Monday, October 08, 2007

Saurian Sagacity Power Poll

I’ve decided a Power Poll is in order. This is not my BlogPoll voting (which is resume based), but an unbiased Power Poll based on NCAA statistics.

Why do this? Well it all comes from a comment left in my “BlogPoll – My Response” piece.

The comment was –

KevinH said...

Here's a question about resume ranking. Say Florida beats LSU. Is it better to have a quality loss (LSU) or better to have a quality win (FL)? This is similar to the Canes over FSU in 2000 (in which they chose quality loss).
Sadly, I didn’t get to address the question directly because the Gators lost in the last minute. But KevinH got me thinking – At what point is a quality loss better than a win?

To determine this, I first had to devise a system where, to the extent possible, I could look at the teams from some sort of Power perspective. In that way, I could decide just which teams are “quality” for the purposes of “quality wins or losses”.

So what is quality or power? I wanted something with totally objective input, so I went to the NCAA stats. I chose –

1. Net Average Yardage – Simply the team’s average yards per game on offense minus the average yards surrendered on defense. I did not choose anything based on points because points are not necessarily the whole story. For instance USC out gained Stanford significantly. I have little doubt USC would beat Stanford 9 out of 10 times. Yet Stanford scored more points due to circumstance, including USC turnovers. I wanted a stat that reflected power absent the sometimes arbitrary results of scoring. Plus, scoring does show up in my next input, which is;

2. Win-Loss Record – It is what it is. Also;

3. Strength of Schedule – We use the NCAA stats for this. Notably, we are using the Past Opposition category.

As far as individual teams in each category

1. Net Average Yardage – The top teams here might surprise you –
1. Kansas
2. Texas Tech
3. West Virgina
4. LSU
5. Hawaii
6. Ohio State
All have over 200 yards net average per game.

2. Win-Loss Record – Anyone paying attention knows.

3. Strength of Schedule – Year to date the ten hardest past schedule belong to –
Stanford
Illinois
South Carolina
Oregon
North Carolina
Kansas St.
Auburn
Oregon St.
Florida
South Fla.
After 6 games in the season, I decided this could work although some teams (U Conn for instance) have still not played a competitive schedule, and as a result may have over inflated statistics. I do feel this Power Poll will get better as the season wears on.

Lastly, any bias in this poll would reflect the weightings given the stats. I’m going to spare everyone the formula used here in the interest of brevity, but know that the stats are essentially “equal weighted” as ratios then added together.

Introducing Saurian Sagacity’s Week 6 Power Poll
1 LSU
2 Ohio St.
3 Kansas
4 West Virginia
5 Texas Tech
6 Oklahoma
7 Missouri
8 Arizona St.
9 Boston College
10 Oregon
11 Florida
12 Illinois
13 Rutgers
14 South Carolina
15 Kentucky
16 Cincinnati
17 South Fla.
18 Hawaii
19 Southern California
20 Boise St.
21 Connecticut
22 Oregon St.
23 Colorado
24 Texas
25 Wisconsin
And your bottom 10 –
110 La.-Lafayette
111 Eastern Mich.
112 Southern Methodist
113 Middle Tenn. St.
114 Northern Ill.
115 Rice
116 Utah St.
117 North Texas
118 Syracuse
119 Florida Int'l
For the most part, I actually like the way this looks, especially because it is relatively objective.

What does this mean relative to KevinH’s question? Well, it is going to help me determine what a quality loss or win is. Florida’s loss to the number 1 and number 26 (Auburn’s spot) on the Power Poll looks better than, say USC’s loss to the 77th ranked team on this poll (Stanford, of course).

One final note – For those of you wondering why Oregon St (3-3) and Colorado (4-2) are so high in the Power Poll, and why Cal is not in the top 25, know that–

Oregon State – +110 Net Average Yardage, Past Schedule Rank 8th
Colorado - +71.67 Net Average Yardage, Past Schedule Rank 12th
Cal - +37.8 Net Average Yardage, Past Schedule Rank 89th

In other words, the first have played well, and against difficult opponents, despite the losses. Cal has played marginally positive in the net yardage column, against teams with an 8-11 record, good for only 89th nationally.

12 comments:

Henry Louis Gomez said...

Batten down the hatches.

Anonymous said...

Ha! I read your column almost everyday, and that's the first time I've commented! And I get a whole article! Thanks! By the way, I love the site and the statistical analysis. I've gotten sick of the same old shit from the announcers regurgitating the same thing the last guy said. Thanks for the originality and Go Gators!

Anonymous said...

Oh, and one comment about the power poll. You have to weight the win/loss much higher than that (or however you did it). That is THE most important stat out there by far, and you know it (aka resume based polling)

Anonymous said...

Have you thought about using the DVOA% and DAVE% statistics that the guys at FootballOutsiders.com have applied to weighting NFL teams? It's some really cool stuff, and you're bound to get better comparison numbers from 120 teams than 32.

Mergz said...

Kevinh - I agree wins are the most important, that's why I weight it so heavily in my BlogPoll. I was looking for something else here, something to indicate teams "quality" based on some recognized stats.

Anon- I'll take a look at it. Sounds interesting.

