Code and Football

I’ve been reading a ton of books. One of these is Robert W. Peterson’s “Pigskin”, which has been an interesting read so far. I’m roughly in the late 1940s in this book, which starts with the beginning of professional football and ends with the NFL championship in 1958. What has caught my eye are Mr. Peterson’s comments about the spread of the T formation in the 1940s. He describes the Bears 73-0 NFL Championship victory over the Redskins. Later, when describing the switch of the Redskins to the T in 1944, he gives this accounting of the state of the football world in 1944: (1)

By that year, more than 50 percent of college teams has converted to the T formation. So had most pro teams. Henceforth, the old single-wing formula of “three yards and a cloud of dust” as the ideal offensive play would go the way of the rugby ball in pro football

The adoption was not immediate upon the end of the 1940 season, however, and teams, coaches, and whole conferences that were successful with the single wing (or Southwestern spread) tended to stick with it. For example, in Tom Landry’s autobiography, he notes that Texas made the switch in 1947, after Dana Bible retired.(2) Y. A. Tittle’s memory of the conversion is (3)

If I remember correctly, the first Southwestern conference team to switch to the T formation from the single- and double-wing formations was Rice University, followed by Georgia and Louisiana State.

The quote above mixes the SEC and the Southwest conference, but still.. LSU switched in 1945. I’m just not sure which of the 50% of college football teams were converting. Army and Notre Dame are well known early adopters, but as a counterexample, in 1947, Fritz Crisler won a national championship with a single wing offense at Michigan.

Dan Daly, when discussing the effects of the 73-0 Bears win over the Redskins, noted:(4)

Only one other NFL team, the Philadelphia Eagles, switched to the T the next season. And as late as 1944, both clubs that played in the championship game, the Green Bay Packers and the New York Giants, used the single wing or some variation.

Paul Brown, the head coach of Ohio State from 1941 to 1943, was the first coach to see Don Faurot’s split T in action, in his very first game as Ohio State’s head coach, but then says of his game with Clark Shaughnessy’s Pittsburgh squad in 1943 (5)

It was my first real look at the T formation with flankers and men in motion, however, and it was the kind of football I later assimilated into my own system with the Browns.

So from 1941 to 1943, the “Bears” T was largely unknown in the Big 10. Paul Brown then learned the T while serving in the armed services. In 1946 and 1947, in the first two AAFC championships, Brown’s T was pitted against the single wing offense of the New York Yankees.(6)

As Dan Daly notes, the lack of players trained in the new offense slowed the T formation’s spread.(7)

In the early ’40s, the Bears and the Eagles – the only two T-formation teams – drafted an unusual number of Shaughnessy’s Stanford players because the Cardinal were the lone major college team using the offense.

Dan Daly later writes (8)

By the end of the decade, though, five out of seven college teams played some form of the T. Suddenly it was the single-wing Steelers who were having trouble finding players to fit their system.

And it does make sense. There were some early adopters who ran into Luckman, or Shaughnessy, or former Bears quarterbacks and coaches, but a lot of coaches learned the T while serving in the armed services during the war, coaching or playing in service teams. So it wasn’t the early 1940s when the transition occurred, as far as I can tell. Instead, it was the mid to late 1940s when the T became dominant. The conversion was not “immediate”. It took 3-4 years to gain steam, and a decade for it to dominate.

Notes

There were only ten pro teams in 1944, and it’s entirely possible that most NFL teams were running a T by 1944 (By my count, Chicago, Philadelphia, Washington, and Cleveland are using the T by 1944. Green Bay and New York are not. The other four – Brooklyn, Boston, Detroit, and Card-Pitt – I’m not sure of). Green Bay switches to the T in 1947, New York in 1949.

Army’s first use of the T is in the 1941 Army-Navy game.(9) Notre Dame had Halas’s players assist with the conversion in 1942. Clark Shaughnessy coaches Maryland in 1942 and then Pittsburgh in 1943.

1944 is an unusual year to use as a baseline, because so many coaches and players were in the armed services. That may in fact have aided the transition, as so many coaches with a traditional single wing background found themselves coaching alongside experts in the T on service teams.

For those who have never read Ron Fimrite’s article in Sports Illustrated about the Stanford Indians’ 1940 season, just do it. It’s one of the great short articles on football. The link is given in the bibliography.

References

1. Peterson, Chapter 8.

2. Landry and Lewis, p. 74.

3. Tittle, Chapter 5.

4. Daly, Chapter 3.

5. Brown and Clary, p. 101.

6. Brown and Clary, pp. 181-182.

7. Daly, Chapter 3.

8. Daly, Chapter 3.

9. Roberts, Chapter 2.

Bibliography

Brown, Paul, and Clary, Jack, PB: The Paul Brown Story, Atheneum 1980.

