2008 NFL Week 7 Beatpaths Graph

Interesting week – a better pick record at 10-4, and the graph looks a little more orderly to me now, pretty much for the first time all season. I was 1-2 on my personal picks, so I’m back to even for the year.

Games and beatpaths removed from the graph due to beatloops: OAK->NYJ->ARI->BUF->OAK, SD->NE->DEN->SD, STL->DAL->PHI, TB->CAR->NO, CHI->MIN, WAS->CLE.

Other games that used to be in beatloops that have been restored to the graph: CHI->IND->MIN->NO, CAR->ATL->GB, and a few beatwins that are already reinforced by existing beatpaths. So there’s one less team in the beatloop graph this week.

And here are the graphs, first the beatpaths graph, then the beatloop graph.



14 Responses to 2008 NFL Week 7 Beatpaths Graph

  1. doktarr says:

    I agree that we appear to finally be coming into focus. I had a feeling this week would create a bit more vertical graph.

    With the return of CHI->IND->MIN, Indianapolis now has all six of its games represented on the graph. I bet that’s pretty rare at this point in the season.

    I’m curious what the iterative method is going to do with St. Louis, since it will first remove the PHI->STL game, and leave the wins over Washington and Dallas. Washington will be fine thanks to the Arizona win, but I’m not sure how Dallas will get rid of the remaining half-strength STL->DAL path.

    I don’t want to overcomplicate things here, but have you ever thought about drawing a distinction between “path picks” and “ranking picks”? That is, path picks are picks where there is a beatpath one way and not the other, while ranking picks are where there is no beatpath and you are relying on the ranking methods to resolve the ambiguity.

  2. The MOOSE says:

    My graphs are up: http://www.twomuffin.com/BeatPaths.htm

    I hope Shannon Sharpe was watching the game last night since he has the National Guard on speed dial.

    Anyway, looking at the Standard graph I’m surprised to notice that of the seven teams on the middle level, only IND has any BeatLosses.

    The NFC South suddenly has three teams in the top four also. I’m reluctant to believe that ATL, TB, and CAR are the best teams in the league so we’ll see how long they can stay up there.

  3. JT says:

    Just doing a quick scan for well-ordered divisions, I see that the following divisions have arranged themselves into a pecking order:

    AFC North
    AFC South
    NFC West (with a long chain from ARI to SF)

    And I know the placement on the graph of beatloops doesn’t mean much, but that AFC West chain at first glance has the makings of a pecking order, but then it got looped away by two separate 3 team beatloops.

  4. The MOOSE says:

    One thing I find interesting about the graphs right now is that you hear all the pundits on TV talking about how “wide open” the NFL is right now because the dominant teams (NE, DAL, IND) are all struggling. This is shown explicitly in the graphs as you can compare this week to the same week last year.

    After 7 weeks this year the longest BeatPath is 10 teams long. There are 11 teams without a surviving BeatLoss and 6 without a surviving BeatWin.

    After 7 weeks last year the longest BeatPath was 16 teams long. Only 3 teams had no surviving BeatLosses and only 4 didn’t have a surviving BeatWin.

    This shows that there is a lot more ambiguity this year that there was last at this time. As the weeks passed, the longest BeatPath fluctuated, but the teams without BeatWins and BeatLosses stayed small. Until we start linking teams on both ends this year, we won’t really know who is better.

  5. doktarr says:

    I have to say, this year is the most reasonable the weighted (i.e. points-based) beatpath graph has ever looked to me. There’s a handful of teams (Chicago, NYJ, NE, Denver) that look misplaced to me, but overall it seems like a very good reflection of what I’ve seen this year. Moose, have you considered posting an “iterative wieghted” graph?

  6. The MOOSE says:

    No I haven’t. Do you think it would change results in any significant way?

    What I have done though, was run college numbers through the system since I think that BeatPaths would actually be a fantastic alternative to the BCS. Instead of rewarding record (which incents teams to play bad teams), it rewards defeating teams that have defeated other teams. Therefore the #1 team is considered the team that had the best record against the hardest schedule. It also gives no value to blowing the snot out of overmatched teams, and could bring sportsmanship to the college level. This would change the dynamics of college football for the better in my opinion.

    Here’s the top 25 after 8 weeks of college football using my Standard Rating method.

    1] Texas (4.42)
    2] Oklahoma (3.73)
    3] Alabama (3.49)
    4] TCU (3.17)
    5] Georgia (2.86)
    6] Pittsburgh (2.64)
    7] USC (2.61)
    8] Utah (2.55)
    9] South Florida (2.54)
    10] Florida State (2.49)
    11] Oklahoma State (2.19)
    12] Vanderbilt (1.94)
    13] South Carolina (1.88)
    14] Kansas (1.86)
    15] Tulsa (1.86)
    16] BYU (1.83)
    17] Colorado (1.82)
    18] Ball State (1.81)
    19] New Mexico (1.75)
    20] Navy (1.74)
    21] Air Force (1.73)
    22] Ohio State (1.73)
    23] Arizona (1.73)
    24] Wake Forest (1.71)
    25] California (1.71)

    Compared to the AP top 25, many of these are similar which gives some credence to it all. There are also some major differences. The AP has undefeated Penn State at #3, but the Standard method only rates them 36th based on the teams they have defeated. I won’t go into any detail since I’m not really into college football, but I thought this would be an interesting exercise.

