2008 NFL Week 14 – Divisions

Playing around with divisions this week.

Since hidden lines are more important, I start with a full graph of the week:


And then I start with the divisions, starting with the NFC South like last time. The first rev has some extraneous stuff in it:


But I’ll clean that up after doing the rest of the divisions. I finish the NFC and clean up the most obvious arrows (the looping ones where a division beats itself) and I get this:


I do all divisions with some cleanup (removing the looping arrows where a division beats itself) and I get this:


Then I start simplifying. For instance, I see that the AFC West has three arrows to the NFC South, but that the NFC South has plenty more to the AFC West. So I remove three from each:


I do the rest – (AFC South and AFC East are tied at 2-2, so they cancel out).


I finally remove the redundant arrows and get:


What are your conclusions? To me it looks like the NFC South is good, but not great. Even the AFC North is arguably better than them.

11 Responses to 2008 NFL Week 14 – Divisions

  1. doktarr says:

    Interesting analysis. I like the all-paths graph at the top. Funny how 8-5 Denver has 10 games beatlooped away, leaving only its NFC South wins. But the inconsistency title is actually shared by Washington and Philadelphia, both of whom only have two games not involved in a loop.

  2. Kenneth says:

    Interesting little parallel structure there–the two conferences are aligned neatly in order, right next to each other. And the intraconference rankings look pretty close to conventional wisdom to me; the only possible arguments seem between AFC North and AFC East, which some might switch, and NFC South and NFC East, which again some might switch.

    As for the rankings…it’s possible that the NFC South is 5th, but it’s also possible that it’s 2nd. I wonder if removing the redundant arrows makes them look weaker than they are; the NFC South only has one incoming arrow, while the AFC North has 5.

  3. mm says:

    it makes me think that in the 3 NFCsouth-NFCeast games so far, the NFCeast has had homefield advantage and 2 of them have been quite close anyway.

    In the NFC, the connections between the South and East as well as that between the West and North are pretty week because we only have 4 pairings scheduled (and not all of those have been played yet!). Home field advantage only goes one way in these matchups as well. (for the same reason, the connections between the North and East as well as between the South and West are weak, but I have more confidence in what the graphs says there).

    Some years you could get a more clear graph going through links to the other conference, but it doesn’t seem to be happening this year.

    In the AFC, I have more confidence in the South’s place at the top because it was paired with the North (who comes out behind it here), and both divisions have lines to the East and West.

  4. The MOOSE says:

    As a fan of an AFC East team, I’d have to say that right now the AFC North is likely the better division. I realize that our division may not have a lame team in it, but most of the reason is because our intra-conference matchup this year was against the AFC West and our inter-conference matchup was against the NFC West, both at the bottom of the graph. And we all know that these graphs are about who you beat. 😀

  5. Eddo says:

    It seems like you did this in a really complicated way. Wouldn’t it have been easier to (1) remove all intradivision games, (2) change each team’s id to represent their division (i.e. ATL->NFCS, TAM->NFCS, DEN->AFCW, etc), (3) re-run the standard beatpath method using the division as teams?

    Maybe it gets the same result, I don’t know, but if you’re lumping all the teams together into divisions, shouldn’t that be done before removing any beatloops?

    Or am I not understanding your method?

  6. Boga says:

    I am a fan of that graph. Seeing how that changes every couple of weeks would be sweet.


  7. ThunderThumbs says:

    Eddo, well, I deliberately did it that way to give more data people to discuss on. No one’s seen the “redundant line” beatpath graph for a long time. But you’re right, I could have created the beatpath graph after doing the substitutions, rather than before. I’m actually not sure, it might lead to a slightly different graph – perhaps more beatloops.

  8. Eddo says:

    Thanks, TT. I hope I didn’t come across as combative; I love this site and eagerly await every post, but I was confused by your method and trying to wrap my head around it.

  9. ThunderThumbs says:

    Nope, you’re fine! 🙂 Comment away, I love the ideas everyone has.

  10. Tom says:


    I have the same difficulty: the data linking some divisions is weaker than the data linking other divisions. The NFC East has played a lot of games against the NFC West, but far fewer games against the AFC North, NFC North, and NFC South. In addition, the data relating divisions to one another isn’t binary (win-loss) but subject to slightly finer gradations (e.g. 3 wins – 5 losses).

    I like TT’s initial cut at this problem—it’s simple, and the results correspond pretty well to our intuitions about which divisions are more difficult. For instance, it isn’t difficult for the Cardinals to rack up a good record when their division rivals are some of the weakest in the entire league. As such, it makes sense that the NFC West is at the bottom. The NFC East seems to do well against every division it plays against. Its wins against the NFC West oughtn’t count for much, since most divisions should (and do) do well against the NFC West, but winning against the NFC North, AFC North, and NFC South as well makes the NFC East look pretty good.

    So where do we go from this initial picture?

    I wish I had an easy answer to write in here, but I’m going to have to think about it for a little while more. I just picked up 40 ten page papers that need to be graded this afternoon–a task that will likely take me the next 2.5 days to complete. Absolutely mind-numbing.

    In any case, I’m excited about this new type of graph, and think there’s a lot of potential here.

  11. Tom says:

    I tried an experiment (taking a break from grading papers), and simply summed the BeatPower scores of each division, arriving at the following ranking:

    NFC East – 299.7
    NFC South – 275.9
    AFC South – 228.3
    AFC East – 215.8
    AFC North – 202.3
    NFC North – 146.5
    AFC West – 129.2
    NFC West – 88.5

    This is interesting for two reasons: first, it confirms the view that the NFC East is the toughest division, and that the AFC West and NFC West are the weakest divisions. Second, it ranks the NFC South above the AFC South, despite the graph (visually) placing the AFC South above the NFC South (albeit without a beatpath between the two).

    The problem with my quick & dirty methodology is that the BeatPower score includes all of the team-vs-team beatpath strength (including division rivals), rather than just division-vs-division beatpath strength.

    TT: is there a way of applying the ranking system you use for individual teams to the divisions? We know that the Beatpath graph often places teams visually higher than their actual place in the Beatpath rankings — perhaps the same thing is happening here, causing the seeming AFC South-NFC South discrepancy.

Leave a Reply

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