Aikman efficiency ratings

[BlindFaith]

Interested if anyone has taken a look at these?
http://www.aikman.com/AikmanEfficie...an-Efficiency-Ratings-2014-as-of-Week-11.aspx

For the first 7 weeks of the season, that traditional rate was lagging behind in 2014. At that point teams with higher Aikman Game Ratings were winning at only a 74% clip.

But that has changed in a big way in the last 4 weeks, as teams with the higher Aikman Game Rating are 54-1, with an undefeated score of 40-0 in the last three weeks. It puts the season winning percentage back up at its traditional 83%.

That’s three consecutive undefeated weeks after having only six in recorded Aikman history which goes back to 2005.

Rank Record Team Combined Off Def Off Def Combined
1 7-3 Broncos 166.1 95.1 71 3 5 8
2 7-3 Packers 164.3 92.1 72.2 9 25 34
3 7-3 Chiefs 162.6 90.3 72.2 23 8 31
4 8-2 Patriots 161.2 93.3 68 7.5 16 23.5
5 6-4 Ravens 161.1 84.6 76.5 12 11 23
6 9-1 Cardinals 160 81.8 78.2 20 13 33
7 7-3 Cowboys 157 88.6 68.4 6 15 21
8 6-4 Dolphins 156.5 79.6 76.9 16 2 18
9 6-4 Seahawks 155.9 89.1 66.9 11 3 14
10 5-5 Texans 155.6 81 74.5 15 30 45
11 7-3 Lions 154.9 76 78.9 21 1 22
12 6-4 Browns 154.9 81.5 73.4 14 22 36
13 6-4 Colts 153 89.9 63.2 1 24 25
14 4-6 Saints 151.2 90.4 60.7 2 20 22
15 6-3-1 Bengals 151 82.8 68.2 17 28 45
16 6-4 Chargers 150.1 85.1 65 22 9 31
17 7-4 Steelers 150.1 84.2 66 5 14 19
18 6-4 49ers 150 77.2 72.8 18 4 22
19 4-6 Falcons 148.3 85.3 63 10 32 42
20 5-5 Bills 147.8 73 74.8 25 6 31
21 7-3 Eagles 146.9 77.7 69.2 4 26 30


Based on this, I'm not sure why the Ravens and Browns aren't getting more love for playoff hopes. Based on remaining schedule and applying the Aikman ratings, I see the top 6 for each conference as follows

Patriots
Broncos
Chiefs
Ravens
Colts
Browns

Packers
Cardinals
Cowboys
Lions
Seahawks
Eagles

 

[percyhoward]

These are the factors that go into Aikman's rating. (what I wish he'd use instead)

yards per rush (first down percentage per rush)
yards per pass play (pass rating, or adjusted yards per pass)
turnovers per game (turnovers per drive)
first downs per game (drive success rate)
third-down percentage (eliminate - use above stat instead)
points per game (points per drive)
points per red zone possession (keep it!)

Still, AER is much better than total yards any day.

 

[BlindFaith]

These are the factors that go into Aikman's rating. (what I wish he'd use instead)

yards per rush (first down percentage per rush)
yards per pass play (pass rating, or adjusted yards per pass)
turnovers per game (turnovers per drive)
first downs per game (first downs per drive)
third-down percentage (eliminate - use above stat instead)
points per game (points per drive)
points per red zone possession (keep it!)

Still, AER is much better than total yards any day.

I could agree with you on points per game. Points per drive is better. Points per play would be best.

 

[percyhoward]

I could agree with you on points per game. Points per drive is better. Points per play would be best.

No way. If you're going to score on a drive, you'd rather it take as many plays as possible. Points per play treats the offense that scores on a 9-play drive the same as the one that goes three-and-out twice then scores on a three-play drive.

 

[DallasDW00ds0n]

I agree... The more plays the better if we're talking about drives ending in TDs

 

[BlindFaith]

No way. If you're going to score on a drive, you'd rather it take as many plays as possible. Points per play treats the offense that scores on a 9-play drive the same as the one that goes three-and-out twice then scores on a three-play drive.

I know you favor per drive stats. I prefer per play stats.

