The 2015 Dallas Cowboys and the Myth of DeMarco Murray

[percyhoward]

I'm not saying that there isn't value with looking at stats on a per drive basis. But there are too many factors that play into making that as valuable as per play stats. Not all drives are equal. Two minute warning drives, drives where you are forced to pass, drives that end due to turnover. There becomes an inherit weighting to drive based stats. Per play stats are the least weighted way of looking at things.

Not all plays are equal either, and it makes no sense to treat 10 plays spread out over four drives and three punts the same as 10 plays all together in one drive. Obviously. At least, this should be obvious.

Yards per play is really just yards per pass attempt + yards per rushing attempt. If you were only looking at yards per pass play, that would at least give you something with a decent win correlation. But since yards per rushing play has no correlation with wins whatsoever, you're combining a "pretty good" stat with a horrible stat. You end up with a fairly mediocre stat, and that's no way to judge the performance of a defense. Check in with the real world to see how "yards per play" ranks the teams.

Compare the two stats, using the differential (offense minus defense). The better stat will rank the teams with the most wins at the top. There were five NFL teams that finished with 12 wins in 2014. Patriots, Seahawks, Packers, Cowboys, and Broncos. Follow them through these two sets of rankings.

YARDS PER PLAY
1 Denver
2 Seattle
3 Green Bay
4 Baltimore
5 Indianapolis
6 Detroit
7 Dallas
8 Philadelphia
9 Pittsburgh
10 Kansas City
11 New England

Notice how "yards per play" takes these five teams and spreads them out over the rankings, even valuing the performances of 10 other teams over that of the world champions. Were the Eagles and Chiefs really better than the Patriots in 2014? "Yards per play" says they were. But "yards per play" doesn't care how many points you score or give up, because it makes no distinction between points and yards. It doesn't measure scoring at all. Got a great red zone team? This stat can't see it. Got a defense that gives up big plays, but that's top 10 in scoreless drives faced (as Dallas was in 2014)? This stat is blind to that too.

Keep following those same five NFL teams. This is how they ranked according to Points per Drive. All five 12-win teams made up the top five.

POINTS PER DRIVE
1 Green Bay
2 New England
3 Seattle
4 Denver
5 Dallas

Football Outsiders, Advanced Football Analytics. I'm sure you're familiar with them. I put much more stock in what they have done to really evaluate defense, offense and specific positions.

Then you already know that Football Outsiders has the defense moving up 13 spots from the end of 2013 to the end of 2014, using the weighted ranking that gives less importance to early-season games.

Football Outsiders Weighted Defensive Efficiency
Cowboys' NFL rank after week 17
2013 32nd
2014 19th

Who knows how this correlates to wins, but it's probably a pretty good stat just based on the fact that it basically agrees with the defense's move up the points per drive rankings from 30th to 16th.

 

[BlindFaith]

Not all plays are equal either, and it makes no sense to treat 10 plays spread out over four drives and three punts the same as 10 plays all together in one drive. Obviously. At least, this should be obvious.

Yards per play is really just yards per pass attempt + yards per rushing attempt. If you were only looking at yards per pass play, that would at least give you something with a decent win correlation. But since yards per rushing play has no correlation with wins whatsoever, you're combining a "pretty good" stat with a horrible stat. You end up with a fairly mediocre stat, and that's no way to judge the performance of a defense. Check in with the real world to see how "yards per play" ranks the teams.

Compare the two stats, using the differential (offense minus defense). The better stat will rank the teams with the most wins at the top. There were five NFL teams that finished with 12 wins in 2014. Patriots, Seahawks, Packers, Cowboys, and Broncos. Follow them through these two sets of rankings.

YARDS PER PLAY
1 Denver
2 Seattle
3 Green Bay
4 Baltimore
5 Indianapolis
6 Detroit
7 Dallas
8 Philadelphia
9 Pittsburgh
10 Kansas City
11 New England

Notice how "yards per play" takes these five teams and spreads them out over the rankings, even valuing the performances of 10 other teams over that of the world champions. Were the Eagles and Chiefs really better than the Patriots in 2014? "Yards per play" says they were. But "yards per play" doesn't care how many points you score or give up, because it makes no distinction between points and yards. It doesn't measure scoring at all. Got a great red zone team? This stat can't see it. Got a defense that gives up big plays, but that's top 10 in scoreless drives faced (as Dallas was in 2014)? This stat is blind to that too.

