27 May 2011
by Mike Tanier
It's third-and-2. Your team breaks the huddle and lines up in an empty backfield formation. You groan. Way to take away any threat of a run, you think. Way to limit your pass protection options. In the name of getting one more receiver off the line quickly, the team has telegraphed its intentions. What a horrible trade-off.
That's what I think, at least. I have written in the past about my frustration with empty backfield alignments on third-and-short. Down-and-distance situations like third-and-2 or third-and-3 may now be passing downs, but that empty backfield set takes running plays completely out of the equation unless Michael Vick or Tim Tebow is on the field. And with those guys in shotgun, the quarterback draw isn't going to fool anyone.
But is emptying the backfield on third-and-short really a bad strategy? Since I have the Football Outsiders' Game Charting Database at my fingertips and little else to do, it's a small matter to find out. Our Game Charters mark the number of players in the backfield (besides the quarterback) on every play, making it easy to filter out empty backfield formations. From there, it's easy to sort out all third-and fourth-down situations with four or fewer yards to go. Beyond four yards, the offense is in an obvious enough passing situation that the threat of a run won't affect the defense much anyway.
I found 184 empty backfield plays in short yardage situations, which I will call EBSY plays from now on. That's fewer than six per team; while we notice the EBSY strategy when it happens, it isn't taking the league by storm. Of those plays, 156 were pass attempts, 13 were sacks or scrambles, and the others were runs of various types, including direct snap plays to guys like Michael Robinson and Dexter McCluster. The offense converted a first down or scored a touchdown 67 times, 54 times through the air and 13 times running or scrambling. The "or touchdown" caveat is significant, because some teams use the empty backfield set frequently in red zone situations, particularly the two teams in the Super Bowl.
The first-down percentage on EBSY plays is 36.4 percent. Here's a breakdown of how that compares with normal short yardage situations, where a team keeps a back or two in the backfield:
|Normal Short-Yardage Situations|
|Plays||First Downs||First Down %||Yards||Yards Per Play|
The asterisk row removes rushes on third-and-1 and fourth-and-1. When sifting the data, I noticed that empty backfield plays with one yard to go are uncommon, so I was comparing one set of data with a lot of third-and-4 situations with another set loaded with extreme short-yardage situations. There are a lot of ways to adjust for that problem, but I opted for a hardheaded one. With ultra-obvious rushing situations out of the mix, the two data sets look much more alike in terms of down-and-distance composition. I could have done something more technical, sorting the data by field zone and breaking everything down by yards-to-go, but that 184-item data set isn't very big, and I didn't want to run the risk of measuring the instruments.
The chart shows that first-down percentages are lower for EBSY plays than regular short-yardage plays, even when the one-yard plunges are removed. From a yards-per-play standpoint, having backs in the backfield is better than going empty once you take out those plunges.
There are all sorts of gum in this data, from sneaks to scrambles to sacks. To get a clearer picture, let's look at the raw passing numbers:
|Raw Passing Numbers, Third Down, 2-4 Yards To Go|
|Attempts||Completions||Completion %||Yards||Yards Per Attempt||Sacks||Sack %|
There were 12 touchdown passes thrown in EBSY situations, 114 thrown with backs in the backfield; I didn't put that in the chart because the number of goal-to-go plays in the data skews the results. Losing 10 points of completion percentage and 0.8 yards per throw is enough to scare me away from using a particular tactic. It's interesting that sack rates go down in EBSY situations. That probably has to do with play design: Teams opt for quick throws when they need short yardage from an empty backfield, whereas there are a lot of slow-developing play-action passes in the other set. Still, the data contradicts one of my fears -- pass protection does not suffer much from EBSY tactics.