If you look at the top offenses on a point per play basis, you cant argue the rankings. And the formula takes account for third down % and first downs per game. In your scenario, the team with 2 three and outs would be ranked lower because of the poor third down conversions and low first downs.

If you look defensively at points per drive, currently Arizona and Detroit are the leaders. Then you have Philly and Dallas. The 49ers, Bills, Texans, Browns, Ravens and Vikings round out the top 10.

Aikman has the defenses ranked as below:
Lions
Cardinals
Dolphins
Ravens
Bills
Texans
Browns
49ers
Chiefs
Packers
Broncos
Eagles
Rams
Vikings
Cowboys

Points per play rank the teams as follows:
1Detroit
2Miami
3Cleveland
4Arizona
5Kansas City
6Baltimore
7Houston
8Cincinnati
9Buffalo
10San Diego
11Green Bay
12Denver
13New England
14San Francisco
15Minnesota
16Philadelphia
17Dallas

Clearly there is a big discrepancy between the top teams at (points per drive) vs (Aikmans efficiency and points per play). The Dolphins aren't in the top 10 for points per drive. Philly and Dallas are 3 and 4 and there is no way either of those defenses should be that high. 49ers and Vikings are high in points per drive but are not as high in the other ratings.

Hey, I get it. Dallas looks good on a points per drive ranking. So you want to stick with that. I get it. But I personally feel that that the other two rankings, not to mention the DVOA rankings that I have not listed, clearly show that Dallas is not a top 4 defense as the points per drive stat you keep bringing up shows.

 

[big dog cowboy]

3 7-3 Chiefs

That didn't take long.

 

[BlindFaith]

That didn't take long.

lol I guess that's why they play the games.

 

[kramskoi]

I know you favor per drive stats. I prefer per play stats.

If you look at the top offenses on a point per play basis, you cant argue the rankings. And the formula takes account for third down % and first downs per game. In your scenario, the team with 2 three and outs would be ranked lower because of the poor third down conversions and low first downs.

If you look defensively at points per drive, currently Arizona and Detroit are the leaders. Then you have Philly and Dallas. The 49ers, Bills, Texans, Browns, Ravens and Vikings round out the top 10.

Aikman has the defenses ranked as below:
Lions
Cardinals
Dolphins
Ravens
Bills
Texans
Browns
49ers
Chiefs
Packers
Broncos
Eagles
Rams
Vikings
Cowboys

Points per play rank the teams as follows:
1Detroit
2Miami
3Cleveland
4Arizona
5Kansas City
6Baltimore
7Houston
8Cincinnati
9Buffalo
10San Diego
11Green Bay
12Denver
13New England
14San Francisco
15Minnesota
16Philadelphia
17Dallas

Clearly there is a big discrepancy between the top teams at (points per drive) vs (Aikmans efficiency and points per play). The Dolphins aren't in the top 10 for points per drive. Philly and Dallas are 3 and 4 and there is no way either of those defenses should be that high. 49ers and Vikings are high in points per drive but are not as high in the other ratings.

Hey, I get it. Dallas looks good on a points per drive ranking. So you want to stick with that. I get it. But I personally feel that that the other two rankings, not to mention the DVOA rankings that I have not listed, clearly show that Dallas is not a top 4 defense as the points per drive stat you keep bringing up shows.

Nope, not a top 5 defense by DVOA and I doubt that they have been since the last SuperBowl run. A bit unnerving actually. FO currently has them ranked 21 (4.4%), which is slightly worse than they were before the two losses. They could go 6-0 down the stretch and still not likely break into the top 5, unless those six wins are really defensively dominant like the Seattle game, which I think is still the highest DVOA performance by a defense this year...so they are certainly capable of such a feat but for six straight weeks I would say not. Unless you watched those SuperBowl defenses, you can't appreciate the level of their dominance at the time. The current edition has a way to go before getting to such levels.

That said...the Colts, and Packers showed that you don't need an "all-world" defense to get some hardware. That may become the norm in the parity-driven, salary cap era of football.

 

[Beast_from_East]

So based on the Aikman ratings, we are the 3rd best team in the NFC behind Arizona and Green Bay?

Am I reading this correctly?

