Third-Down Efficiency Key For Cowboys' Offense

[TrailBlazer]

I can guarantee dallas will hit a lot of 3rd downs bc they run a lot on first and second. Their 3rd down conversion rate will be very indicative of how the season plays out.

 

[ShiningStar]

im saying this more tongue in cheek than ignorance, but anyone watching our games would get it.

Want to be successful at 3rd down, play Dallas.

 

[percyhoward]

I don't understand why converting on 3rd down is any more important than converting on first or second down. If an offense goes down the field 9 times out of 10 without needing to convert a 3rd down and scores TDs each time and yet 1/10 times gets a 3rd down that it fails to convert it would be a great offense and yet have a 0% 3rd down conversion percentage.

Drive Success Rate is a much better stat than 3rd-down conversion %. It's the percentage of down series that result in a first down (or TD), so it ignores whether the conversion happens on 1st, 2nd, 3rd, or 4th down.

Cowboys' Offensive Rankings
Drive Success Rate
2014 4th
2013 9th

3rd Down Conversion %
2014 2nd
2013 26th

 

[Eskimo]

Drive Success Rate is a much better stat than 3rd-down conversion %. It's the percentage of down series that result in a first down (or TD), so it ignores whether the conversion happens on 1st, 2nd, 3rd, or 4th down.

Cowboys' Offensive Rankings
Drive Success Rate
2014 4th
2013 9th

3rd Down Conversion %
2014 2nd
2013 26th

The thing I don't like about DSR is that it is throwing out a lot of useful information. For example, a drive that starts on your 1-yard line and ends at the opponents 1-yard line has generated about 4 points worth of expected value but it is a negative drive. Now take a drive that starts at the opponent 1-yard line and ends in a FG. That is a successful drive but actually generated -2.5 points compared to expected result when starting with that field position. Or for a more common scenario, the offense gets the ball on the 20-yard line of the opponent but gets stopped and kicks a FG. That is a "successful" drive accoring to DSR but really didn't result in positive points compared to the expected value of that drive.

Nevertheless, I like the stat you showed. The name of the game is scoring points, not converting on 3rd downs.

 

[percyhoward]

The thing I don't like about DSR is that it is throwing out a lot of useful information. For example, a drive that starts on your 1-yard line and ends at the opponents 1-yard line has generated about 4 points worth of expected value but it is a negative drive. Now take a drive that starts at the opponent 1-yard line and ends in a FG. That is a successful drive but actually generated -2.5 points compared to expected result when starting with that field position. Or for a more common scenario, the offense gets the ball on the 20-yard line of the opponent but gets stopped and kicks a FG. That is a "successful" drive accoring to DSR but really didn't result in positive points compared to the expected value of that drive.

DSR is simply number of first downs, divided by the number of down series. (Dallas had 313 first downs on 420 down series last year, for a season DSR of 74.5%.) So in your last scenario, the DSR would be 0% (0 of 1), unless that offense got a first down at the 10 before kicking the FG. Then it would be 50% (1 of 2). The drive that starts at the opponent's 1-yard line and ends in a FG would have a DSR of 0% because it didn't result in any first downs or TD.

A drive that starts on your own 1 and ends at the opponent's 1 is anything but a negative drive. If you make 7 first downs along the way, you'd end up with a DSR of 88% (7 of 8) after getting stopped at the 1. The longer the drive, the greater the potential for a high DSR. The closer the drive begins to the opponent's goal line, the lower the potential for a high DSR on that drive. If you start a drive with 1st-and-goal, the only possibilities on that drive are a DSR of 0% or 100%.

Every drive ends with either a DSR of 100% (like 6 of 6, for example), or a "first downs" number that's one less than the "down series" number (like 5 of 6). If you compare say, 10 random drives, none of which end in a TD, the drive with the most first downs will have the highest DSR. Any 3-and-out (or turnover on the drive's initial down series) results in an "0 fer" for that particular drive Defensively, our best game was against Seattle, who we held to 50% (9 of 18) for the game.

 

[Eskimo]

DSR is simply number of first downs, divided by the number of down series. (Dallas had 313 first downs on 420 down series last year, for a season DSR of 74.5%.) So in your last scenario, the DSR would be 0% (0 of 1), unless that offense got a first down at the 10 before kicking the FG. Then it would be 50% (1 of 2). The drive that starts at the opponent's 1-yard line and ends in a FG would have a DSR of 0% because it didn't result in any first downs or TD.

A drive that starts on your own 1 and ends at the opponent's 1 is anything but a negative drive. If you make 7 first downs along the way, you'd end up with a DSR of 88% (7 of 8) after getting stopped at the 1. The longer the drive, the greater the potential for a high DSR. The closer the drive begins to the opponent's goal line, the lower the potential for a high DSR on that drive. If you start a drive with 1st-and-goal, the only possibilities on that drive are a DSR of 0% or 100%.

