Pick “Confidence” for Week 13

Tom here. A little late in the day, but I thought I’d continue to look at BeatPower and the confidence of the model in making each week’s picks.

I take the standard picks for Week 13, and then compare the BeatPower numbers for each team from the most recent rankings. The greater the difference between the two teams’ Beatpower, the great “confidence” the model has in making the pick.

I’ve ranked the teams for Week 13 below, and I’m filling in the results as they come in. If the model is generally correct, the incorrect picks should be clustered toward the bottom of the chart.

Matchup
(Winner-Loser)
“Confidence”
(out of 100)
BeatPower comparison
(predicted winner – predicted loser)
Result
Tennessee-Detroit 95.2 95.2-0.0 CORRECT
NY Jets-Denver 87.5 100.0-12.5 WRONG
Dallas-Seattle 76.5 80.8-4.3 CORRECT
Atlanta-San Diego 71.3 91.3-20.0 CORRECT
Pittsburgh-New England 54.7 83.3-28.6 CORRECT
Baltimore-Cincinnati 42.5 48.1-5.6 CORRECT
Carolina-Green Bay 30.2 94.0-63.8 CORRECT
Arizona-Philadelphia 26.9 86.0-59.1 WRONG
Houston-Jacksonville 25.9 34.6-8.7 CORRECT
Miami-St. Louis 25.3 31.5-6.2 CORRECT
Indianapolis-Cleveland 21.3 56.7-35.4 CORRECT
Buffalo-San Francisco 13 24.1-11.1 WRONG
Oakland-Kansas City 10.4 12.5-2.1 WRONG
NY Giants-Washington 6 96.0-90.0 CORRECT
Chicago-Minnesota -10.6 60.0-70.6 CORRECT
Tampa Bay-New Orleans -11.6 75.9-87.5 WRONG

Again, note the two negative “confidence” picks at the bottom. Here the BeatPower method disagrees with the standard Beatpaths method. As was the case last week, the standard Beatpath method seems to do better when the two methods disagree.

7 Responses to Pick “Confidence” for Week 13

  1. boga says:

    I would like to see how this stacks up to TMQ’s method of picking wins (Pick the team with the better record, if same record, pick the home team). It would be interesting to see if trying to come up with a prediction based on the beatpaths matches well with the vanilla picking.

    Boga

  2. Tom says:

    Well, hopefully it would do better, as the Beatpaths method is supposed to get beyond the raw record, to look at the strength of the teams that constitute the record.

    TT used to highlight the Beatpaths pick record compared to various other experts and pick systems, but hasn’t done that this year.

  3. boga says:

    I’ll look forward to the comparison at the end of the regular season.

  4. Kenneth says:

    Beatpaths is a part of the Panel O’ Experts compiled by King Kaufman of salon.com this year. I don’t recall where it is ranked among that group, and I don’t know if TMQ was a part of it, too, but at the end of the year we should have a nifty side-by-side comparison.

    And, maybe dinner at King’s house, too! 🙂

  5. Tom says:

    @Kenneth: Ah yes, here we are on the last Panel O’ Experts. Not doing as well as the others, unfortunately, but we don’t have banks of computers running evolution simulators on every single play of future matchups, like “Accuscore” does. At least we’re beating EroticModelPicks…

    @boga: I’m also interested at the end-of-year comparisons. We should get better as the year progresses and more data is fed in. I’ll be interested to see if end-of-year Beatpower graphs do a better job at ‘predicting’ the winners after the fact than the week-on-week picks do.

  6. doktarr says:

    As I said last week – there is no correct ranking position for the Denver Broncos.

  7. Tom says:

    @doktarr: I guess the best metric we can hope for is seeing where the Broncos end up, and whether their final rank, when viewed retroactively, ‘predicts’ their final record as best as possible.

    But you’re overall right, there’s just no good way to force inconsistent teams into a rationalized ranking.

Leave a Reply

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