  7. The MOOSE says:

    College BeatLoops for those interested:

    Eastern Michigan → Bowling Green → Akron → Eastern Michigan
    Connecticut → Virginia → North Carolina → Connecticut
    Georgia Tech → Boston College → Virginia Tech → Georgia Tech
    Vanderbilt → Auburn → Mississippi State → Vanderbilt
    Stanford → Arizona → UCLA → Stanford
    UAB → Marshall → Memphis → UAB
    South Carolina → Mississippi → Florida → LSU → South Carolina
    Stanford → Oregon State → USC → Arizona State → Stanford
    Wisconsin → Fresno State → Toledo → Michigan → Wisconsin
    Wake Forest → Florida State → Miami (FL) → Duke → Navy → Wake Forest
    Wake Forest → Mississippi → Memphis → Arkansas State → Middle Tennessee State → Maryland → Wake Forest
    UTEP → Southern Miss → Arkansas State → Texas A&M → New Mexico → New Mexico State → UTEP
    Middle Tennessee State → Maryland → California → Michigan State → Iowa → Florida International → Middle Tennessee State
    Northwestern → Duke → Virginia → Maryland → California → Michigan State → Northwestern
    Connecticut → Temple → Miami (OH) → Bowling Green → Pittsburgh → Navy → Rutgers → Connecticut

  8. doktarr says:

    I think that’s a great tool for NCAA football, which has the most disconnected schedule of any havily followed sport in the country. I wouln’t accept those results as the sole basis for anything, but it does give you a chance to look back and say, “wait, we don’t really know much about Penn State”. Like you, I’m no college football expert so I won’t comment further.

    I suspect the iterative beatpath resolution would have a pretty heavy punitive effect on the Jets, who seem like the most overrated team in the weighted method. No other teams jump out at me as likely to move a lot, but I could be wrong.

  9. The MOOSE says:

    Preview of the upcoming games pending TT’s official picks. Comments are based on the Standard rankings. They also don’t take into consideration other outcomes for the same week. Results from many games can cause new circumstances and there are too many possibilities to worry about going through all of them.

    Games of the Week:

    #1 TB @ #7 DAL
    These two teams are playing for a BeatPath to each other. If TB wins, the only visible effect on the graph is a link from TB to DAL, but by gaining DAL’s points TB would strengthen their lead over WAS for #1. Should DAL win, WAS would gain a BeatPath to TB through DAL, solidifying WAS at the top.

    #6 ARI @ #3 CAR
    This is another high ranking fight for a BeatPath. Should CAR win, again there will be little difference in the graph but CAR will go to #1 by gaining all of ARI’s points. If ARI wins, they’ll move up to #2 and also reaffirm WAS at #1.

    Mismatch of the Week:

    #2 WAS @ #32 DET
    Almost as bad as it gets in terms of a mismatch. Could this be a trap game? A WAS win makes no difference since DET has no points to gain. If DET pulls the huge upset, it’ll be devastating for WAS as it’ll wipe away 3 wins and leave WAS completely disconnected from the graph. If WAS wants to take advantage of the other teams capable of giving them the #1 slot, they’ll have to win this game.

    Woofers of the Week:

    #29 SEA @ #26 SF
    This looks to be the first divisional rematch of the year with SF having taken the first round being directly responsible for SEA’s awful ranking. A SF win will simply reinforce the link and likely doom SEA to the bottom for the rest of the year. If SEA wins, they’ll break the link raising both them and STL towards the middle of the pack. SEA may have an awful record, but aside from the loss to SF they’re not against bad teams.

    #31 CIN @ #22 HOU
    While these teams are both near the bottom and not connected to each other, a HOU win is mostly meaningless as CIN has no points for them to gain. If CIN were to manage their first win of the season, the team to suffer would be MIA due to their loss to HOU. HOU would fall to #30 dropping MIA to #31.

  10. The MOOSE says:

    BTW Doktarr.. If DET beats WAS this week, it’ll make for a very interesting Iterative graph due to the 3 loops that it would create.

  11. doktarr says:

    Yeah… I don’t even want to try to speculate on the effects of a WAS loss in iterative, both because it’s way too hard to do in my head, and because I’m a ‘skins fan.

    But I know they won’t end up disconnected from the graph in the iterative version even if they lose. You can only end up disconnected from the graph in the iterative version if you have an even record, and the ‘skins would be 5-3 even if they blow this one.

    So… I guess the beatpaths rankings are STRONGLY advising me to pick Washington in my eliminator league. By the other measures I look at, it’s basically a tossup between that game and the Jets/Chiefs game. It’s down to two people and there’s a lot of money on the line…

  12. SB says:

    Moose, thank you for running the college data – I had been certain that my beloved Terps had found their way into some pretty absurd beatloops, and it appears I am correct. Three seven team beatloops through seven games. So inconsistent.

  13. the silent speaker says:

    Is there any reason the beatloop graph has to be so vertical? I think if you made it squarer it would be easier to read. There seem to be three clusters of loops: one with only JAC–>DEN–>TB in and out, one with only DEN–>TB in and CHI–>PHI out, and one with only CHI–>PHI in and JAC–>DEN out.

    Also, and admittedly this may be easier to visualize on paper than to draw in a computer program, but if you moved PIT to CHI’s level directly over PHI and JAC just above them, the PHI–>>DEN chain would take up a lot less space and be much shorter. Likewise with NYJ, MIA and ARI on the same level or so, and with SD–>NE crossing NYJ–>MIA instead of circling around, SD can be pulled up to about level with OAK and about four arrows would be a lot shorter.

  14. JT says:

    I don’t think there is any particular reason why the beatloop graph is so vertical, since there really isn’t any meaning to the vertical placement of teams on the graph. I suspect the graphing software is set up to draw the arrows downward as much as possible, with a minimal set of arrows going up. I happen to like that, as that makes it easier to spot the loops.

    But I mostly suspect that as the beatloop graphs are a pretty recent addition to the site, they just haven’t been modified much.

Leave a Reply

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