Daly, Dan, The National Forgotten League: Entertaining Stories and Observations from Pro Football’s First Fifty Years, University of Nebraska Press, 2012. [ebook]

Fimrite, Ron, “The Melding of All Men, Suited to a T”, September 5, 1977. “Sports Illustrated”. retrieved July 28, 2013.

Holland, Gerald, “The Man Who Changed Football”, February 3, 1964. Sports Illustrated. retrieved July 28, 2013.

Johnston, James W. ,The Wow Boys: A Coach, a Team, and a Turning Point in College Football , University of Nebraska Press, 2006.

Landry, Tom, and Lewis, Gregg,Tom Landry: An Autobiography, Harper Paperbacks, 1990.

McGarr, Elizabeth, “The Top 20 Greatest Moments”, August 20, 2008. “Sports Illustrated”. retrieved July 28, 2013.

Peterson, Robert W., Pigskin: The Early Years of Pro Football, 1997. [ebook]

Roberts, Randy, A Team for America: The Army-Navy Game That Rallied a Nation at War , Houghton Mifflin Harcourt, reprint ed 2011. [ebook]

Tittle, Y. A, and Clark, Kristine S.,Nothing Comes Easy: My Life in Football ,Triumph Books, 2009. [ebook]

Zimmerman, Paul, in “Letters”, December 22, 1997. “Sports Illustrated”. retrieved July 28, 2013.

In a previous post, we noted that the “two gap” 3-4 doesn’t extend back to the origin of the 5-2 Oklahoma, as Jones and Wilkinson, 1957, didn’t coach a pure two gap system. It was a hybrid 1 gap 2 gap system, with only the nose guard two gapping.

Looking for post-Wilkinson 5-2 Oklahoma philosophies is difficult. The article by Norris and Harper in Defensive Football Strategies (American Football Coaches Association), dated 1974, already shows the influence of Jimmy Johnson’s “upfield pressure” philosophies. Linemen 1 gap to the play side and purse from the offside. Jimmy Johnson was the Oklahoma defensive line coach from 1970 to 1972. Fogie Fazio’s article on the 50, dated 1980, in the same book, has clear 1 gap responsibilities.

Since we know Parcells coached a two gap 3-4, that’s our starting point. We’ll consider Parcell’s career, using Bill Gutman’s biography as a reference.

The first mention of a defense is on page 36, where they discuss the 52 Invert Defense. Quoting Tom Godfrey

It was a defense that dictated to the offense, not a traditional sit-and-read defense. When the ball was snapped we were moving, and that caught people off guard. All five linemen were moving one way or the other, and the secondary moved opposite to them.

Bill Parcells was an assistant at Wichita State at the time, and he taught this defense to his old high school. But confusing, slanting, pressure 5-2s aren’t read-and-react two gap 3-4s. This isn’t the defense he took to the Giants.

His next stop is at Army, where his old high school coach had assembled a terrific staff, and where he meets Bobby Knight. From there, he goes to Florida State, where he runs into Steve Sloan, whom he follows for some time. There is a stay at Vanderbilt, and then three years at Texas Tech (1975-1977). In discussing the middle of this period, Steve Sloan says

In 1976 he became more creative. He went to an even front, something not many college teams were doing then. He ran a lot of slants and gave the offense different looks.

Needless to say, the 3-4 isn’t an even front.

Parcells then spends a year at the Air Force academy as their head coach, spends a brief period as a linebacker coach with the Giants, gets out of football briefly, and then in 1980 joins Ron Erhardt with the Patriots, where Hank Bullough was the defensive coordinator, and one of the architects of the conversion of the Pats to the 3-4. This moment seems seminal to me.

As Bullough says of the Pats introduction to the professional 3-4,

We were the first. We had gone through a tough season in ’73 and our defensive line wasn’t very good. We had drafted Steve Nelson and Sam Hunt and they were two good-looking kids at linebacker, and I said to Chuck, “Let’s go to the 3-4,” and that’s what we did.

I believe the trail of evidence now moves from Bill himself to something known by the awkward phrase Fairbanks-Bullough 3-4 defensive system, a kind of gobbledegook that makes “arm talent” seem svelte by comparison. Again, the questions that come to mind are: did it start out as a pure two gap 3-4 or was it a hybrid 1-gap 2-gap system, like Bud Wilkinson’s? If the latter is true, when did it evolve into a two gap 3-4? Could there have been prior art?

