week, so maybe it’s going downhill again.
“But on the other hand, beatpaths went 7-7 last week, so maybe it’s going downhill again.”
Don’t feel too bad about that. Peter King, one of SI’s experts, only went 6-8. And he’s paid to pick teams. I’d take your objective system over his opinion any day.
It’s so weird to see Chicago ahead of Pittsburgh. Pittsburgh is a top 5 team, and Chicago is a bottom 10 team.
It’s also possible that last week had a fair amount of upsets, and that ANY system which was choosing the winner based on the better team (which is what most predictors do, accounting somewhat for matchups) was going to do badly. That would be my guess.
After all, remember that the better team doesn’t always win…