Keep following those same five NFL teams. This is how they ranked according to Points per Drive. All five 12-win teams made up the top five.

POINTS PER DRIVE
1 Green Bay
2 New England
3 Seattle
4 Denver
5 Dallas


Then you already know that Football Outsiders has the defense moving up 13 spots from the end of 2013 to the end of 2014, using the weighted ranking that gives less importance to early-season games.

Football Outsiders Weighted Defensive Efficiency
Cowboys' NFL rank after week 17
2013 32nd
2014 19th

Who knows how this correlates to wins, but it's probably a pretty good stat just based on the fact that it basically agrees with the defense's move up the points per drive rankings from 30th to 16th.

You keep referring to correlation to winning. The only Stat that matters in that regard is that the team that scores more than the other team wins 100 percent of the time.

I use stats to evaluate performance. Not try and find some magic number that will tell me who will win.

 

[DandyDon52]

Not all plays are equal either, and it makes no sense to treat 10 plays spread out over four drives and three punts the same as 10 plays all together in one drive. Obviously. At least, this should be obvious.

Yards per play is really just yards per pass attempt + yards per rushing attempt. If you were only looking at yards per pass play, that would at least give you something with a decent win correlation. But since yards per rushing play has no correlation with wins whatsoever, you're combining a "pretty good" stat with a horrible stat. You end up with a fairly mediocre stat, and that's no way to judge the performance of a defense. Check in with the real world to see how "yards per play" ranks the teams.

Compare the two stats, using the differential (offense minus defense). The better stat will rank the teams with the most wins at the top. There were five NFL teams that finished with 12 wins in 2014. Patriots, Seahawks, Packers, Cowboys, and Broncos. Follow them through these two sets of rankings.

YARDS PER PLAY
1 Denver
2 Seattle
3 Green Bay
4 Baltimore
5 Indianapolis
6 Detroit
7 Dallas
8 Philadelphia
9 Pittsburgh
10 Kansas City
11 New England

Notice how "yards per play" takes these five teams and spreads them out over the rankings, even valuing the performances of 10 other teams over that of the world champions. Were the Eagles and Chiefs really better than the Patriots in 2014? "Yards per play" says they were. But "yards per play" doesn't care how many points you score or give up, because it makes no distinction between points and yards. It doesn't measure scoring at all. Got a great red zone team? This stat can't see it. Got a defense that gives up big plays, but that's top 10 in scoreless drives faced (as Dallas was in 2014)? This stat is blind to that too.

Keep following those same five NFL teams. This is how they ranked according to Points per Drive. All five 12-win teams made up the top five.

POINTS PER DRIVE
1 Green Bay
2 New England
3 Seattle...........**
4 Denver
5 Dallas

Who knows how this correlates to wins, but it's probably a pretty good stat just based on the fact that it basically agrees with the defense's move up the points per drive rankings from 30th to 16th.

Well with seattle # 3 that means wilson has to be a top QB all of the top 5 have vg QB's

 

[percyhoward]

I use stats to evaluate performance. Not try and find some magic number that will tell me who will win.

To evaluate means to find the value of something. You can't evaluate anything without a standard. If that standard isn't "things that lead to wins", then what is it? A performance that leads to losing couldn't possibly have more value than one that leads to winning. It's not magic, it's just science. And it doesn't necessarily have to be predictive -- it might be descriptive (not "who will win," but "why they won.") You can analyze teams that win to see what they have most in common.

A 15-year study (3,720 games) found that winning teams usually had more passing yards per play than losing teams, for example. But almost as often as not, they had fewer rushing yards per play. That alone tells us that yards per play is not a good stat, since those are the only two ways to gain yards from scrimmage. Below, note that the win correlation of yards per drive differential is only .69, which explains why the Chiefs and Eagles rank higher than the Patriots in the "per play" version of this stat. It's not a good stat. Yards don't win games, points do.

Correlation Coefficient to Win %
(all are differentials)
Points Per Drive .92
Pass Rating .81
Yards Per Drive .69
3rd Down Conversions .64
Turnover Differential .64

You keep referring to correlation to winning. The only Stat that matters in that regard is that the team that scores more than the other team wins 100 percent of the time.

