Looking back at FEI's preseason projections, we find that most teams did about what they were supposed to do -- but not in the Big Ten, where things got screwy.
02 Dec 2005
by Mike Tanier
Carson Palmer took the center snap, turned quickly, and fired a pass into the flat to T.J. Houshmandzadeh. The receiver tried to turn upfield, but he was quickly corralled by Ravens cornerback Samari Rolle. The Bengals lost one yard on the play.
Palmer leads the league in completion percentage. His receivers haul in 69.7 percent of his passes; at this rate, he could post one of the highest figures in NFL history. But only a few of those passes are 50-yard bombs to Chad Johnson. Many are quick screens to Hooch that lose yardage. Or out-patterns by Kevin Walter that lose yardage. Or flairs to Chris Perry for one yard. Or screens to Rudi Johnson that travel about 10 feet through the air and net just five yards on the field.
Palmer threw all of those short passes against the Ravens in one of his best games of the year. How many more crummy little throws has he executed this year? How much is that 69.7 percent figure inflated by glorified handoffs, throws so short that they don't help the offense a lick?
And Palmer's not the only offender. Quarterbacks around the league are padding their stats by dumping off two-yard passes to their fullbacks. That's why NFL passers complete an average of nearly 60 percent of their passes. In the mid-1970s, when men were men and quarterbacks threw deep, that total was closer to 50 percent.
Modern quarterbacks execute sissy throws so they can pump up their rankings in that namby-pamby "efficiency rating" statistic that doesn't work and nobody believes in. Or so some critics suggest. But just how prevalent are these short completions? Are players like Palmer really building gaudy stats out of useless three-yard gains? And if we isolate the weenie receptions and remove them from the equation, can we separate the men from the boys and the downfield bombers from the dump-off pretenders?
Through Week 11, a total of 1,560 passes were completed for gains of five yards or less. That figure includes 76 completions for no gain and 121 completions for a loss. In an average game, each team completes about five passes that net five yards or less.
The Ravens lead the league with 75 of these short passes. They are followed by the Cardinals (70), Eagles (68), Patriots (66), Buccaneers (65), Packers (64), Lions (60), Bengals (60), Rams (56), and Redskins (54). Palmer, clearly, is among the league leaders in dink-and-dunk completions.
The Giants are at the bottom of the list; Eli Manning and company have completed just 30 passes of five yards are less. The bottom five also includes the Steelers (31), Niners (31), Falcons (34), and Panthers (37).
The raw data on short passes is polluted by several factors. First, we are dealing with the results of plays, not the actual lengths of the throws. If Tiki Barber catches a 36-inch toss and runs 36 yards, that play is not in the data. But that's not a very big problem. After all, a quarterback who spots a player two yards away with an acre of open field in front of him deserves credit for an alert read and a wise decision. The "short completions" we are looking for are plays that went nowhere, the ones that had almost no offensive value.
But some short receptions are important. First downs and touchdowns are included in the totals above; while three-yard completions aren't very impressive in most situations, they can be extremely important on third-and-2 or at the three-yard line. The Eagles threw 15 short passes that resulted in first downs or touchdowns, the Patriots 14. These passes aren't the weenie throws we're looking for.
Discard the first downs and touchdowns and the order of the top 10 changes somewhat. The Ravens and Cardinals still lead the way with 69 and 63 "useless" short passes. The Buccaneers rise to third with 55, followed by the Packers (54), Eagles (53), Lions (53), Rams (53), Patriots (53), Rams (52), Bengals (49) and Vikings (48). The Redskins drop to 11th with 46. The Giants threw just 21 useless short passes, the Steelers 23, the Falcons and Panthers 27 each.
The data contains another, more powerful pollutant: many of the teams with the most short passes are simply the teams that throw the ball the most. The same is true at the bottom of the list. The Cardinals (420) lead the league in pass attempts, followed by the Eagles, Rams, and Packers; the Ravens (believe it or not) and the Patriots are also among the top 10. The Niners (231) and Steelers (240) have attempted fewer passes than any other team, with the Panthers and Falcons among the bottom ten. In other words, there's a self-selection bias at work: obviously, when one team throws 190 more passes than another, they'll complete a few more three-yarders.
The best way to filter out this bias is to simply divide by attempts. By dividing completions of five or fewer yards (minus first downs and touchdowns, so teams aren't penalized for short-but-clutch throws) and dividing by attempts, you get Short Completion Percentage. Here are the top five and bottom five in this category.
