NFL Power Rankings (experimental)

[Dark Horse]

Based on just two games, so still very fluid.
Formula: ranking offense in pts per drive (1x); plus ranking defense in pts per drive (x1.5); plus ranking in net pts per red zone drive (x2). Divided by 10:

1) BAL 1.35
2) SD 1.5
3) JAX 3.3
4) ATL 3.5
5) CHI 3.8
6) PHI 3.95
6) NE 3.95
8) SEA 4.6
9) DEN 4.9
10) CIN 5.5
11) NO 6.05
12) ARI 6.3
13) MIN 6.6
14) NYG 6.7
15) STL 7.6
16) BUF 8.05
17) MIA 8.1
18) DAL 8.2
19) PIT 8.25
19) CAR 8.25
21) CLE 8.45
22) NYJ 8.75
22) GBP 8.75
24) TEN 9.55
25) KC 9.6
26) IND 10.4 (terrible defense)
27) HOU 10.8
28) SF 10.95
29) WAS 11.1
30) DET 11.7
31) TB 12.1
32) OAK 13.25

Home team with biggest edge: SEA 4.6 against NYG 6.7 (2.1 pt)
Road teams with biggest edge: JAX, BAL, PHI (all 7 pts or more)

[bigboydan]

This should be interesting in the comming weeks. I'm curious to see how this will change in just 2-3 more weeks.

[Hulu]

Looks very interesting Dark Horse. I tried to come up with a power ranking formula years ago while in college but after the initial year at about 58% ATS it failed miserably the second year and returned back to a tw0-year average of about 50.2% or something like that. I gave it up after that but have always had it in the back of my mind that if you could find the right mix of statistical categories, it could be a valuable tool in handicapping.

[McBa1n]

There's so many factors to consider, ranking formulas are rough.
Turnovers and penalties DEFINITLY have to be considered, IMO. I would say rushing o/d, also, but with Vick running around, it makes it moot for one team on offense (those rushing numbers are scewed).

The thing about rankings is there has to be a 'tilt' for things like injuries (key or not), home advantage/disadvantage - and how much do those things weight numberically? Not to mention play during 'garbage time' - as in the 4th quarter when nothing really matters.

I wish you well on experimenting with a system, it sure isn't easy.

[Dark Horse]

Just an experiment. If it does stick, I would add it to the toolbox and only use it along with other tools. If not, I'll probably learn something in the process. Reasoning behind this is that defense wins championships, so that has to count somewhat more than offense. And red zone efficiency decides games, so that has to weigh in heavier.

Probably need about four weeks in the books to stabilize. At this point the weakness of the Indy defense stands out.

[Justin7]

Any chance you'll explain the general theory of how you make your power rankings?

[Dark Horse]

I get the starting numbers from 2minutewarning.com. lol. Then do the little adjustment.

They're actually more like yardage rankings, translated to a dynamic game reality. I had been looking for something like that.

[Jay Edgar]

Looks promising, DH. If you can get this data for any prior years, it'd be revealing to work backwards -- rank the teams like this for last year, and compare those rankings with the straight up 2005 W/L standings and 2005 ATS raknings. My hunch: that exercise would confirm that you are onto something, or at least you are in the right neighborhood.

In the past, one way I have tried to get at roughly the same thing is by yards-per-point ratios. The best teams ATS seem to have the lowest YPP ratios on offense (they are efficient with the yards they do gain) and the highest YPP ratios on defense (they bend but don't break). (It's also nice because the teams with huge raw offensive yardage stats are the public teams, and this analysis points you elsewhere). Your formula would probably put the teams in roughly the same places as a YPP ratio analysis. (And probably ditto for a ranking by turnover differential, which is the simplest of all to do.)

[Philippe27]

I made my own rankings as well and posted my picks for the week, I guess we can compare the two after a couple of weeks and see how things go.

