mesaSteeler

03-28-2011, 05:55 PM

Breakdown of Pittsburgh Steelers 2010 Offensive Performace by Set

http://www.behindthesteelcurtain.com/2011/3/28/2073805/breakdown-of-2010-offensive-performace-by-set

by Chileburger on Mar 28, 2011 4:00 PM EDT

Those of you who had the patience to work through my previous posts are aware of the fact that I put together an Excel data base on our offensive plays and I've tried to use it to analyze certain aspects of the Steelers 2010 performance.

Now that the season is on the books, I thought that some of you might be interested in an analysis that I performed of offensive performance by Set for the entire 19 game campaign. You'll find plenty of data and tables after the jump.

Before continuing, though, I'd like to give a tip of the hat to Swissvale, whose work gave me the idea of keeping track of the sets in the first place. I should stress, though, that the set classifications are mine, so responsibility for any errors is entirely on me.

Star-divide

Basic Facts and Figures

* During the 19 games we ran a total of 1,190 offensive plays. For the effects of the analysis, I decided to leave out all spikes (8) and all kneels (19). There were 5 plays that I could not watch and therefore could not classify (I couldn't download two games via torrent and while I recorded them on the DVR, they were cut off). That still leaves us a sample size of 1,158 plays.

* During those 1,158 plays, we ran 547 times, passed 560 times and our QBs were sacked on 51 occasions. Counting the sacks as pass attempts, that gives us a 47 run / 53 pass split for the year.

* Total yardage gained was 6,489; we racked up 4,513 through the air and 2,307 on the ground. We lost 311 yards on the 51 sacks.

* We gained 5.6 yards per offensive play. This beaks down as 4.2 yards per attempt on the ground; 8.1 yards per pass attempt and 12.8 yards per completion.

Set Breakdown

My definitions of "Sets" are quite simple. I am merely interested in observing how many players lined up in a Receiving position ("R") and how many players we have in the Backfield ("B"). This classification depends upon the assignment of a player during a particular play, not his "normal position." What do I mean by that? If a certain play called for Hines Ward to line up in the backfield, for example, I counted him as a Backfield player. When Mewelde Moore lined up in the slot, I tallied him as a Receiver. So a play recorded as "R2/B2" means we had two players lined up as receivers and two in the backfield with the QB.

Using this system, it is possible to break down the 1,158 plays into 16 categories. Having said that, here's our first table.

SET Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

R0/B1 3 0 3 0 2 0 2 0

R0/B2 7 0 7 0 4 0 4 0

R0/B3 10 0 10 0 10 0 10 0

R1/B1 39 13 25 1 124 69 59 6

R1/B2 103 8 94 1 442 90 359 6

R1/B3 36 4 32 0 175 43 132 2

R2/B1 203 90 111 2 1.248 802 460 57

R2/B2 119 36 82 1 758 284 477 27

R2/B3 8 1 7 0 40 21 19 1

R3/B0 2 2 0 0 5 5 0 1

R3/B1 366 203 140 23 2.035 1.649 555 127

R3/B2 65 50 14 1 352 314 43 29

R4/B0 35 26 5 4 122 119 23 16

R4/B1 98 74 10 14 680 675 89 46

R4/B2 1 1 0 0 5 5 0 1

R5/B0 63 52 7 4 487 437 75 33

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* The first thing we notice is that the most common formation we used in 2010 was the R3/B1 set. That is, three receivers and one back. It was used 31.6% of the time. And when we were in this formation, our play split was 38 run / 62 pass.

* Next comes the R2/B1 set, which accounted for 17.5% of our plays. Here the split was 55 run / 45 pass.

* The set most likely to result in a sack was the R4/B1 (which we used on 8.5% of our plays), when our QBs got a taste of the turf 14.3% of the time.

* In overall terms, the "safest" sets for a QB were those with no receivers at all. We ran plays like this 20 times. They were all goal-line plays from the 1 or 2 yard line.

* The most "successful" set in terms of yards per play was the much-maligned R5/B0. Yes, that Empty backfield spread set. We picked up 7.7 yards per play. The lowest gaining set was the R0/B2, which averaged a mere 0.6 yards. (As we'll see later, it was the least successful of all the short-yardage sets in terms of outcome.)