No way is it the only stat that matters, it's just the one that matters most. If you're doing something that wins games 90% of the time (or 10% of the time), you think that's equally important as doing something that wins 50% of the time?

You said you "put much more stock" in the analysis of Football Outsiders. You should know that Football Outsiders doesn't measure pure yardage anyway. The closer the yards are to a first down or touchdown, and the greater the likelihood that the yards gained extend a drive, the more they're worth. That's how they reward scoring. That's why our defense went up from 32nd to 19th -- they ranked in the top 10 in percentage of scoreless drives faced. Is the evaluation of our defense by Football Outsiders suddenly not important anymore, just because it doesn't agree with yours (that ranks the Chiefs and Eagles above the Patriots)?

 

[percyhoward]

Well with seattle # 3 that means wilson has to be a top QB.

That's the whole team -- not just the offense.

Seattle, 2014
Points (and points allowed) per drive
offense 2.19 (9th)
defense 1.44 (2nd)
differential +.75 (3rd)

With an average defense, the Seahawks wouldn't even be a top 10 team, much less top 5.

 

[BlindFaith]

To evaluate means to find the value of something. You can't evaluate anything without a standard. If that standard isn't "things that lead to wins", then what is it? A performance that leads to losing couldn't possibly have more value than one that leads to winning. It's not magic, it's just science. And it doesn't necessarily have to be predictive -- it might be descriptive (not "who will win," but "why they won.") You can analyze teams that win to see what they have most in common.

A 15-year study (3,720 games) found that winning teams usually had more passing yards per play than losing teams, for example. But almost as often as not, they had fewer rushing yards per play. That alone tells us that yards per play is not a good stat, since those are the only two ways to gain yards from scrimmage. Below, note that the win correlation of yards per drive differential is only .69, which explains why the Chiefs and Eagles rank higher than the Patriots in the "per play" version of this stat. It's not a good stat. Yards don't win games, points do.

Correlation Coefficient to Win %
(all are differentials)
Points Per Drive .92
Pass Rating .81
Yards Per Drive .69
3rd Down Conversions .64
Turnover Differential .64


No way is it the only stat that matters, it's just the one that matters most. If you're doing something that wins games 90% of the time (or 10% of the time), you think that's equally important as doing something that wins 50% of the time?

You said you "put much more stock" in the analysis of Football Outsiders. You should know that Football Outsiders doesn't measure pure yardage anyway. The closer the yards are to a first down or touchdown, and the greater the likelihood that the yards gained extend a drive, the more they're worth. That's how they reward scoring. That's why our defense went up from 32nd to 19th -- they ranked in the top 10 in percentage of scoreless drives faced. Is the evaluation of our defense by Football Outsiders suddenly not important anymore, just because it doesn't agree with yours (that ranks the Chiefs and Eagles above the Patriots)?

I agree with everything you said.

But football is a team game. Its really three teams. Offense, defense and special teams. You could have a great offense but a bad defense. You could have the opposite.

You can have a great oline and a awful running back. You can have a great WR but a terrible QB. Awesome secondary but no pass rush.

Points per drive is much better than overall points. But to me, its still too subjective to be used in evaluation. It's perfect to use in showing that the team that scores more than the other team wins.

But when you start breaking down drives, you see there are many variables that can be deceiving. Take our Cowboys last year. Second in the league in takeaways. So that means many otherwise good offensive drives ended without a score. We know that the defense was less than average. And can you count on takeaways every game? Points per drive factor in takeaways. Points per play eliminate some of that.

http://www.sportingcharts.com/nfl/stats/points-per-play-differential/2014/
Points Per Play Differential
TeamPoints Per PlayPoints Per Play AgainstPoints Per Play Differential
1Green Bay Packers0.4860.3300.155
2New England Patriots0.4360.3030.133
3Seattle Seahawks0.3860.2750.111
4Baltimore Ravens0.4010.2920.109
5Denver Broncos0.4520.3430.108
6Dallas Cowboys0.4610.3600.101
7Kansas City Chiefs0.3670.2740.093
8Houston Texans0.3500.2850.065
9Philadelphia Eagles0.4210.3590.061
10Buffalo Bills0.3360.2810.055

That's a pretty good indicator there.

