Monday Night: No change to the graph or rankings after New England defeats the Jets. Last week’s power rankings were 10-6 this week.
Sunday Games: Chicago wins, Minnesota loses, Minnesota is three games back of Chicago, and the beatpaths have Minnesota ranked ahead of Chicago.
The reason is because Buffalo’s win caused Chicago to lose their beatpath to Carolina, which was a huge part of their ranking. Meanwhile, Minnesota still has their beatpath to the Giants. I definitely believe that Chicago is a better team than the whole of their season would indicate at this point, and should probably be considered better than Minnesota because of Grossman. However, strictly in terms of the season’s wins and losses, Minnesota has actually played a slightly stronger season when you factor in MIN=>NYG and BUF=>CIN. Chicago could put the doubts to bed by beating Minnesota next week, but the beatpaths will be picking Minnesota to win.
Big movers:
Up: WAS, DAL, SD, ARI, BUF
Down: CIN, CHI, CLE, STL
Read on for Saturday games.
Afternoon Games: Indianapolis’ loss, even though the starters sat, results in Denver taking over the top ranking. Denver was extremely close to Indianapolis even before the game.
Big movers:
Up: WAS, DAL, SD, ARI, SF, BUF
Down: CIN, CHI, CLE, STL
Early Games: Some crazy results in the NFL this week. Buffalo’s upset win affects the standings of some teams you wouldn’t expect, including San Diego. The Giants missed their chance to prove their worth. The standings have them on top in their division, but according to the beatpaths, the Giants are third in the NFC East. Incidentally, that would be true even if Minnesota hadn’t beaten the Giants.
Big movers:
Up: WAS, DAL, SD, ARI, SF, BUF
Down: CIN, CHI, STL
I find it very interesting that week after week, San Diego, despite losing doesn’t seem to get any arrows pointing toward them. I am wondering which wins they played earlier in the year are the games that seem to be protecting them, despite a couple recent losses. You figure sooner or later if they keep playing .500 football, someone is going to have a beatpath toward San Diego. The whole methodology of this is to determine relationships of “who is better than whom”. You wouldn’t think with San Diego’s loss count this late in the season, there hasn’t been at least 1 team you could say is better than SD.
Here are San Diego’s beatloops after Saturday’s late games:
SD split with KC
SD=>WAS=>PHI=>SD
SD=>NYG=>PHI=>SD
SD=>NE=>PIT=>SD
SD=>NE=>MIA=>SD
SD=>OAK=>DAL=>SD
SD=>NYG=>DEN=>SD
SD=>WAS=>DAL=>SD
SD=>IND=>PIT=>SD
So that accounts for all six of their losses: KC, PHI, PIT, MIA, DAL, and DEN.
If Denver beats them next week, Denver will have a beatpath to them.
I should point out that in the beatflukes variant, San Diego does suffer a beatloss to Denver. The reason is because San Diego is only protected from Denver by beating the Giants. The Giants beat Denver, but Denver has an alternative beatpath to the Giants. So the beatflukes variant cancels out the NYG=>DEN portion of the beatloops, which restores Denver’s beatpath to San Diego.
I was pondering your Beatpaths system (which I really like, in case my occasional critiques obscures), and I think I have honed in on what I think is its primary limitation– its nature puts undue significance on in-conference, out-of-division games.
Since a team plays each divisional opponent twice, and every division-mate shares games against the same out-of-conference foes, these games provide a bevy of opportunities for beatloops and the resulting ambiguities.
Conversely, in-conference/out-of-division games are not generally shared by many in-division foes, and as such they more frequently result in beatpaths that have few opportunities to be eradicated.
Looking at the current graphic, and only at the direct arrows rather than those ‘downpath’, here is what I see:
Denver: three outbound arrows; two to in-conference, out-of-division teams (ICOODT).
Washington: one outbound arrow, to an ICOODT.
Indy: two outbound arrows, both to ICOODT.
Seattle: two outbound arrows, both to ICOODT.
Jax: two outbound arrows, both to ICOODT.
NE: two outbound arrows, one to ICOODT.
Dallas: one outbound arrow, to an ICOODT.
Pit: three outbound arrows, one to an ICOODT.
TB: two outbound arrows, one to an ICOODT.
Car: one outbound arrow, to an ICOODT.
Chi: one outbound arrow, not to an ICOODT.
Atl: one outbound arrow, to an ICOODT.
Min: two outbound arrows, one to an ICOODT.
Cin: two outbound arrows, neither to an ICOODT.
Det: two outbound arrows, one to an ICOODT.
NYG: three outbound arrows, one to an ICOODT.
Cle: Two outbound arrows, one to an ICOODT.
Bal: three outbound arrows, two to ICOODT.
KC: Two outbound arrows, both to ICOODT.
SD: Three outbound arrows, two to ICOODT.
Ari: One outbound arrow, not to an ICOODT.
Phi: One outbound arrow, to an ICOODT.