Suddenly, the data seem intuitive. Anyone who has watched a Ravens, Bills, or Lions game knows that those teams are constantly dumping passes to running backs or asking their receivers to run short hitches or drag routes. Steelers and Panthers fans know that when their quarterbacks drop back to pass, they mean business. The Eagles may have completed 68 short passes, but take away the touchdowns and first downs and divide by attempts, and they earn a Short Completion Percentage of 12.8, almost precisely league average. The Bengals, at 14.8 percent, are sixth: Palmer may have a little padding in his numbers, but he's not a chronic offender.
Other players and teams have much more helium in their figures. Critics of the quarterback rating statistic certainly have a case when talking about the Ravens, Lions, and other teams near the top of the Short Completion Percentage table. Think about it: 19.2 percent of Kyle Boller or Anthony Wright's pass attempts result in a meager gain. That works out to 31.5 percent of their completions. Baltimore quarterbacks complete 60.8 percent of their passes, Bucs quarterbacks 61.7 percent; they are above the league average, and the NFL's antiquated passer rating system judges them to have outstanding accuracy and efficiency. Our eyes tell us otherwise.
In fact, it might be informative to wipe out all of these short passes by subtracting each team's Short Completion Percentage from its real completion percentage. The result isn't quite a "long completion percentage" -- remember that many 20-yard passes begin life as one-yard dumpoffs, and incomplete short passes aren't removed from the data -- but it's a decent estimate of a team's effectiveness when actually throwing down the field.
The table below shows the top teams in Non-Weenie Completion Percentage, for want of a better term. Also included is each team's rank in real completion percentage
|Non-Weenie Completion Percentages|
Some of the league's most effective offenses top the non-weenie list, but they also are among the leaders in old-fashioned completion percentage (and quarterback rating, for that matter). The Giants are an interesting case, as Eli's low completion percentage seems out-of-whack with the rest of his stats. The numbers above suggest that the younger Manning isn't completing many dump-off passes, opting instead to take shots downfield or throw the ball away. His high completion percentage on longer passes suggests that he's very effective when putting a little distance on his throws.
The Dolphins' completion percentage drops to 38.4 (from 50.9) when the short tosses are removed. The Niners, Bears, Jets, Ravens and Lions join them near the bottom of the list. Only the Ravens are above the league average in actual completion percentage. Take out the dump-offs, and they take their rightful place among the league's weakest offenses.
As for Palmer, well, he completes some weenie passes, but he completes a heck of a lot of long throws as well.
But wait, you say: why are four and five-yard passes being lumped into this study? A five-yard pass on 1st-and-10 is a pretty good play. Two five-yard passes equal a first down. What happens if you limit the study to really short completions?
The first thing that happens is that the data sample shrinks, and it doesn't shrink proportionally. Those 68 short passes thrown by the Eagles? It turns out that 40 of them earned four or five yards. Of the Patriots' 66 shorties, 41 were relatively productive. Meanwhile, only 32 of the Ravens' 75 short passes earned four or five yards. It seems like some teams use the "long handoff" more effectively than others.
The teams that complete the most "super-short" or "eenie-weenie" passes are the Ravens (11.9 percent of attempts), Lions (11.8), Bills (11.7), Vikings (10.7), and Bears (10.5). All bad offenses, to be sure, but the Bengals (10.0) and Chargers (9.2) are among the top ten. Only 4.2 percent of the Steelers' pass attempts net three or less yards.
If you count only passes that gain more than three yards, the Colts have a completion percentage of 63.6. The Rams, Seahawks, Bengals and Chargers round out the top five. The bottom five looks familiar: Miami (42.7 percent), San Francisco (43.7), Chicago (44.4), the Jets (45.6), and the Bills (47.3). The Ravens are seventh from the bottom
So who is responsible for all of these mostly harmless little dinks and dunks: conservative offensive coordinators, skittish quarterbacks, or receivers who can't make moves in the open field? That's a knotty question; the answer is certainly all three, to a certain degree.
When a team changes quarterbacks, sometimes its percentage of short completions changes drastically. In Buffalo, Kelly Holcomb completed a short pass (five yards or less) on 18.9 percent of his attempts. J.P. Losman's Short Completion Percentage is 9.6 percent. Holcomb's regular completion percentage is significantly higher than Losman's, with short throws comprising much of the difference.
A similar split occurs in Tampa, where Brian Griese completed 20.1 percent of his attempts for shorties, Chris Simms 12.7 percent. On the other hand, Daunte Culpepper's Short Completion Percentage was 13.4; Brad Johnson's is 15.3. The numbers are similar, though the players themselves have different strengths, and Culpepper completed about 10 percent more throws overall. The Vikings playbook must emphasize short passes to some degree, with decidedly mixed results.
