Beatpaths

14 Responses to “2009 NFL Week 11 Beatpath Graph”

1. doktarr Says:
   November 24th, 2009 at 12:02 pm

   Yeah, it’s funny how, despite the addition of many new loops, the overall graph is much more vertical. The addition of a couple key paths (notably NYG->ATL and DET->CLE) makes all the difference.

   I know the graph is automatically generated, but personally, aesthetically, I think it would be cool if the longest path always ran straight down the middle of the graph – a sort of a “spine” for the graph. In this case that would mean IND=>ARI=>NYG=>ATL=>SF=>SEA=>DET=>CLE=>BUF=>TB would be the “spine”.

2. The MOOSE Says:
   November 24th, 2009 at 1:33 pm

   What do you do when there are multiple longest paths? Last week there were 10.

3. doktarr Says:
   November 24th, 2009 at 2:12 pm

   Yeah, good point. It breaks down in that case.

   Hey, no commenting here until you get the iterative graph up!

4. Tom Says:
   November 24th, 2009 at 4:12 pm

   Did you do something different with the graph this week? The font looks crisper than in previous weeks.

   Atlanta seems to be propping up a lot of the top teams. It will be interesting to see how that position holds up through the wildcard race.

   Also, looks like only the NFC West has a clear pecking order at this point in the season.

5. ThunderThumbs Says:
   November 24th, 2009 at 4:14 pm

   There’s all sorts of tiny things you can do with graphviz and I keep experimenting. I think what I did this time is rather than having graphviz create a medium sized graph, I made it create a really huge one, and then I’m just letting other processes resize it down later. Seems to work a lot better.

6. Jason R Says:
   November 24th, 2009 at 6:51 pm

   How automatic is the graph? Look at New England, Tennessee, and new England’s arrow to Atlanta. Wouldn’t it be more efficient (is that the word?) to have Tennessee a little more to the left, and have New England’s arrow to Atlanta pass on the right of Tennessee?

7. The MOOSE Says:
   November 24th, 2009 at 7:11 pm

   Fair enough doktarr, graphs are up. No commentary again this week. I’ll be moved this weekend, but having internet won’t be guaranteed so there’s a chance next week will be late too.

8. The MOOSE Says:
   November 24th, 2009 at 7:26 pm

   Jason: The graph is very automatic. With graphvis, you can tell it relationships…

   ARI->SEA [color=blue];

   Define the nodes…

   MIN [fillcolor="#b3e2cd"][color="blue"];

   Even force certain teams to be on the same level…

   {rank=same; DAL PIT NE NO}

   There are many options to graphvis, which you can find in their documentation. But the purpose of the application is to draw graphs without having to figure it out yourself. There may or may not be controls to fine tune a graph, but doing so would reduce the automation in our programs and require weekly tweaking that can be a pain.

9. ThunderThumbs Says:
   November 24th, 2009 at 9:34 pm

   Yep, thanks Moose. Do you use graphviz for your graphs, too? The team image icons are really nice.

10. Tom Says:
    November 24th, 2009 at 10:52 pm

    Pretty interesting that Standard, Iterative, and Weighted all agree on the top two teams: Indy and NO.

    Strange that iterative has preserved Oakland’s win over Philly for four weeks now. Can the Eagles credibly be ranked as the 22nd team in the NFL?

11. The MOOSE Says:
    November 25th, 2009 at 11:39 am

    TT: Yes I use graphviz. The team logos I made myself in Photoshop with a trial license which has expired. Otherwise I would have updated the Lions’ new logo.

    Tom: Yeah, there are always going to be teams that are out of place. Especially in the NFL with so few games, the schedule doesn’t give us a whole lot to work with. One or two games can have a large influence. PHI can shed the BeatLoss to OAK though if they beat DAL or NYG in the rematch games.

12. Tom Says:
    November 25th, 2009 at 5:26 pm

    Moose,

    I thought this was the exact type of eventuality that iterative was meant to remedy—i.e. making sure that one ‘flukey’ game doesn’t persist for weeks on end. Standard got rid of it pretty quickly in comparison. Why does it show up in iterative and not in standard?

13. The MOOSE Says:
    November 25th, 2009 at 6:15 pm

    Because Iterative sees these two loops

    PHI → NYG → DAL → PHI
    PHI → NYG → OAK → PHI

    and considers the PHI → NYG game to be the fluke

14. doktarr Says:
    November 29th, 2009 at 7:23 pm

    Right, the games we consider flukes aren’t necessarily obvious flukes from the algorithm’s perspective. According to the graph, there just isn’t a lot of data conflicting with that Oakland loss. I mean really, look at the Eagles’ schedule. They’ve beaten the Bucs, Chiefs, ’skins (twice), Panthers, and Bears. Those are all bad teams. I happen to think the Eagles are pretty good, but from a beatpaths perspective, they’re a team that’s beaten a bunch of bad teams, and the Giants.

    The easiest chances to shed that Oakland loss for the Eagles come on week 14. If they beat the Giants again, then they can eliminate the Oakland loss. Also, the Redskins play Oakland that week, and if the ’skins win, that generates another loop involving Oakland that get the Eagles off the hook.

    If they can’t pull it off then, then they can shed the loss by beating either the Broncos, or by splitting the season series with the Cowboys. And if they fail at all of those… why shouldn’t they be ranked below the Raiders?

Leave a Reply

Name (required)

Mail (will not be published) (required)

Website