If we just Google’s ngram viewer to investigate, we see that the phrase “3-4 defense” first appears in 1970.

The phrase “3-4 defense” enters the corpus of books that Google has scanned in 1970, and then steadily gains usage.

There isn’t a lot of time to develop prior 3-4 art, if the ngram is correct. And for all we know, the ngram is initially tracking the discussion of three man lines in prevent defenses.

Bibliography

Cardofo, Nick, “Recurring Scheme“, September 5, 2003, Boston Globe, retrieved July 14, 2013.

Gutman, Bill. “Parcells: A Biography“, Carroll and Graf, 2000.

Bill Parcells has an enormous hold on the  hearts and souls of football fans, ranked, for example, in this poll as the sixth best coach of all time. People take his declarative statements as edicts. Parcellisms, made more accessible to the  masses, comprise a substantial part of Pat Kirwin’s book,  “Take Your Eye Off the Ball”.  And  one of the notions of that’s beginning to take hold is that Parcell’s theories of  how the 3-4 should  be played are getting etched in stone as the way it always was played. Without proving it, people are labelling the two gap 3-4s as the “true” 3-4. Other 3-4s are somehow, “not true”.

The origin of the 3-4 is well known. It’s a 5-2 Oklahoma where the defensive ends can stand up and have pass responsibilities. The differences between the two are merely semantic. So if you want to know what the true 3-4  is, you need  to know what the  true 5-2 is, and  the best place for that is the 1957 text of Gomer Jones and Bud Wilkinson, labelled  “Modern Defensive Football.”

In  it they describe some amazing defensive concepts. They are excellent teachers, and I’ve never seen anyone explain force and contain concepts as well as they do. They do it for four man backfields and three man backfields (what they called the four and five spoke contain units). And they also describe the 5-2, which they call defense 72.

Jones and Wilkinson’s Defense 72. Players are named as if they are playing a 6-2 defense, which was the recommended front for a single wing opponent. Note the hybrid defense end/cornerback in this scheme.

Some things to  note about these older defenses. Linebackers are much closer to the line  of scrimmage, pretty tight in fact. This can more easily be seen in old newsreel footage. If you can find, say, video of early 1950s Oklahoma and Notre Dame, linebackers are often within a couple yards of the line of scrimmage. An Oklahoma from this period looks something like this.

Now, what about the gap assignments? We’ll note that the modern notion of gap control is a relatively recent phenomenon. I explored this in a letter exchange with Coach Hugh Wyatt. He would date the phrase to about 1979, with Monte Kiffen the first known user of the term. The phrase “gap responsibility” is used by late 1960s, so notions of gap management in the late 1950s are stated in terms of  things  the offense must never do.

And other  than the nose guard, what must never be done is the opponent must never block you in. This makes it clear: of the 7 line assignments, 6 are one gap assignments. The original 5-2 is largely a one gap defense.

It is interesting to look at the responsibilities of the nose guard, the only one with responsibilities on both side of his opponent. As Jones and Wilkinson say (note that since they labeled all the players as if  they were in a 6-2 front, the nose guard is called the right guard in this scheme):

Right guard:  Line up  head up with the offensive center about 2 to 2 1/2 feet off the line of scrimmage. Vary the strength of your charge from play to play. Occasionally, charge hard into the center and attempt to knock him back. Most of the time, charge with enough force to control the offensive center. Basically, you must never allow the center to cut you either way. You must control both sides of the center, maintaining your ability to move to either side.

Not only is the original  5-2 largely a one gap defense, the center is flexed. This is a far cry from the modern version of 300 to 340  pound behemoths maintaining gap control across the whole front. In other words, the two gap 3-4 was an evolution from the original 3-4, and further, the one gap 3-4s are more in tune with the  original 5-2 than is the “true” 3-4.

So when did the two gap 3-4 evolve? Now, I’m curious. It’s documented, for example, that Parcell’s notion of outside run contain isn’t the same as older notions. Older contain strategies (such as the one in Jones and Wilkinson) tend to  keep runners inside the contain unit. Parcells was content. however, to run the sweep or pitch out of bounds.

Did Parcells invent it? Or was it already common in colleges by the time he arrived in the NFL? Going from one one gapper to three isn’t that much of a reach, especially if your team has three large powerful linemen. But glancing at a couple articles in AFCA’s “Defensive Football Strategies”, articles on the 5-2  originally published in 1974 and 1980,  you see one gap responsibilities being taught, even for the nose guard.

I’m not an old coach, who would have this information buried in his bones. But there is more to college football defense than a two gapping 5-2, and the paths from the past to modern times more complex than many realize.