Mia: One outbound arrow, to an ICOODT.
SF: One outbound arrow, not to an ICOODT.
Buf: One outbound arrow, to an ICOODT.
Oak: One outbound arrow, not to an ICOODT.
Stl: Two outbound arrows, one to an ICOODT.
Ten: One outbound arrow, not to an ICOODT.
There are 49 arrows on the most recent chart. Of these, 27 are linking directly in-conference/out-of-division opponents– 55% of the arrows. That seems very, very high to me. Losing an ICOODT game to a team you should not would have a very high cost that you would not likely be able to shake, while winning one would prop you up when you probably shouldn’t.
“The reason is because Buffalo’s win caused Chicago to lose their beatpath to Carolina, which was a huge part of their ranking. Meanwhile, Minnesota still has their beatpath to the Giants.”
A huge part of Chicago’s ranking was due to a win over an ICOODT. They lost it, but the opportunities to lose such a beatpath are fewer than the opportunities to lose an in-division or an out-of-conference beatpath. Meanwhile Minnesota is still enjoying their win over the ICOODT the Giants– and there have not been many ways (although there have been ways) for them to lose that Beatpath over the past few weeks.
Thanks for the comment – you’re right, that is a quirk with the NFL version of the beatpaths.
I’m really not sure whether it’s a bug or a feature, though.
It can be made up for in some ways. For instance, it doesn’t really mean that it is harder to make beatloops, it means that it is harder to make small beatloops (although, the fact that small beatloops get resolved before large beatloops affects this). Also, any beatwin is evidence of the pecking order. So I think the flaw isn’t so much with the evidence that is created, as it is the lack of other opportunity to contradict that evidence (the existing beatwins). What helps this is the playoffs. The NFL doesn’t have an interest in nailing down the rankings of the bottom half of the league, but the playoffs will create the opportunity to create more beatloops and make things shake out a bit more. As for the Giants – even though the Giants aren’t given an opportunity to make up for their beatloss to Minnesota every week, it’s still true that the Giants haven’t managed to develop a beatwin to *anyone* that is considered better than Minnesota – and there are quite a few of them. I’m not sure if I have pointed it out here (I did over on footballoutsiders), but even if I remove the MIN=>NYG win from the schedule, the Giants are still ranked third in the NFC East, and I believe, still below Minnesota in my current tiebreaking algorithm.
I also am not sure that it’s a bigger flaw than the NFL’s current playoff scheme. The playoffs are good for working out who’s better than whom among the teams that make the playoffs, but that doesn’t mean it’s sending the right teams to the playoffs. It’s impossible for the NFL to predict strength of schedule, so they end up with situations where someone piles up a great record from playing easier opponents, while a competitor has a worse record from playing tough opponents. I do believe that Washington has had more “win value” this year than the Giants, for instance.
But, you are absolutely right that some beatwins that dictate a team’s ranking are stickier than others. This is even more true with college football, but in college football people are used to the idea of one loss (to anyone) just completely decimating a team’s future. It’ll be really interesting, however, to see how this works with basketball and baseball as we start focusing more on those leagues.
“For instance, it doesn’t really mean that it is harder to make beatloops, it means that it is harder to make small beatloops”
I think it does, because of what you go on to note:
“(although, the fact that small beatloops get resolved before large beatloops affects this).”
As for if it is a bug or a feature, I like your word: it is a quirk. The passer rating does a good job, but it does a poor job of really nailing down Michael Vick’s value. It is good for a lot, but it has a quirk. This one for beatpaths is a bit more frequently encountered within it than Michael Vicks are encountered within the NFL’s passer rating, though. It is just something to keep in mind when looking at your charts, rankings, and data.
Hmm. Might be interesting to see one thing, though. Or rather, four things. I would be curious to see what it would look like if you made four charts, for the divisions who are playing each other. Have one chart, for example, be of the NFC East/AFC West, and include in it just the in-division and inter-conference games (in other words, just the games between the 8 teams in these two divisions). Do the same for the AFC South/NFC West, the AFC North/NFC North, and the AFC East/NFC South. It would just be interesting to see how the paths shape up just within these verticals, and then compare these stackings to the one that has developed in the full chart.
I think (because I don’t have the software) that if one looked just at the AFC West/NFC East as I suggested above, the beatpath chart looks something like Denver, NYG, and SD with no beatlosses; Denver with direct beatpaths to Dallas, Washington, and Oakland, the Giants with direct beatpaths to Philly and KC, Dallas with one to Philly, and KC with one to Oakland. Philly, Washington, and KC would have no direct beatpaths to anyone. It would give a pecking order along the lines of Denver, Giants, San Diego, Dallas, Kansas City, Washington, Philly, Oakland. Or, within divisions, of NYG->Dal->Was->Phi, and Den->SD->KC->Oak. Pretty close to the NFL standings, which makes sense since the NFL schedule is 62.5% comprised of these games.
Haven’t forgotten about this. At some point I’ll take a closer look at division versus division.