Sunday, August 4, 2013

2013 NFL Rankings - First Look

As they did last year, Cantor Gaming has released point spreads for every NFL game this season, up to week 16. My rankings are based on what the betting market thinks, as opposed to actual game data, so I can get a jump start the interminable "who's better" debates before we've even seen a complete game of pre-season football.

Opening up the Cantor Gaming Black Box


With 16 weeks of data to work with, it's a fairly straightforward analysis to derive an implied team power ranking from the point spreads. However, I aim to take things one step further and tease out implied team-specific home field advantage from that same data set. To do this, I treat each team as if it were really two teams: the team they are when at home, and the team they are when they're on the road. Each of those teams will have a ranking, as implied by the point spreads, and the gap between the two can be considered a measure of that team's home field advantage.
The table below summarizes the results of my analysis.  An explanation of the columns:

  • 2012 - The team's ranking as of the end of the regular season, according to my betting market rankings
  • GPF - Generic Points Favored.  What you would expect the team to be favored by on a neutral field against a league average opponent
  • aGPF - Away Generic Points Favored.  The implied GPF of the team if you looked only at their road games.
  • hGPF - Home Generic Points Favored.  The implied GPF of the team if you looked only at their home games.
  • HFA - Home Field Advantage.  This is just the difference between aGPF and hGPF, along with the corresponding rank.

2013 NFL Rankings
Rank Team 2012  GPF  aGPF hGPF HFA
1  DEN 3 4.6 3.0 6.2 3.1 1
2  SEA 5 4.5 3.5 5.5 2.0 17
3  SF 4 4.4 3.2 5.6 2.4 6
4  NE 1 4.3 3.5 5.1 1.6 26
5  GB 2 3.1 2.0 4.1 2.1 13
6  ATL 7 2.5 1.2 3.8 2.6 2
7  BAL 10 2.2 1.0 3.4 2.4 4
8  NO 11 2.1 1.1 3.0 1.9 18
9  PIT 12 1.7 0.9 2.5 1.6 27
10  HOU 6 1.5 0.4 2.5 2.0 15
11  CHI 9 1.1 0.0 2.1 2.1 12
12  NYG 8 1.0 -0.2 2.2 2.4 5
13  CIN 14 1.0 0.1 1.9 1.8 22
14  WAS 15 0.9 0.1 1.8 1.7 25
15  MIN 16 0.4 -0.7 1.5 2.2 9
16  DAL 13 0.4 -0.5 1.3 1.9 20
17  DET 18 0.2 -0.8 1.1 1.9 21
18  CAR 17 -0.3 -1.0 0.4 1.3 30
19  MIA 21 -0.5 -1.6 0.6 2.2 10
20  TB 20 -0.9 -1.6 -0.1 1.4 29
21  PHI 27 -1.0 -2.1 0.0 2.1 14
22  SD 19 -1.1 -1.7 -0.5 1.1 31
23  IND 24 -1.3 -2.4 -0.2 2.1 11
24  STL 22 -1.5 -2.7 -0.4 2.3 8
25  NYJ 23 -1.9 -2.9 -1.0 1.9 19
26  KC 32 -2.2 -2.9 -1.5 1.4 28
27  BUF 26 -3.0 -3.9 -2.1 1.7 23
28  CLE 25 -3.1 -3.9 -2.2 1.7 24
29  ARZ 28 -3.6 -4.6 -2.7 2.0 16
30  TEN 29 -4.2 -4.7 -3.7 1.1 32
31  OAK 30 -5.0 -6.3 -3.8 2.5 3
32  JAC 31 -6.1 -7.3 -5.0 2.4 7

Some observations:
  • The top 5 (Broncos, Seahawks, Niners, Patriots, Packers) are unchanged from last year, just somewhat reshuffled.
  • Not a whole lot of movement in general, with the Chiefs and Eagles moving the most, both jumping six spots from their lowly 2012 positions
  • The home field advantage numbers don't appear to track that closely to historical results.  The Broncos are at the top, but their home away splits are below average (since 2006, their average scoring margin is 3.7 points higher at home, which is 19th overall).  In contrast, the Seahawks have had the strongest home/away split, with home scoring margins 11.8 points higher than on the road for the same time period.  But Cantor doesn't appear to put too much credibility in those results. They have the Seahawks right in the middle in terms of implied home field advantage.

9 comments:

  1. The home field advantage on that list is pretty odd with respect to historical data in a bunch of ways.

    Historical mean home field advantage is around 2.7, but Denver is the only team on that list with a HFA that's higher than that. The listed average is around 2, so I'm not sure how profitable 'just take the home team' would be expected to be.

    Listing NYG as a team with high home field advantage (+2.4, 5th highest) seems pretty staggering when they've got the
    smallest home field advantage in the league.

    ReplyDelete
  2. Looking through the list the home field advantage is too low in general. The usual home field advantage suspects seem like a good value - since the HFA is too low, Dallas, San Diego, and and Kansas City are better values that typical.

    San Diego - in particular - has a large difference between the typical home field advantage in the spread, and the Cantor Gaming HFA. That suggests taking bets on them now is relatively likely to set up opportunities to play the middle later.

    ReplyDelete
    Replies
    1. I'm not sure what to make of those home field numbers. As you said, they look small. Is anybody even taking action on the Week 2 through Week 16 games? I was never sure whether Cantor was just publishing these for entertainment purposes.

      Delete
    2. I had considered taking a recreational motorcycle ride to the Venetian in Vegas this weekend to find out, but a phone call is probably a more sensible option.

      Delete
  3. The range of expected margins against average teams also seems oddly narrow. It's possible that Cantor oddsmakers believe the parity will be high this year. I understand that they use software that simulate games to come up with their lines, at least at this stage of the season - maybe it isn't as accurate as they hoped. :)

    ReplyDelete
    Replies
    1. A simulation approach seems most consistent with the data, as the spreads don't seem to follow any kind of simple mathematical formula. Although I would question the value of simulation at this point in the season. It's only as good as your inputs, and there's not much to go on right now.

      Delete
  4. Well, there's always players' past performance. It'd be interesting how they would input "non-skill" players - perhaps 3-year worth of profootballfocus.com stats? :) I'm actually flirting with the idea of using those to make predictions for O-line & Def players' performances.

    ReplyDelete
    Replies
    1. By the way, Cantor supposedly use a mysterious software called "Mirage" to stimulate games' outcomes. According to googled articles, it was created by "hackers". Lol.

      Delete
    2. I was going to make a hot air and fata morgana joke, but I'm too lazy and unsophisticated to work backwards and figure out what they're actually doing.

      Delete