Every drive ends with either a DSR of 100% (like 6 of 6, for example), or a "first downs" number that's one less than the "down series" number (like 5 of 6). If you compare say, 10 random drives, none of which end in a TD, the drive with the most first downs will have the highest DSR. Any 3-and-out (or turnover on the drive's initial down series) results in an "0 fer" for that particular drive Defensively, our best game was against Seattle, who we held to 50% (9 of 18) for the game.

Okay, I misunderstood the statistic. I thought it was scoring drives were the criteria for success.

It makes more sense to me now and I think the statistic is a reasonable one for characterizing offensive efficiency. It certainly does a much better job than the myopic 3rd down efficiency statistic. It gets rid of prioritizing 3rd downs over the other downs which really makes no sense. Isn't it just better to minimize the number of 3rd downs you face?

I still think there is room for some sort of a statistic that converts yards gained to points scored for each offensive series compared to the expected value of the starting field position. It would also take into account non-scoring drives and turnovers which would make it a very encompassing statistic.

For example, start on your own 1 and score a TD: that is worth 9 points.

Start on your own 25 and score a TD: 7 points

Start on the opponents 1-yard line and score a TD: 1.5 points

Start on your own 1-yard line and drive to mid-field: 3 points

Start on at mid-field and turn the ball over: -4 points

This is just a rough idea about how it would work as I do believe there is quite a bit of data and how much certain field positions are worth. It would nicely tie together the concepts of yards gained and turnovers while taking out the effect of starting field position to judge the true efficiency of an offense.

The reason I bring it up is I know a lot of people have been talking about points per drive as an offensive statistic. But this only counts actual points rather and forgets the impact of yards gained, turnovers and field position (both where you start and where the opposition starts). But all these things have an expected value that can be estimated from the data collected over the years.

 

[percyhoward]

The reason I bring it up is I know a lot of people have been talking about points per drive as an offensive statistic. But this only counts actual points rather and forgets the impact of yards gained, turnovers and field position (both where you start and where the opposition starts). But all these things have an expected value that can be estimated from the data collected over the years.

You can't get any stronger win correlation than the differential of points per drive. I'll always take the descriptive stat over the predictive one because I like to try to understand exactly why something happened, and I think a team that scores a lot with great starting field position would also generally tend to score a lot at the end of a really long drive of their own. Penalize them if it happens once maybe, but the more it keeps happening penalize them less and less. I don't put a whole lot of importance of the effect of strength of schedule either.

Last year some time around week 10 or so I took all the teams that points per drive said would make the playoffs and compared them with all the teams DVOA said would make the playoffs. The reality ended up being right in the middle of what the two metrics said.

 

[jnday]

Maybe we do maybe we dont, no way to tell yet. What we do know is there are 5 guys who can get some serious push to go ahead and convert in those short yardage situations.

Everything about their past history supports the claim that they are not able to do it. Those are the only facts that are proven at this point.

 

[Brooksey]

Hopefully this will mean more passing on 1st down in particular against those base defenses with stacked boxes.

I'm sure the defense would adjust immediately, add a defensive back and start bringing pressure on first down. All of a sudden instead of 2nd and 6 (from that boring run) you may be looking at 2nd and 20 if Romo holds the ball and takes a sack, or 2nd and10 on an incompletion.

The sack and incompletion most likely forces two more passes. If we fail to pick up a first down, we just went three and out in about :45 seconds.

We tried this theory for years with JG before becoming a run team first. Were not going to change now.

 

[jobberone]

I like Sc% and RZ% as well as points scored and 3rd down conversions. The first two say most per drive And PPG overall.

 

[Eskimo]

I'm sure the defense would adjust immediately, add a defensive back and start bringing pressure on first down. All of a sudden instead of 2nd and 6 (from that boring run) you may be looking at 2nd and 20 if Romo holds the ball and takes a sack, or 2nd and10 on an incompletion.

The sack and incompletion most likely forces two more passes. If we fail to pick up a first down, we just went three and out in about :45 seconds.

We tried this theory for years with JG before becoming a run team first. Were not going to change now.

The issue that I have is we strive to be balanced and run the ball. But we were actually very unbalanced in how we operated on first down. This gave the defense a huge advantage of subbing in personnel who are better suited to defend the run. It only makes sense in the chess match part of the game to then start passing more when we have an advantage in the passing game.

If they start putting in another DB then we have a huge advantage in the box and we run the ball more in those situations.

If they start putting in a bunch of coverage LBs who struggle against the run then we run the ball more in those situations.

I think we were letting the defense off the hook when we let them sub into a run defense alignment and still choose to run the ball 75% of the time. That is not smart football. That is fear of being unable to execute offensively in the passing game more than it is confidence in the running game.

The best teams in the league now score a lot - the best way to score a lot is to pass the ball around 52-55% of the time since the rules favor passing and you can gain about 8.0 yards per attempt if you have elite players at QB, WR and TE like we do. This is in contrast to a good running game which typically only nets you about 4.5 yards per attempt.

Anyhow I'm very curious to see if our offensive strategy is tweaked a bit this year now that they don't have to hide the defense anymore. I suspect we'll see a more aggressive and efficient offense that will put up more points than we have ever dreamed of - maybe 33 points per game.