Cal Fans (I know you are coming) - I have nothing against Cal, they are a top 3 team in my BlogPoll. But they haven't played a tough schedule.

This is a work in progress. I plan on tinkering with it.

Anonymous said...

I realize that it's blog and you can poll if you want to, and the idea is intriguing.

However, while the net yardage does quantify the relative strengths of the defenses and offenses, as you correctly note, that does not always lead to points.

One wonders what the impact of a % of possessions scored less opponents % of possessions scored might yield or including a giveaway /takeaway ratio as a category.

Certainly this methodology is as valid as any other methodology used to prove the unprovable.

Anonymous said...

The NCAA strength of schedule numbers are incorrect. Cal's opponents have gone a combined 8-16 not 8-11 versus D-1A competition. I understand that this makes Cal's schedule look worse, but it also calls into question the accuracy of the statistics employed.

Anonymous said...

Net yards completely ignores an important aspect of the game: Special Teams. Teams can win and lose games on special teams, yet special teams play would only have an incidental effect on the net yardage numbers. Any effect would be woefully insufficient to account for such a crucial part of the game. I could attempt to refute your rankings, but I will not bother myself over rankings so fundamentally flawed.

I hope that other Cal fans follow suit and not dignify this analysis with a response.

Anonymous said...

I'll ignore my advice and continue ripping this to shreds.

The strength of schedule stats as calculated by the NCAA encourage teams to play D-1AA teams instead of bad D-1A teams. Is a winless D-1A team better than a winless D-1AA team? Probably, but not if you listen to the NCAA. On one hand you have 12 losses added to your opponent's winning percentage, and on the other you have a non-existent game. Both are cupcakes, but the lesser is ignored.

Is an 85 yard touchdown run better than a 40 yard touchdown run? No, unless there is some chance of the opponent catching the player before he reaches the end-zone; however, the Net Yardage Calculation disagrees. A TD should be a TD, but apparently one is 45 yards better than the other, and thus indicative of a better team.
Consider this situation: Two teams have identical records, and play identical games except the proceeding situation occurs twice in each of their games which results in one team having an additional 90 yards of total offense per game (lets assume that special teams play accounts for the differences in field position). A simple numerical analysis of the two teams would result in a significant advantage for one team. Is this advantage justified? Absolutely not. Statistics have meaning, but not when you fail to account for outside variables.

Mergz said...

Anon – The NCAA strength are correct. Cal’s opponents have gone 8-11 when you take out Cal’s record against them (5-0). The NCAA calculates SOS this way, and it is the only way it makes sense because you can’t get an accurate SOS when you include your own efforts against them.

You might want to check out what you are complaining about before you put it to print, so to say. Ignorance tends to “shred” one’s authority.

As for Net Yardage, it doesn’t account for special teams play that it true. Obviously Win-Loss records do to a certain extent as them encompass “everything”. How exactly does one weight special teams? They certainly are not equal to the overall offensive-defensive effort.

In regard to Cal, where do their special team stats rate? Well,

Punt Returns – 27th nationally
Kickoff Returns – 28th nationally
Net Punting – 21st nationally

I’m not sure these stats would do much to pull Cal into the top 25 were they added.

Sorry to drive you to such uncontrollable rage that you just had to keep posting. These stats weren’t devised as some way to take a slap at Cal. The facts are that Cal has played less than a challenging schedule, and given up a ton of yards and points on D. Yet, they are 5-0, which speaks for itself.

Isn’t that the statement you want to make?

Anonymous said...

The NCAA numbers might be correct, yet it still ignores D-1AA competition (I know for a fact that it ignores them, because Cal's opponents would have a 10-11 record). Any calculation that ENCOURAGES playing teams from a lower division is absurd.

Secondly, what is the value of the record of one's opponents? Ohio State defeated a previously undefeated Purdue, and thus gets 5 wins added to their opponents win record. Do these 5 wins imply that Purdue is a good team? Their past opponent's record ranks #76 thus calling into question the merit of Purdue, yet it is weighted equally with Missouri's win over Illinois (Illinois' schedule being ranked #2). That is an attempt to calculate SOS by looking at one number when there are other important factors.

Special Teams are not equal to overall offensive or defensive effort? I'm not sure I can agree with that. I have seen too many instances where games have been won or lost on special teams to flatly ignore them. How many blocked kicks did Florida have against USC last year? If special teams play was as trivial as you portray it, then Florida would not have been National Champions last year.

The point I want to make is that abstract numbers mean nothing. Football games are not played in a vacuum, so do not try to look at numbers in a vacuum.

Mergz said...

Anon - there "ain't" nothing perfect out there, and this qualifies, though it is as good as anything else I have seen. Plus, I think your worry is aimed in the wrong direction. Take a look at our actual BlogPoll vote in the post above. You will see we put Cal 3rd (and it was a very close call against OSU at 2nd). Our own Gators are unranked.

This Power Poll (as I call it) exercise is a help to me in assessing quality wins and losses in my BlogPoll voting. I think it shows strength of many programs perhaps overlooked (I think Kansas and Arizona State are better than anyone seems to think they are). But it isn't my opinion as far as "rankings" go. That is,again, the post above.