I realize this kind of goes against the entire point of this site – focusing on wins and losses and no other stats whatsoever – but I also like experimenting with the system, and the latest experiment is to figure in point differentials. Partly to relieve the angst of all the Pat-heads here that talk about how the Patriots are blowing everyone out 
but also just to see how things shake out.
What I did is treat each point as a win. So if A beats B 17-13, the system treats it as thirteen ties, and then A beating B four times. So every point is a win – beatloops are resolved the same way, etc.
What it basically means is that while there the exact same beatloops, the beatloops eventually get broken at one link, creating a chain, so it’s like the tiebreaker is point differential. On top of that, it’s also used as a tiebreaker for choosing between multiple teams with no beatpaths to them.
There are a lot of differences in the rankings. It compares the rankings to how I otherwise have them ranked this week.
This also would flip the picks for Atlanta over New Orleans and Detroit over Tampa Bay.
| Rank | Team | Notes | Vanilla | BeatPower |
1 |
 |
(Beat WAS) … |
2 |
100.0(23/23 – 0/23) |
2 |
 |
(Bye) … |
1 |
100.0(21/21 – 0/21) |
3 |
 |
(Beat DAL) The Patriots are still ranked #3, even when taking point differential into account. |
3 |
100.0(24/24 – 0/24) |
4 |
 |
(Bye) The Steelers have had some dominating victories, so they climb quite a bit. |
10 |
100.0(22/22 – 0/22) |
5 |
 |
(Lost to NE) … |
7 |
95.8(23/24 – 1/24) |
6 |
 |
(Beat ATL) … |
8 |
88.0(22/25 – 3/25) |
7 |
 |
(Lost to GB) … |
9 |
84.0(21/25 – 4/25) |
8 |
 |
(Beat NYJ) … |
11 |
80.0(20/25 – 5/25) |
9 |
 |
(Bye) Detroit’s a top-ten team. |
16 |
76.0(19/25 – 6/25) |
10 |
 |
(Beat CHI) Minnesota’s a top-ten team. What? |
21 |
72.0(18/25 – 7/25) |
11 |
 |
(Lost to NO) … |
15 |
95.5(21/22 – 1/22) |
12 |
 |
(Beat TEN) … |
4 |
87.0(20/23 – 3/23) |
13 |
 |
(Lost to TB) … |
5 |
82.6(19/23 – 4/23) |
14 |
 |
(Beat HOU) … |
6 |
78.3(18/23 – 5/23) |
15 |
 |
(Lost to NYG) … |
25 |
54.8(17/31 – 14/31) |
16 |
 |
(Lost to JAC) …. |
13 |
51.6(16/31 – 15/31) |
17 |
 |
(Beat CIN) … |
18 |
42.9(12/28 – 16/28) |
18 |
 |
(Beat OAK) … |
12 |
39.3(11/28 – 17/28) |
19 |
 |
(Bye) … |
17 |
33.3(9/27 – 18/27) |
20 |
 |
(Lost to SD) … |
20 |
26.9(7/26 – 19/26) |
21 |
 |
(Beat ARI) … |
14 |
15.8(3/19 – 16/19) |
22 |
 |
(Beat MIA) … |
24 |
24.1(6/27 – 20/27) |
23 |
 |
(Lost to KC) … |
26 |
19.2(5/26 – 21/26) |
24 |
 |
(Lost to CAR) … |
30 |
11.9(1/21 – 17/21) |
25 |
 |
(Beat SEA) … |
19 |
0.0(0/17 – 17/17) |
26 |
 |
(Bye) … |
22 |
9.5(2/21 – 19/21) |
27 |
 |
(Lost to MIN) … |
23 |
0.0(0/18 – 18/18) |
28 |
 |
(Beat STL) … |
27 |
16.7(4/27 – 22/27) |
29 |
 |
(Lost to PHI) … |
28 |
4.0(1/25 – 24/25) |
30 |
 |
(Bye) … |
29 |
6.0(1/25 – 23/25) |
31 |
 |
(Lost to CLE) … |
31 |
0.0(0/25 – 25/25) |
32 |
 |
(Lost to BAL) … |
32 |
0.0(0/26 – 26/26) |
Why is it surprising that this doesn’t change New England’s rating? If I understand correctly, the only difference with this methodology is that it eliminates beatloops via a point difference tie breaker. But NE hasn’t lost and so are not in any beatloops. Whether it won all its games by 1 point or 100 would not change its beatpaths, right? The same is true of the Colts. The Packers benefit from this, jumping above NE and Ind because it apparently effectively eliminates their loss.
I am not sure this does anything to address the issue of your system failing to capture the information contained in margin of victory.
It’s also a tiebreaker for teams that don’t have beatlosses. And the system hardly “fails” to capture that information, since it deliberately doesn’t.
Woohoo! The Bears beat the best team in the league!
*goes back to crying in his cornflakes*
No matter how you slice it, the Rams and Dolphins are at the bottom of the list…seems right to me!
Uck! This does go against the point of the site.
I love Beatpaths – not because looking at only wins and losses is the best way to judge teams, but because it is interesting to see just how much you can do with so little data. Once you add in point differentials, you are just another lame football ranking site.
If you want to experiment with taking margin of victory into account, try this: treat any close game as a tie. You could call the graph “BeatDownPaths”.
Defining “close” is up to you. I would call anything within 8 points “close” – in that the losing team could have tied it with one score. You could use 6 points instead, since the losing team could have won it with one score (assuming no 2 point tries.) 2, 3, and 7 points could also work, for similar reasons.
Ha! I agree, Nat – don’t worry, it’s just a what if. In the past I’ve experimented with figuring in home field advantage and other things like that, but the official version that I’ll be continuing with is still just going to focus on only wins and losses.
I’m confused. From the graph, it looks like the Pack and Pats have the same beatpaths, except the Pats also have the Cowboys in theirs. Why do the Pack come out on top? They couldn’t have a higher point differential, right?
The tiebreaker is actually still “strength of beatwin”, kinda sorta as measured by the average beatpath length of the teams they’ve directly beaten. And Green Bay has still beaten more highly ranked teams than New England. If there’s a huge point differential, it means that they’ve beaten them “many times”, which ups their average for a highly ranked team – but in New England’s case, it’s not enough to surpass Green Bay’s tougher schedule.
Okay, I’ll relax and enjoy. I have nothing against “what ifs” – I enjoyed the “what if Buffalo beats Dallas” chart, and was rooting for the Bills to pull off the upset. Even this experiment wasn’t as bad as I made it out to be. Keep up the good work.