Some teams never go EBSY. I couldn't find any Titans or Raiders plays, and I only found one Jaguars play, which was a pass on third-and-four. (The Jaguars almost never empty their backfield. If you had their roster, you wouldn't let Maurice Jones-Drew leave the field, either.) The Rams ran one quarterback sneak at the goal line from an empty backfield; again, there's no good reason to take Steven Jackson out of the game when you only need a yard or two. Here are some other notes from my EBSY study:
Most of the successful EBSY plays came from a handful of teams, like the Packers and Saints. Obviously, if your wide receivers are great and your backs are so-so, and it's third down and a long two, there's some wisdom to emptying the backfield. Conversely, if you are the Falcons, with one good wide receiver and a hammer of a running back, going EBSY is counter-intuitive. It's not a good strategy to attempt three or four times in a season, because most of the teams that did so were 1-for-3 or 1-for-4 on conversions. It's also not a good way to isolate Vick or a Wildcat guy on a draw, because there isn't much evidence of defenses getting fooled.
That was fun. In a few weeks, I will isolate another down-and-distance tactic, the second-and-short play action bomb. How common is it? Does it work? And do you have any other suggestions? Leave them in the comment thread!
If you follow me on Twitter @FO_MTanier, you know I spent last week counting down the Top 100 Players Off the Top of My Head.
Some readers assumed that my list was a parody or satire of some other list, like the NFL Network's Top 100. Really, it was purely spontaneous, a culling of players who popped into my mind while I researched for Football Outsiders Almanac and some ESPN articles. Ranking players across various positions and teams, based on arbitrary criteria, is more of a move for attention than analysis. I'm not above attention getting moves, particularly during a lockout; I have books to sell and readership to attract. The Top 100 kept me out of trouble last week.
If you missed the list, here are some highlights. Will I do more Twitter countdowns? Who knows? But you have to follow to find out!
100. Michael Jenkins: The most ordinary player I have ever seen. Like a randomly generated Madden avatar, but in real life.
97. Jairus Byrd. Poor kid plays so deep that only Nietzsche can scout him.
95. Cliff Avril. If the Lions D-line were a boy band, he would be the shy boy-next-door type.
90. Alex Barr ... oops, false start! Rodger Saffold.
82. Chris Kluwe. S 14 I 12 W 9 D 15 C 10 Ch 14. AC -1 (pads, helmet), Chaotic Neutral, Cleats of Insidethetwentydom.
73. Lousaka Polite. The Dolphins offense is an ice cream truck trying to climb a steep hill in February. Polite is the soft-serve machine.
68. Alphonso Smith. Getting traded for nothing by Josh McDaniels is like getting a hickie from Kenickie.
48. Mike Tolbert. Imagine Pete Johnson and Andra Franklin had a baby and ... Wait, don't picture that.
33. Tom Brady. I was just being silly with this selection, because no one in their right mind would rank Brady 33rd on any list of active players.
28. Chris Johnson. New Titans playbook is just a cartoon flipbook of him running across the bottom of the page.
19. Some guy on the Packers squad we never heard of who will be starting by Week 9 and make two big plays in the 2012 playoffs.
The Rams had the last Walkthrough all to themselves. Now it's time to look at the other NFC West team with a big debate at the top of its list: the San Francisco 49ers.
1. Joe Montana. The "black ink test" isn't very kind to Montana. He led the league in completion percentage a few times and efficiency rating twice. He led the league in touchdowns in two strike-shortened years, and his 1987 total is a little shady because he played in two replacement games, throwing five touchdown passes. If you didn't know Montana, and you just took a cursory look at his statistical profile, you would probably see a near-Hall of Famer, not arguably the best quarterback ever.
Montana had a habit of finishing second in statistical categories. Dan Marino had a lot to do with that, as did Montana's habit of missing a game or two with injuries, or to rest for the playoffs, or both. In 1984, when Montana threw for 3,630 yards, 28 touchdowns and 10 picks, Marino topped him in yards and touchdowns, while Tony Eason beat him in interception percentage and Steve Bartkowski topped his 64.6 completion percentage in an 11-start season. Montana took the end of the 1989 season off to rest injuries before the playoffs, so his 26 touchdowns finished fourth in the NFL. He threw just 10 interceptions, but was bested in interception percentage by Jim Miller. There was always a Marino-type trumping him in volume stats, and some short-season or short-career guy nudging him in percentages.
Montana was 14-5 in the postseason with the Niners, and of course no one judges him based solely on black ink or statistical totals. It's often a good idea to look at a familiar player's record with fresh eyes.