 

[percyhoward]

Hey, I get it. Dallas looks good on a points per drive ranking. So you want to stick with that...I get it. Dallas is not a top 4 defense as the points per drive stat you keep bringing up shows.

You're wrong on both points. Defensively, Dallas ranks 12th in points per drive -- not 4th.

Dallas could rank 1st or 32nd, but that wouldn't change the fact that per drive stats have stronger win correlations than per play stats, for the following reason: An 8-play drive that results in a TD (7 points, 0.88 PPP) will win more games than a 3-play drive that results in a FG (3 points, 1.00 PPP).

You don't pick a stat based on where your favorite (or least favorite) team ranks according to that stat. You go by the stats that lead to wins.

 

[BlindFaith]

You're wrong on both points. Defensively, Dallas ranks 12th in points per drive -- not 4th.

Dallas could rank 1st or 32nd, but that wouldn't change the fact that per drive stats have stronger win correlations than per play stats, for the following reason: An 8-play drive that results in a TD (7 points, 0.88 PPP) will win more games than a 3-play drive that results in a FG (3 points, 1.00 PPP).

You don't pick a stat based on where your favorite (or least favorite) team ranks according to that stat. You go by the stats that lead to wins.

What stats are you using for your points per drive?

 

[percyhoward]

What stats are you using for your points per drive?

12th
11th

If you do some research, you'll find out that points per drive differential (offense's PPD - defense's PPD) has the highest win correlation of any metric. In the best study I've seen so far (1995-2012), the correlation coefficient to win % of PPD was .92, pass rating differential was .81, and nothing else was as high as .70. Drive Success Rate differential was not included in that 18-year study, but is always up there with those other two in any seasonal study. So those three metrics give you a pretty good idea of where your defense ranks.

Through Week 11, the Cowboys' defense is at 12th in PPD, 13th in DPR, and 9th in DSR.

 

[BlindFaith]

12th
11th

If you do some research, you'll find out that points per drive differential (offense's PPD - defense's PPD) has the highest win correlation of any metric. In the best study I've seen so far (1995-2012), the correlation coefficient to win % of PPD was .92, pass rating differential was .81, and nothing else was as high as .70. Drive Success Rate differential was not included in that 18-year study, but is always up there with those other two in any seasonal study. So those three metrics give you a pretty good idea of where your defense ranks.

Through Week 11, the Cowboys' defense is at 12th in PPD, 13th in DPR, and 9th in DSR.

http://www.sportingcharts.com/nfl/stats/opponent-scoring-efficiency-percentage/2014/

Thats what I used. Scoring efficiency. Intereesting that the number of drives are different. And Im not sure why the big difference in rankings. Youd think scoring efficiency would correlate with points per drive.

And I bet if you could do that same study but use points per play differential youd result in the same, if not better correlation to wins.

 

[percyhoward]

That's what I used. Scoring efficiency. Interesting that the number of drives are different. And I'm not sure why the big difference in rankings. You'd think scoring efficiency would correlate with points per drive.

Yeah, that's not points per drive. That link you provided is a stat that treats all scores the same, whether they're touchdowns or field goals. They call the stat "scoring efficiency," which sounds good, but they don't tell you they're making no discrimination between drives that result in 3 points and drives that result in 7 points. That site appears to shoot for quantity over quality. It's a useless stat, IOW.

Before you draw any conclusions about a stat, it's a good idea to know what's actually being measured.

And I bet if you could do that same study but use points per play differential you'd result in the same, if not better correlation to wins.

You'd lose that bet, because it doesn't matter how many plays it takes you to score, and in fact, in normal game situations, you'd prefer it to take as many plays as possible because you'd rather have your offense on the field than your defense. The idea is to score points, however many plays it takes. There's no advantage to be gained by limiting your number of plays on offense. In fact, high-scoring offenses typically have more plays and more points, and yet you're pitting those two factors against each other with "points per play." All of this probably goes toward explaining why per-drive stats always perform better than per-play stats with respect to win correlation.

It's early in the 3rd quarter of a tie game, and you have the ball. Would you rather score 3 points or 7 points on this drive?

Now, does your answer really change if the 3 points come at the end of a 6-play drive, and the 7 points come at the end of a 15-play drive?