Alas, but I don't have John Stephen's skills with HTML tables, so that first one, with 16 different categories, may generate more than one headache. So here's the same information, but summarized, and probably more manageable. First, by number of receivers:

SET GROUPS Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

R0 20 0 20 0 16 0 16 0

R1 178 25 151 2 741 202 550 14

R2 330 127 200 3 2.046 1.107 956 85

R3 433 255 154 24 2.392 1.968 598 157

R4 134 101 15 18 807 799 112 63

R5 63 52 7 4 487 437 75 33

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* Previously we saw that the R3/B1 was the single most utilized Set in our arsenal of 16. So it's probably not too surprising that the above table with the groups shows that overall the three receiver sets were the most common, and we used them for 37.4% of all plays (433 / 1,158).

* And the R3 sets ended in sacks 5.5% of the time. That made them considerably safer than the R4 sets, which ended badly for the QB 13.4% of the time.

* We passed 89% of the time out of the R4 and R5 sets... and most of the running there was done by our QBs when they were either forced to run for their lives or had an option and took off.

And now by Backfield sets... "Empty," of course, is when we don't have anyone back there with the QB:

SET GROUPS Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

Empty 100 80 12 8 614 561 98 50

B1 709 380 289 40 4.089 3.195 1.165 236

B2 295 95 197 3 1.561 693 883 63

B3 54 5 49 0 225 64 161 3

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* Now we can see that a solid majority of our plays (709 of 1,158, or 61.2%) featured only one back.

* We only used the Empty sets 8.6% of the time (100 plays) the entire year.

Outcome Analysis

Now let's throw outcomes into the mix. It's one thing to say that we used "Set X" so many times, but I'm interested in analyzing how successful we were in each one. For this post, we'll consider the following "outcomes":

* Positive outcomes: 1st downs; TDs; and "Big Gains," which are those that end in a gain of 20 yards or more. It is perhaps arbitrary that I have pegged Big Gains at +20 yards, but it seems to me that if you can advance one-fifth of the field on one play, you've done something pretty special.

* Negative outcomes: sacks, interceptions, fumbles and the times we got "stuffed." I remember Michael using the term "stuffed" to define a running play that resulted in zero or negative yards. I decided to expand this term to include pass completions that resulted in zero or negative yards, but NOT pass incompletion. An incomplete pass is "just a misfire," but a completed pass for negative yardage was a play that didn't work, be it due to our own problems with execution or because the defense anticipated what we were up to and adjusted accordingly.

* Overall, of our 1,158 plays, 29.1% ended in a first down, which includes the 4.0% that led to a TD as well as all by 2 of the Big Gain plays (yes, two times during the season picking up 20 yards was not good enough for a 1st down.)

* If we group the sacks, stuffs, interceptions and fumbles together, we can determine that 17.1% of our plays ended badly (i.e. had a negative outcome). Admittedly, we can agree that some plays end "more badly" than others: an interception is a worse outcome than getting stuffed for no gain on a 1st and 10. So some of you may object to me lumping them together to measure an overall negative outcome score for each offensive set. But I think that we can all agree that those are still undesirable ends to a particular play.

Once again, for the sake of simplicity, I won't present the information by all 16 categories. Let's stick with the groups:

SET GROUPS Total Plays 1st Downs Big Gains TDs Sacks Stuffed INTs Fumbles

R0 20 12 0 12 0 6 0 0

R1 178 43 6 4 2 34 1 2

R2 330 88 29 9 3 40 3 5

R3 433 118 32 9 24 42 0 3

R4 134 45 12 8 18 4 2 1

R5 63 31 6 4 4 0 3 1

TOTALS 1.158 337 85 46 51 126 9 12

* The sets most likely to result in positive outcomes were actually the R0s, but we only used them 20 times when we were on the 1 or 2 yard line. We punched it in for a TD 60% of the time. However, we also had negative outcomes 30% of the time, which is the highest of the five categories presented (from getting stuffed on 6 tries).

* The R5 Empty Backfield Set plays ended in a first down 49% of the time; 6.3% were TDs. This makes them the second most successful set category overall. In terms of the Big Gain, the R5 sets reigned supreme in percentage terms, at 9.5% occurrence.