I’ve been getting some decent feedback from the pass defense images I’ve made, so I’ve decided to extend this series for now.

Cover Zero and Man Free

In Cover Zero, all the defensive backs have assignments, and so there is no “free” safety. This is good for blitzes, but can be weak if your defensive backfield lacks the ability to cover for any length of time. In this image, the stippled lines represent an assigned ‘man’.

Cover Zero, Tampa Under front, ace backfield.

The coverage “man free” or “one free” is a defense where the free safety is a free agent, able to defend or double cover or safety blitz, as the need arises.

man free, Miami 43 over front.

Cover  1

Cover 1 keeps the free safety back in a deep zone. Otherwise, coverage beneath is man to man, or perhaps a mix of man and zone.

Miami 43, shade front, man plus cover 1 by the free safety.

Cover 2 and Tampa 2

Back in the day, Cover 2 was also called the double zone, because both outside receivers had a form of double coverage. There are references that claim the 1963 Chicago Bears played a form of a double zone and confused the heck out of people. This is significant because most folks were only playing rotating zones, if that (see the Cover 3 section).

Cover 2 from Miami 4-3 over front. Cornerbacks jam then fall into zones.

Cover 2 is famous for having a hole in the middle. But if you have a  fast, agile middle linebacker, as the 1970s Pittsburgh Steelers did, then you can have him race down the middle and split the deep zone into three, forming  what is known now as the Tampa 2 defense.

Tampa under front, Tampa 2 zone defense. Modeled on the diagram in Matt Bowen’s Tampa 2 article.

Cover 3: rotating zones versus the modern Cover 3.

In the early 1960s, when you said zone coverage, by default you meant  this, and only  this:

Tom Landry’s 4-3 Inside, showing a 1960s era strong side rotating zone. Strong side linebacker and left cornerback jam before falling into zone.

This kind of defense was abused in Super Bowl III, where Baltimore’s rotating zone became a sitting duck for a still  mobile Joe Namath. By the 1970s, usage of this defense fell away, as it was too easy to diagnose.

The Cover 3 we will show here comes from a Stack 44 setup, achieved when a 4-3 Stack (Miami 4-3) overshifts the secondary. Some people call the defensive back at linebacker depth a monster or rover, and these kinds of defenses, with three defensive backs at backfield depth, naturally lend  themselves to Cover 3, with three deep backs.

Cover 3 from a Stack 4-3 monster.

QQH coverage

This is quarters-quarters-half coverage, what some folks call Cover 6 (Cause Cover 2 plus Cover 4 equals Cover 6). Note how it changes the complexion of the related Cover 3 from above.

Quarter quarter half coverage from a Stack 4-3 Monster.

Cover 4

Also called quarters coverage. This is a kind of prevent defense.

Quarters coverage, from a Tampa 4-3 under front.

There are two well known adjusted yards per attempt formulas, which easily reduce to simple scoring models. The first is the equation  introduced by Carroll et al. in “The Hidden Game of Football“, which they called the  New Passer Rating.

(1) AYA = (YDs + 10*TDs- 45*INTs)/ ATTEMPTS

And the Pro Football Reference formula currently in use.

(2) AYA  = (YDs +20*TDs – 45*INTs)/ATTEMPTS.

Scoring model corresponding to the THGF New Passer Rating, with opposition curve also plotted. Difference between curves is the turnover value, 4 points.

Formula (1) fits well to a scoring model with the following attributes:

• The value at the 0 yard line is -2 points, corresponding to scoring a safety.
• The slope of the line is 0.08 points per yard.
• At 100 yards, the value of the curve is 6 points.
•  The value of a touchdown in this model is 6.8 points.

The difference, 0.8 points, translated by the slope of the line,  (i.e 0.8/0.08) is equivalent to 10 yards. 4 points, the value of a turnover, is equal to 50 yards. 45 was selected to approximate a 5 yard runback, presumably.

Pro Football Reference AYA formula translated into a scoring model. Difference in team and opposition curves, the turnover value, equals 3.5 points.

Formula (2) fits well to a scoring model with the following attributes:

• The value at the 0 yard line is -2 points, corresponding to scoring a safety.
• The slope of the line is 0.075 points per yard.
• At 100 yards, the value of the curve is 5.5 points.
• The value of a touchdown in this model is 7.0 points.

The difference, 1.5 points, translated by the slope of the line,  (i.e 1.5/0.075) is equivalent to 20 yards. 3.5 points, the value of a turnover, is equal to 46.67 yards. 45 remains in the INT term for reasons of tradition, and the simple fact this kind of interpretation of the formulas wasn’t available when Pro Football Reference introduced their new formula. Otherwise, they might have preferred 40.