 

[Brooksey]

The issue that I have is we strive to be balanced and run the ball. But we were actually very unbalanced in how we operated on first down. This gave the defense a huge advantage of subbing in personnel who are better suited to defend the run. It only makes sense in the chess match part of the game to then start passing more when we have an advantage in the passing game.

If they start putting in another DB then we have a huge advantage in the box and we run the ball more in those situations.

If they start putting in a bunch of coverage LBs who struggle against the run then we run the ball more in those situations.

I think we were letting the defense off the hook when we let them sub into a run defense alignment and still choose to run the ball 75% of the time. That is not smart football. That is fear of being unable to execute offensively in the passing game more than it is confidence in the running game.

The best teams in the league now score a lot - the best way to score a lot is to pass the ball around 52-55% of the time since the rules favor passing and you can gain about 8.0 yards per attempt if you have elite players at QB, WR and TE like we do. This is in contrast to a good running game which typically only nets you about 4.5 yards per attempt.

Anyhow I'm very curious to see if our offensive strategy is tweaked a bit this year now that they don't have to hide the defense anymore. I suspect we'll see a more aggressive and efficient offense that will put up more points than we have ever dreamed of - maybe 33 points per game.

I understand your theory, there are some others on the board who share your idea's as well. However, I don't think anyone here is smarter than JG or Linehan when it comes to (the chess match) why they ran as much as they did it on first down. This was by design and it worked very well, even against stacked boxes we were picking up 4-5 yards. I personally love running it on first down. I would like to see some more play action/deep throws to Dez on first down to keep them honest.

Second if running the ball 50% of the time worked well to keep the defense off the field, and you improved your defensive personnel why would you change your offensive philosophy? We were ranked 15-22nd in most categories not 1st or second.

The best team in the NFC has been Seattle for the last several years and they run it 55% pass it 45%.

 

[percyhoward]

I think we were letting the defense off the hook when we let them sub into a run defense alignment and still choose to run the ball 75% of the time. That is not smart football. That is fear of being unable to execute offensively in the passing game more than it is confidence in the running game.

I don't think we were worried about the passing game at all. If there was a fear about passing, it was that if we scored too quickly, we might give the opponent an extra possession against our defense. We ran so much on first down so we could eat clock and keep our D off the field.

There's no question that we could score a ton of points throwing the ball, because our highest-scoring quarter was also the only quarter when we passed more than ran.

Here is our pass/run ratio, broken down by quarter:
1st 49.6% pass / 50.4% run
2nd 56.3% pass / 43.7% run
3rd 46.7% pass / 53.3% run
4th 45.9% pass / 54.1% run

Here is the scoring margin by quarter, not including return TD:
1st +33 (95 - 62)
2nd +55 (140 - 85)
3rd +52 (110 - 58)
4th -17 (105 - 122)

 

[RS12]

Third-Down Efficiency Key For Cowboys' Offense

Even more for the defense IMO.

 

[percyhoward]

Even more for the defense IMO.

29th in the league last year, and more importantly 24th in DSR.

 

[Eskimo]

I don't think we were worried about the passing game at all. If there was a fear about passing, it was that if we scored too quickly, we might give the opponent an extra possession against our defense. We ran so much on first down so we could eat clock and keep our D off the field.

There's no question that we could score a ton of points throwing the ball, because our highest-scoring quarter was also the only quarter when we passed more than ran.

Here is our pass/run ratio, broken down by quarter:
1st 49.6% pass / 50.4% run
2nd 56.3% pass / 43.7% run
3rd 46.7% pass / 53.3% run
4th 45.9% pass / 54.1% run

Here is the scoring margin by quarter, not including return TD:
1st +33 (95 - 62)
2nd +55 (140 - 85)
3rd +52 (110 - 58)
4th -17 (105 - 122)

It isn't even necessarily that I want us to pass more although I would like us to pass a bit more.

The big thing is that I want to pass more when it is easier to pass. When the defense is playing us heavily to run because we are running the ball 75% on 1st down and then we run into a stacked box for a 1-yard gain we are making things too easy for the defense. If they want to load up on players who are better defending the run I'd rather we pass 60% of the time and run 40% of the time. We run enough to keep them honest but pass more often since it is to our advantage to pass more in those situations.

I would argue exactly the opposite if teams came out and played us in the nickel while we came out in 12 personnel we should be running the ball about 75-80% of the time until they commit to playing the run game more neutrally.

 

[percyhoward]

It isn't even necessarily that I want us to pass more...When the defense is playing us heavily to run because we are running the ball 75% on 1st down and then we run into a stacked box for a 1-yard gain we are making things too easy for the defense.

Understand your point completely, I'm just saying why I think we did it. Not because we thought we couldn't execute in the passing game, but because quick scores were not the goal of the offense last year. FWIW, I also think we overdid it just a bit with all the running on first down.

But I don't believe that the goal was to maximize scoring. I think it was to maximize the scoring margin. We were willing to trade points for time, in order to give the opposing offense fewer opportunities to score.