* Were the R5s particularly prone to negative outcomes? Readers of my previous post on the subject won't be surprised by the answer: actually, they weren't. They ended badly only 12.7% of the time, which is the best showing of the five categories shown above. They were the most turnover prone, at 6.3%.

Here's the outcome data grouped by number of players lined up in the backfield:

SET GROUPS Total Plays 1st Downs Big Gains TDs Sacks Stuffed INTs Fumbles

Empty 100 39 6 6 8 1 3 2

B1 709 202 61 23 40 64 4 6

B2 295 77 14 8 3 48 1 4

B3 54 19 4 9 0 13 1 0

TOTALS 1.158 337 85 46 51 126 9 12

* The Empty sets were most likely to end in a first down (39% of the time), but also most likely to end in a sack (8.0%) or an interception (3.0%). However, since it was extremely rare for a play from these sets to end in a "stuff" or a fumble, they were still the least likely to end badly (14.0% overall).

* The B3 sets of course were used for "regular" short yardage (34 of 54) or goal-line situations (20 of 54). We picked up the 1st down only 35.2% of the time and had negative outcomes 25.9% of the time. This initially surprised me. Then I realized that while the R0/B3s on the goal line were remarkable successful (80% of those plays ended in a touchdown), the R1/B3 and R2/B3 sets had very high "stuff rates" which is why as a category the B3s were prone to negative outcomes.

* Interestingly enough, from this angle, the best formation at generating a "Big Gain" is one in which we have one player in the backfield. The B1 sets generated Big Gains 8.6% of the time; the Empty sets only 6.0%. This is consistent with the findings I presented in the Empty Backfield analysis previously.

Lord knows that it's possible to slice and dice the data in myriad ways, so I could go on for a long time. If this post gets a positive reception, I probably will present more findings. But I think I'll wrap it up now in order to avoid wearing out my welcome. But I hope that this analysis has provided a breakdown of 2010 offensive performance in a way that's a bit different from other takes on it.

http://www.behindthesteelcurtain.com/2011/3/28/2073805/breakdown-of-2010-offensive-performace-by-set

by Chileburger on Mar 28, 2011 4:00 PM EDT

Those of you who had the patience to work through my previous posts are aware of the fact that I put together an Excel data base on our offensive plays and I've tried to use it to analyze certain aspects of the Steelers 2010 performance.

Now that the season is on the books, I thought that some of you might be interested in an analysis that I performed of offensive performance by Set for the entire 19 game campaign. You'll find plenty of data and tables after the jump.

Before continuing, though, I'd like to give a tip of the hat to Swissvale, whose work gave me the idea of keeping track of the sets in the first place. I should stress, though, that the set classifications are mine, so responsibility for any errors is entirely on me.

Star-divide

Basic Facts and Figures

* During the 19 games we ran a total of 1,190 offensive plays. For the effects of the analysis, I decided to leave out all spikes (8) and all kneels (19). There were 5 plays that I could not watch and therefore could not classify (I couldn't download two games via torrent and while I recorded them on the DVR, they were cut off). That still leaves us a sample size of 1,158 plays.

* During those 1,158 plays, we ran 547 times, passed 560 times and our QBs were sacked on 51 occasions. Counting the sacks as pass attempts, that gives us a 47 run / 53 pass split for the year.

* Total yardage gained was 6,489; we racked up 4,513 through the air and 2,307 on the ground. We lost 311 yards on the 51 sacks.

* We gained 5.6 yards per offensive play. This beaks down as 4.2 yards per attempt on the ground; 8.1 yards per pass attempt and 12.8 yards per completion.

Set Breakdown

My definitions of "Sets" are quite simple. I am merely interested in observing how many players lined up in a Receiving position ("R") and how many players we have in the Backfield ("B"). This classification depends upon the assignment of a player during a particular play, not his "normal position." What do I mean by that? If a certain play called for Hines Ward to line up in the backfield, for example, I counted him as a Backfield player. When Mewelde Moore lined up in the slot, I tallied him as a Receiver. So a play recorded as "R2/B2" means we had two players lined up as receivers and two in the backfield with the QB.