Because you are curious: Joey Harrington completes 14.5 percent of his passes for short gains, Jeff Garcia 18.8 percent. Garcia's data sample is really too small to be useful.
Most short passes are thrown to running backs, so it makes sense that two of the best receiving backs in the NFL lead the league in short receptions: LaMont Jordan (17) and Brian Westbrook (16).
Both Jordan and Westbrook are capable of breaking free any time they touch the ball, so it makes sense that their quarterbacks keeping feeding them short throws. Third-down backs and all-purpose backs also catch lots of short passes, so Chris Perry (14), Chester Taylor (14) Obafemi Ayanbadejo (14), Ronnie Brown (13), Marcel Shipp (12) and Shawn Bryson (12) are all among the league leaders in short receptions. LaDainian Tomlinson also has 12.
Passes to Tomlinson and Jordan are worth the risk of a two-yard gain, but what about passes to Justin Peelle, Josh Parry, John Paul Foschi, or Derrick Wimbush, all of whom caught multiple short passes this year? Wimbush had three receptions as of Week 11, none of which cleared the five-yard mark. Sometimes, the quarterback would have been better off taking a shot downfield or trying to run; Peelle, for example, lost yardage on one third-and-12 catch. But a closer look at the data shows that even fullbacks and backup tight ends can help the offense by gaining five yards on first down or four yards on second-and-5. One of Wimbush's three catches, for example, converted a fourth down.
Wide receivers don't run many routes shorter than five yards, but drags, hitches, and "bubble screens" are popular this year. Derrick Mason has caught 13 short passes, paying a price after many of them. Deion Branch, Larry Fitzgerald, and Muhsin Muhammad each have 11 short receptions.
The Patriots throw more dump-offs to wide receivers than most teams; Tom Brady's quick tosses to Branch, Givens, and others are clearly designed to gain a few yards while stretching the defense horizontally. When Mason or Muhammad catches a four-yard pass, it looks more like a desperation strategy: get the ball to the playmaker, and maybe he'll make 20 moves and score a touchdown.
For the record, the shortest completion of the year was a catch by Jamal Lewis that lost nine yards. It was one of five completions by the Ravens that lost yardage. The Vikings lead the league in receptions for a loss, with nine, split mostly between Michael Bennett and Mewelde Moore. The Lions have completed eight passes for a loss.
Obviously, the straw-man criticism of Carson Palmer at the start of this essay was unwarranted. Palmer completes a lot of passes, period: short ones, medium ones, long ones. Peyton Manning, Tom Brady, and Drew Brees can say the same thing. Take away the occasional two-yard dunk to the fullback, and these players still have great completion percentages and excellent stat lines.
On the other hand, some players and teams are clearly over-reliant on short completions. Players like Anthony Wright and Joey Harrington complete a lot of passes, but so many of them yield meager gains that they aren't doing their teams any good. You can argue that even a two-yard completion is better than an incompletion, but what about a two-yard completion versus a 10-yard Wright scramble? Three two-yard completions, or three shots further down the field, one of which could gain 20 yards?
Many of the league's weakest offenses -- the Lions, Ravens, Bills, Bears, and others -- appear to be gorging on these non-nourishing little plays. But is this cause or effect? Does throwing too many short passes make their offenses bad? Or does offensive ineptitude lead to lots of three-yard catches?
The Patriots, Bengals, and Chargers are able to handle these weenie completions just fine. The Eagles used them wisely before everyone was hurt. The Panthers, Falcons and Steelers eschew them altogether in favor of something called a running game. And of course, the 49ers complete so few passes in general that they throw everyone else's stats off.
There's more data here than conclusions. And there are other questions. What about short completions in wins versus losses? On first down versus other downs? What about six-yard completions?
What we need is a way to measure the value of all of these plays. How useful is that three-yard completion on second-and-6? How many of a quarterback's completions are essentially useless? Maybe each individual pass can be rated, compared to league averages, and adjusted based on the game situation. That would be an excellent measure of pass offenses and of quarterbacks. It would take away most of the biases and numerical padding, leaving us with just the true value of a team or player's performance.
Oh, yeah. I just described DVOA.
(Notes: All statistics calculated through Week 11 except for those pertaining to Palmer in the opening paragraphs. Individual quarterback Short Completion Percentages, like the Griese-versus-Simms comparison, do not include short completions for first downs or touchdowns.)
74 comments, Last at 16 Mar 2012, 1:32pm by parissportif