I was saying mid season that our defense wasn't very good. I had that discussion with you. You didn't agree. Turned out our defense wasn't very good. I'll stick with using the stats that I think that matter along with watching the games. If the stats you use work in how you evaluate things, then keep using them. Doesn't mean either of us are wrong. Stats are great. But making sense of them is what matters.

 

[percyhoward]

Points Per Play Differential
TeamPoints Per PlayPoints Per Play AgainstPoints Per Play Differential
1Green Bay Packers0.4860.3300.155
2New England Patriots0.4360.3030.133
3Seattle Seahawks0.3860.2750.111
4Baltimore Ravens0.4010.2920.109
5Denver Broncos0.4520.3430.108
6Dallas Cowboys0.4610.3600.101
7Kansas City Chiefs0.3670.2740.093
8Houston Texans0.3500.2850.065
9Philadelphia Eagles0.4210.3590.061
10Buffalo Bills0.3360.2810.055

That's a pretty good indicator there.

Like I said, points are more meaningful than yards. I'm glad you see that now. Now go back to that page, click the 2nd column (marked "points per play against"), and you'll see that the Cowboys' defense ranked 18th. That's even better than Football Outsiders' 19th. And remember, these are the two sources that you recommended, telling you Dallas had a slightly below average defense.

Points per drive factor in takeaways. Points per play eliminate some of that.

If you want to eliminate the influence of Dallas' takeaways on points per drive, you can go to PFR's Drive Finder and filter out the drives that ended in takeaways. Dallas was still 14th in percentage of scoreless drives faced.

I was saying mid season that our defense wasn't very good. I had that discussion with you. You didn't agree. Turned out our defense wasn't very good.

No, you were saying that the defense wasn't much better than 2013, and that it only looked better because of our running game. Which wasn't true at the time, and wasn't true at the end of the season either.

Like I've said several times in this thread, we were top 10 in percentage of drives stopped (scoreless drives faced). Remember when Adam explained why our yards per play numbers made us look bad, despite our high ranking in points per drive? Because we stopped drives without doing the kinds of things that affect yards (sacks, negative plays), and we allowed a lot of 3rd-down completions short of the first down. I assume you understood his post, because you gave it a like.

Not much has changed since then, except that you're no longer extolling the virtues of yards per play.

 

[BlindFaith]

Like I said, points are more meaningful than yards. I'm glad you see that now. Now go back to that page, click the 2nd column (marked "points per play against"), and you'll see that the Cowboys' defense ranked 18th. That's even better than Football Outsiders' 19th. And remember, these are the two sources that you recommended, telling you Dallas had a slightly below average defense.


If you want to eliminate the influence of Dallas' takeaways on points per drive, you can go to PFR's Drive Finder and filter out the drives that ended in takeaways. Dallas was still 14th in percentage of scoreless drives faced.


No, you were saying that the defense wasn't much better than 2013, and that it only looked better because of our running game. Which wasn't true at the time, and wasn't true at the end of the season either.

Like I've said several times in this thread, we were top 10 in percentage of drives stopped (scoreless drives faced). Remember when Adam explained why our yards per play numbers made us look bad, despite our high ranking in points per drive? Because we stopped drives without doing the kinds of things that affect yards (sacks, negative plays), and we allowed a lot of 3rd-down completions short of the first down. I assume you understood his post, because you gave it a like.

Not much has changed since then, except that you're no longer extolling the virtues of yards per play.

Nope, yards per play is still something I factor in. Nothing has changed. And I still hold fast that there is value there, at least to me for what I look at. A team that gives up a lot of yards on a per play basis will get exposed. We managed some of that during the season. We ranked higher in points per play than yards per play. Turnovers played a big part in that. Playing subpar QBs helped with that as well. But look at what a hobbled Rodgers did to this team. Exactly what the stats indicated. They gave up huge chunks of yards. But in the playoffs they weren't able to stop them from scoring.

http://www.sportingcharts.com/nfl/stats/team-yards-per-play-differential/2014/

I obviously don't look at things in a vacuum. It's a combination of many different things. Like I said, use whatever stats you want. I'll use mine. And honestly, I'd much rather just watch the games than to play around with numbers. I only do that when opinions vary from what my eye balls tell me.

As far as last year goes, yes, this defense was better. I've said it many times. But it was so bad that just having healthy bodies that at least somewhat belong in the NFL would make it better. It still wasn't good though.

 

[Bullet22]

let it go.....