Using this system, it is possible to break down the 1,158 plays into 16 categories. Having said that, here's our first table.

SET Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

R0/B1 3 0 3 0 2 0 2 0

R0/B2 7 0 7 0 4 0 4 0

R0/B3 10 0 10 0 10 0 10 0

R1/B1 39 13 25 1 124 69 59 6

R1/B2 103 8 94 1 442 90 359 6

R1/B3 36 4 32 0 175 43 132 2

R2/B1 203 90 111 2 1.248 802 460 57

R2/B2 119 36 82 1 758 284 477 27

R2/B3 8 1 7 0 40 21 19 1

R3/B0 2 2 0 0 5 5 0 1

R3/B1 366 203 140 23 2.035 1.649 555 127

R3/B2 65 50 14 1 352 314 43 29

R4/B0 35 26 5 4 122 119 23 16

R4/B1 98 74 10 14 680 675 89 46

R4/B2 1 1 0 0 5 5 0 1

R5/B0 63 52 7 4 487 437 75 33

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* The first thing we notice is that the most common formation we used in 2010 was the R3/B1 set. That is, three receivers and one back. It was used 31.6% of the time. And when we were in this formation, our play split was 38 run / 62 pass.

* Next comes the R2/B1 set, which accounted for 17.5% of our plays. Here the split was 55 run / 45 pass.

* The set most likely to result in a sack was the R4/B1 (which we used on 8.5% of our plays), when our QBs got a taste of the turf 14.3% of the time.

* In overall terms, the "safest" sets for a QB were those with no receivers at all. We ran plays like this 20 times. They were all goal-line plays from the 1 or 2 yard line.

* The most "successful" set in terms of yards per play was the much-maligned R5/B0. Yes, that Empty backfield spread set. We picked up 7.7 yards per play. The lowest gaining set was the R0/B2, which averaged a mere 0.6 yards. (As we'll see later, it was the least successful of all the short-yardage sets in terms of outcome.)

Alas, but I don't have John Stephen's skills with HTML tables, so that first one, with 16 different categories, may generate more than one headache. So here's the same information, but summarized, and probably more manageable. First, by number of receivers:

SET GROUPS Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

R0 20 0 20 0 16 0 16 0

R1 178 25 151 2 741 202 550 14

R2 330 127 200 3 2.046 1.107 956 85

R3 433 255 154 24 2.392 1.968 598 157

R4 134 101 15 18 807 799 112 63

R5 63 52 7 4 487 437 75 33

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* Previously we saw that the R3/B1 was the single most utilized Set in our arsenal of 16. So it's probably not too surprising that the above table with the groups shows that overall the three receiver sets were the most common, and we used them for 37.4% of all plays (433 / 1,158).

* And the R3 sets ended in sacks 5.5% of the time. That made them considerably safer than the R4 sets, which ended badly for the QB 13.4% of the time.

* We passed 89% of the time out of the R4 and R5 sets... and most of the running there was done by our QBs when they were either forced to run for their lives or had an option and took off.

And now by Backfield sets... "Empty," of course, is when we don't have anyone back there with the QB:

SET GROUPS Total Plays N° Passes N°

Runs N° Sacks Total Yards Pass Yards Run Yards Completions

Empty 100 80 12 8 614 561 98 50

B1 709 380 289 40 4.089 3.195 1.165 236

B2 295 95 197 3 1.561 693 883 63

B3 54 5 49 0 225 64 161 3

TOTALS 1.158 560 547 51 6.489 4.513 2.307 352

* Now we can see that a solid majority of our plays (709 of 1,158, or 61.2%) featured only one back.

* We only used the Empty sets 8.6% of the time (100 plays) the entire year.

Outcome Analysis

Now let's throw outcomes into the mix. It's one thing to say that we used "Set X" so many times, but I'm interested in analyzing how successful we were in each one. For this post, we'll consider the following "outcomes":

* Positive outcomes: 1st downs; TDs; and "Big Gains," which are those that end in a gain of 20 yards or more. It is perhaps arbitrary that I have pegged Big Gains at +20 yards, but it seems to me that if you can advance one-fifth of the field on one play, you've done something pretty special.