2. Steve Young. Some of you would probably rank Young over Montana, particularly if you are a) a younger fan with very dim memories of Montana or b) someone who is very suspicious of "winner sauce" or "most wins" arguments. It is a tough call, and I admit that if I could have either quarterback in his prime to lead me through a 16-game season, I want Young. I want his arm, legs, mind and career-peak durability. Based on what they "could" do, Young is better. Based on what they did, Montana wins hands down. I leave the discussion to the floor.
3. John Brodie. Years ago, I read a brief article about Brodie's statistics by a respected football historian. The historian was influential and was one of my influences in this field, but he had a habit of writing condescending anti-stat straw man arguments late in his career. I cannot find the article now, but its tone went something like this: "Imagine a great quarterback who only completed 55 percent of his passes, who only threw for 3,000 yards twice, and who threw more interceptions than touchdowns for his career. Why, those numbers would make the silly bean counters crazy, but there really was such a player! His name was John Brodie, and your puny statistics could never measure his greatness."
Maybe the tone wasn't quite so mocking, but one theme of the article suggested that statisticians cannot account for changes in era, expectations, rules, season length, or anything else. Brodie, ironically, is a terrible quarterback to make this kind of argument for, because his stats leap off the screen. A 30-touchdown season is still impressive in any context; anyone who puts forth the effort to thumb through an encyclopedia to 1965 knows they are looking at something special in Brodie. Brodie wasn't a great winner who put up old-timey stats. He was a .500 quarterback who put up modern looking stats.
Brodie was not a Hall of Fame caliber player. He was comparable to Sonny Jurgensen in many ways, but he did not put up Jurgensen's numbers. His Pro Football Reference comps start with Norm Snead and John Hadl, and while he was far better than Snead, he definitely belongs at that wedding table. Remember the paradox of football in that era: NFL teams rarely threw when winning, so passing yardage/touchdown leaders often came from middling teams forced to play from behind. When looking at a yardage or touchdown leader from about 1960 through 1977, you are very often looking at a second-tier quarterback, not one of the best of the best. That's the kind of useful observation we make when using stats as evidence, not writing them off as child's play.
4. Y.A. Tittle. Tittle played with the Million Dollar Backfield of Joe Perry, John Henry Johnson, and Hugh McElhenny. Johnson was only in San Francisco for three years, one of them a good one, but the other two players were Niners Hall of Famers all the way. Perry was one of the best backs of his generation; McElhenny was a shake 'n' bake back. It was a T-formation offense, even after Johnson left (replaced by the very good J.D. Smith), so Tittle spent a lot of time faking handoffs to one all-time great and pitching the ball to another. The whole was less than the sum of its parts, and the Niners did some amazing things, like go 4-8 and finish 10th out of 12 teams in points with four future Hall of Famers in the backfield.
Tittle's Hall of Fame credentials rest almost completely with his last few Giants seasons. Take those years away from him, with the championship near misses and the grizzled ole veteran photographs, and he's a forgotten character.
5 Jeff Garcia. Garcia's best statistical year came for a 6-10 team with Jerry Rice and Terrell Owens as his receivers. He was excellent in 2001, very good in 2002, and turned out the lights on the Niners long era of greatness in 2003. He then entered a peripatetic phase where he earned undue praise for everything he did, earning a near Flutie-like reverence from the press. Garcia left a few of his best years in Canada, and it's not his fault that San Francisco was hell-bent on self-destruction by the time he got his feet wet. He earned entry onto two of my franchise lists (the Bucs as well), but I always thought he courted the press too much and lobbied too heavily for starting jobs, and that was before he came to Philly and was heralded as the anti-McNabb.
Frankie Albert was an incredible quarterback, runner, and punter in the 1940s. He threw 56 touchdowns in one two-year stretch, and it was tempting to list him above Garcia. But Albert's best years were in the AAFC. The Browns and Niners were the only AAFC teams that were truly NFL-caliber, so Albert threw a lot of touchdowns against the Chicago Rockets and the original Baltimore Colts, who folded after one season in the NFL. Once Albert started facing real competition, his production fell to earth, and the Niners dropped from 9-3 to 3-9.
116 comments, Last at 04 Jun 2011, 5:20am by Kibbles