Adjusted yards per attempt or adjusted expected points per attempt?

Because these models show a clearly evident relationship between yards and points, you can calculate expected points from these kinds of formulas. The conversion factor is the slope of the line. If, for example, I wanted to find out how many expected point Robert Griffin III would generate in 30 passes, that’s pretty easy, using the Pro Football Reference values of AYA. RG3’s AYA is 8.6, and 0.075 x 30  = 2.25. So, if the Skins can get RG3 to pass 30 times, against a league average defense, he should generate 19.35 points of offense. Matt Ryan, with his 7.7 AYA, would  be expected to generate 17.33 points of offense in 30 passes. Tony Romo? His 7.6 AYA corresponds to  17.1 expected  points per 30 passes.

Peyton  Manning, in his best  year, 2004, with a 10.2 AYA, could have been expected to generate 22.95 points per 30 passes.

This simple relationship is one reason why, even if you’re happy with the correlation between the NFL passer rating and winning  (which is real but isn’t all that great), that  you should sometimes consider thinking in terms of AYA.

A Probabilistic Rule of Thumb.

If you think about these scoring models in a simplified way, where there are only two results, either a TD or a non-scoring result, an interesting rule of thumb emerges. The TD term in equation (1) is equal to 10 yards, or 0.8 points. 0.8/6.8 x 100 = 11.76%, suggesting that the odds of *not* scoring, in formula (1), is about 10%. Likewise, for equation (2) whose TD term is 20, 1.5/7 x 100 = 21.43%, suggesting the odds of *not* scoring, in formula (2), is about 20%.

This is a follow up on my three part article, “Drawing a good diagram of a football field“. After trying for some time to automate arrow drawing, I’ve come to the conclusion that using GIMP, an arrow drawing plug-in, and GIMP’s path tool (a way to draw both straight lines and curves) are adequate to handle this problem.

Needed:

1. Diagram Drawing Tools (see my three part series here, here, and here).
2. A copy of GIMP. It’s free and available on Windows, Mac, and Linux.
3. Some experience with layers on GIMP. Start by searching “How to use layers in GIMP”. There are some nice Youtube videos that can get you started here.
4. This article is also really good: “How to Draw Arrows in Gimp”. Get the arrow plugin, install it, and read the instructions.
5. Look further at GIMP pathing with the search: “using path tool in gimp”.
6. For dashed and dotted lines, search using this phrase: “dashed lines in gimp”.

Some notes about drawing diagrams. If you’re using Windows shell to draw, then you’ll have no issues with Image::Magick. If you want to use the Perl interface to draw, then switch to Graphics::Magick, a fork of Image::Magick. Image::Magick has bugs when used with Perl.

When drawing arrows on a 640×480 diagram scaled in the way I’ve been scaling them, a length of wings setting of 15 and a brush width of 2 works well. This can be paired with path lengths of 5.0 pixels and will do nicely.

Tampa under front, Tampa 2 zone defense. Modeled on the diagram in Matt Bowen’s Tampa 2 article.

Tom Landry’s 4-3 Inside, showing a 1960s era strong side rotating zone, an early Cover 3. SAM and left cornerback jam before falling into zone.

Drawing a zone drop.

Load your diagram. Add a new transparent layer. Make sure you’re drawing on the transparent layer. Use the rectangle select tool (letter “R”). Choose a region to highlight as a zone. At this point select the bucket fill tool (shift B) and then select the color for the bucket fill. Use a foreground fill and set the opacity to about 50%. If your field is light green, use a dark green to fill the zone.

If you have more than one zone to add, add them all now.

Once complete, select all (important) and add another layer. On this new layer, add a path from the middle of the zone to the player. Select Tools -> Arrow (you added the arrow plugin, didn’t you?). Adjust arrow length of wings and brush width and draw the arrow. Repeat as needed.

If the path isn’t straight or you need a bar to create a “jam”, just use the path tool as needed.

If you want to add text or label the diagram, I’d suggest adding another transparent layer and putting the text above the main, zones, and arrows.

Once done, save the file as a GIMP native file and then again as a JPEG file. The second save will cause an export, crunching all the layers down to one image.

Miami 43, shade front, man plus cover 1 by the free safety.

Drawing man to man coverage

The trick here is to use the path tool to make a stippled (dotted) image. Choose a path from the defender to the player to be defended. When you choose stroke path, choose a line type (there are many). I like the stippled pattern, as it’s unlikely to be mistaken for a solid line.