* Negative outcomes: sacks, interceptions, fumbles and the times we got "stuffed." I remember Michael using the term "stuffed" to define a running play that resulted in zero or negative yards. I decided to expand this term to include pass completions that resulted in zero or negative yards, but NOT pass incompletion. An incomplete pass is "just a misfire," but a completed pass for negative yardage was a play that didn't work, be it due to our own problems with execution or because the defense anticipated what we were up to and adjusted accordingly.

* Overall, of our 1,158 plays, 29.1% ended in a first down, which includes the 4.0% that led to a TD as well as all by 2 of the Big Gain plays (yes, two times during the season picking up 20 yards was not good enough for a 1st down.)

* If we group the sacks, stuffs, interceptions and fumbles together, we can determine that 17.1% of our plays ended badly (i.e. had a negative outcome). Admittedly, we can agree that some plays end "more badly" than others: an interception is a worse outcome than getting stuffed for no gain on a 1st and 10. So some of you may object to me lumping them together to measure an overall negative outcome score for each offensive set. But I think that we can all agree that those are still undesirable ends to a particular play.

Once again, for the sake of simplicity, I won't present the information by all 16 categories. Let's stick with the groups:

SET GROUPS Total Plays 1st Downs Big Gains TDs Sacks Stuffed INTs Fumbles

R0 20 12 0 12 0 6 0 0

R1 178 43 6 4 2 34 1 2

R2 330 88 29 9 3 40 3 5

R3 433 118 32 9 24 42 0 3

R4 134 45 12 8 18 4 2 1

R5 63 31 6 4 4 0 3 1

TOTALS 1.158 337 85 46 51 126 9 12

* The sets most likely to result in positive outcomes were actually the R0s, but we only used them 20 times when we were on the 1 or 2 yard line. We punched it in for a TD 60% of the time. However, we also had negative outcomes 30% of the time, which is the highest of the five categories presented (from getting stuffed on 6 tries).

* The R5 Empty Backfield Set plays ended in a first down 49% of the time; 6.3% were TDs. This makes them the second most successful set category overall. In terms of the Big Gain, the R5 sets reigned supreme in percentage terms, at 9.5% occurrence.

* Were the R5s particularly prone to negative outcomes? Readers of my previous post on the subject won't be surprised by the answer: actually, they weren't. They ended badly only 12.7% of the time, which is the best showing of the five categories shown above. They were the most turnover prone, at 6.3%.

Here's the outcome data grouped by number of players lined up in the backfield:

SET GROUPS Total Plays 1st Downs Big Gains TDs Sacks Stuffed INTs Fumbles

Empty 100 39 6 6 8 1 3 2

B1 709 202 61 23 40 64 4 6

B2 295 77 14 8 3 48 1 4

B3 54 19 4 9 0 13 1 0

TOTALS 1.158 337 85 46 51 126 9 12

* The Empty sets were most likely to end in a first down (39% of the time), but also most likely to end in a sack (8.0%) or an interception (3.0%). However, since it was extremely rare for a play from these sets to end in a "stuff" or a fumble, they were still the least likely to end badly (14.0% overall).

* The B3 sets of course were used for "regular" short yardage (34 of 54) or goal-line situations (20 of 54). We picked up the 1st down only 35.2% of the time and had negative outcomes 25.9% of the time. This initially surprised me. Then I realized that while the R0/B3s on the goal line were remarkable successful (80% of those plays ended in a touchdown), the R1/B3 and R2/B3 sets had very high "stuff rates" which is why as a category the B3s were prone to negative outcomes.

* Interestingly enough, from this angle, the best formation at generating a "Big Gain" is one in which we have one player in the backfield. The B1 sets generated Big Gains 8.6% of the time; the Empty sets only 6.0%. This is consistent with the findings I presented in the Empty Backfield analysis previously.

Lord knows that it's possible to slice and dice the data in myriad ways, so I could go on for a long time. If this post gets a positive reception, I probably will present more findings. But I think I'll wrap it up now in order to avoid wearing out my welcome. But I hope that this analysis has provided a breakdown of 2010 offensive performance in a way that's a bit different from other takes on it.