[My3z]

Looks promising, DH. If you can get this data for any prior years, it'd be revealing to work backwards -- rank the teams like this for last year, and compare those rankings with the straight up 2005 W/L standings and 2005 ATS raknings. My hunch: that exercise would confirm that you are onto something, or at least you are in the right neighborhood.

This experiment of yours does look promising Dark Horse
I agree with Jay Edgar as well. Let me offer you this incase you don't have and would like to give JE thought a whirl:

Baltimore Ravens
2005 Stat Rankings

Points per game OFF=16.6 Rank 25 DFF=18.7 Rank 10
RY per game OFF=100.7 Rank 20 DFF=99.6 Rank 9
PY per game OFF=192.6 Rank 22 DFF=185.1 Rank 8
Y per pass attempt OFF=5.48 Rank 27 DFF=5.63 Rank 2
TOTAL Y per game OFF=293.3 Rank 24 DFF=284.7 Rank 5
YARDAGE DIFFERENTIAL OFF=8.6 Rank 17 DFF= - -
TOTAL TURNOVERS OFF=36 Rank28 DFF=30 Rank 11
TURNOVER DIFFERENTIAL OFF= -6 Rank 24 DFF= - -

Stat Logs 2005 (sm. print away)

OPP: PA SU LINE ATS O/U RY PY RYA PYA

IND 7-24 L +3 L 44.5U 21-77 54-324 26-86 36-254
ten 10-25 L -3.5 L 36.5U 13-14 41-168 29-97 37-193
NYJ 13-3 W -7 W 31.5U 45-115 21-144 15-28 28-124
det 17-35 L +1 L 32.5O 30-159 37-228 37-169 23-97
CLE 16-3 W -6 W 34U 33-150 31-201 18-70 30-116
chi 6-10 L +2 L 30U 22-66 32-133 29-143 29-133
pit 19-20 L +12.5 W 35O 27-72 44-246 28-101 31-160
CIN 9-21 L +3 L 36.5U 24-124 31-116 34-98 26-231
jac 3-30 L +6 L 33.5U 17-53 33-110 35-133 30-205
PIT 16-13 W +3.5 W 33.5U 38-104 36-137 25-70 38-212
cin 29-42 L +9 L 37.5O 33-133 32-189 32-135 30-302
Hou 16-15 W -7.5 L 39U 23-73 33-165 34-165 37-133
den 10-12 L +15 W 40U 23-72 40-251 32-96 33-222
GRB 48-3 W -3.5 W 34O 38-182 27-253 19-107 44-181
MIN 30-23 W -3.5 W 36O 32-88 34-279 14-42 36-220
cle 16-20 L -3 L 36.5U 32-129 36-138 23-54 38-178

Hope this is usefull Dark Horse if you haven't this already

[Dark Horse]

Thanks for the input. The other 'yardstick' I use is the X-factor, which compares yardage ranking to points scored/allowed ranking, for both offense and defense. Idea is to play ON teams of 5 or higher against teams of -5 or lower. Again, still very early in season. For teams in action this week:

1) BUF 23
2) BAL 19
3) PIT 15
3) CAR 15
5) CLE 13
6) CIN 9
6) HOU 9
8) CHI 7
9) ARI 6
10) STL 5
11) IND 3
12) NYJ 2
12) SEA 2
14) DEN 0
15) TB -1
15) NE -1
17) NYG -2
18) DET -3
18) TEN -3
18) WAS -3
18) ATL -3
18) NO -3
23) PHI -4
24) MIN -4
25) JAX -5
26) SF -10
26) GB -10
28) MIA -20

Doesn't hurt to have two independent approaches towards the same subject (yardage). BUF and MIA are on opposite ends of the scale here, but almost the same in the other ranking. For now, I'm looking first at the games with teams for which both rankings comply. For instance CHI has an edge in both rankings over MIN.

[onlooker]

Thanks for sharing this Dark Horse. It will be interesting to see how this shapes up down the road.