Adjusting Completion Percentage

by Aaron Schatz

Pro Football Prospectus 2007 officially hits stores today, and it is packed to the gills with information. So packed, in fact, that we couldn't fit in everything we wanted from the Football Outsiders game charting project. You'll find quarterback hits and hurries in the book, and stats that measure defensive backs in coverage, and YAC and pass lengths for both quarterbacks and receivers. You'll find a big long article on when play-action works best, and information on how often teams blitzed or lined up in various formations. Despite all that good stuff, we've barely scratched the surface with what we can do with the game charting information.

One thing we originally planned was a page in the quarterbacks section which would give a variety of tables showing which quarterbacks had the most or fewest incomplete passes of different types: underthrown, overthrown, dropped, thrown away, defensed, and so on. It didn't fit in, so we figured we would look at that information over the next couple of days.

(I should point out at this point that Bill Moore didn't just spend hours coordinating the whole game charting project. He also put together a lot of these tables.)

The first thing we wanted to look at was dropped passes. Dropped passes is one of those stats that we all know is out there, but it isn't easy to find. STATS Inc. keeps it, so you'll find it on some individual player pages on websites where STATS provides the numbers, but not all the individual player pages, and you don't know what kind of passes were dropped -- are we talking long bombs here, or running backs who couldn't get the handle on screen passes?

We have our own count of dropped passes, thanks to the game charting project. By our count, Miami led the league with 46 drops, followed by Seattle with 43, Philadelphia with 40, New Orleans with 39, and Green Bay with 39. Buffalo had the fewest charted drops, just 12, a bit of a surprise since they picked up drop machine Robert Royal before the 2006 season. The book goes into further detail about which players had the most drops, but today we'll look at what drops mean for quarterbacks.

Not counting spikes to stop the clock, last year NFL quarterbacks completed a little over 60 percent of passes. If you were to consider drops as completed passes, completion percentage goes up to about 66 percent. Here's a look at the quarterbacks with the highest completion percentages, with and without including drops as completions. The standard completion percentage will be different from the NFL numbers because it includes only charted passes, without spikes or plays missing from our charting database. Minimum to be ranked is 100 pass attempts.

Hey, what a surprise, Peyton Manning is the best in the league at something. Counting drops moves up the completion percentage for Tom Brady and, surprisingly, David Garrard. We charted 33 drops by Jacksonville receivers, and 21 of those had Garrard at quarterback, even though the Week 17 game with Kansas City is one of the handful of games that was not charted.

Here are the guys at the bottom of the league:

Well, I guess Michael Vick has a little bit of an excuse; adding drops as complete passes means he no longer has the lowest completion percentage in the league. Vince Young also moves up a bit.

Of course, the top quarterbacks in completion percentage are not necessarily the most accurate quarterbacks in the league. We know that David Carr and Charlie Frye had high completion percentages because they were throwing a ton of screens and dumpoffs. What happens if we adjust completion percentage based on the types of throws each quarterback was making?

In the book, we separate passes into four categories. I took each category and determined the league-average completion percentage, including drops as complete passes:

• Short (5 yards or less): 80.5 percent
• Mid (6-15 yards): 65.0 percent
• Deep (16-25 yards): 52.0 percent
• Bomb (26+ yards): 33.6 percent

I then adjusted each quarterback's completion percentage based on how his completion percentage on passes of each distance compared to the league average, and added back in all the passes that don't fall into these categories: balls thrown away, tipped at the line, and so forth. Once again, I included dropped passes as completions.

Here are the best and worst quarterbacks from 2006:

Surprise! David Carr and Charlie Frye remain in the top ten. Carr and Frye didn't just throw a high number of short passes; they also threw very accurate short passes. Frye completed 89.8 percent of his pass attempts of five yards or less, the highest percentage in the league. Carr was second at 87.2 percent. Then again, maybe Carr and Frye weren't just throwing short passes -- they were throwing really short passes, stuff behind the line, or dumpoffs to uncovered receivers on third-and-10. What if we take out the short stuff?

OK, that looks a bit more like a table of the league's most accurate quarterbacks. Tim Rattay is number one, but that's partially due to small sample size (he barely gets over our 100 attempt minimum) and partially due to the nature of Tim Rattay. Rattay's problem has never been that he can't play, it's that he can't play without getting hurt. The two Seattle quarterbacks are a bit of a surprise, but Seattle had a big problem with dropped passes last year (what else is new) and both Hasselbeck and Wallace were quite accurate when it comes to midrange throws of 6-15 yards. That makes a lot of sense in the Mike Holmgren offense. (This is nothing compared to the other huge surprise about Seneca Wallace, but you'll have to buy the book for that one.)

Way down at the bottom of the list, however -- is that really Brett Favre? Yes, it is. Favre was terrible last year when it came to completing long-distance throws. His completion percentage was just 34.4 percent on passes of 16-25 yards, and just 20.4 percent on passes over 25 yards. It makes sense that an older Favre may have lost arm strength and the ability to be accurate when throwing deep. The Packers may want to dial down on those long passes and put more midrange throws into the playbook to take advantage of Favre's remaining strengths and hide his weaknesses.

Perhaps the most interesting player when it comes to this "adjusted completion percentage" statistic is Tom Brady. When we include all passes, Brady has an adjusted completion percentage of 69.2 percent, just outside the top ten. When we take out the short passes, Brady drops to 22nd, with an adjusted completion percentage of 54.1 percent.

Why the drop? Of the league's top quarterbacks, the only two threw more than 50 percent of their passes below six yards: Tom Brady and Drew Brees. When we take out short passes, the quarterbacks whose completion percentages are most dependent on short passes will suffer the most. Brady's completion percentage on passes longer than five yards isn't much different than the league average. You can see that taking out short passes drops Brees in "adjusted completion percentage" as well, but not by anywhere near as much because Drew Brees actually completed more than 65 percent of all passes over 15 yards last year. That includes more than 65 percent of all Bomb passes, passes over 25 yards. Tony Romo was the only other quarterback to complete more than 50 percent of Bomb passes in 2006. Not bad for a guy who was criticized for his arm strength coming out of Purdue.

In Brady's defense -- and perhaps in defense of Carr and Frye as well -- there is a skill to completing short passes, just as there is a skill to completing long passes. Brady had a DVOA of 30.0% on short passes, which ranked fourth in the league behind Trent Green, Carson Palmer, and Marc Bulger. Perhaps nobody in the league has better timing on when to throw a screen pass than Tom Brady. However, if you are expecting lots of long bombs to Randy Moss and Donte' Stallworth this year, you may be disappointed.

(By the way, David Carr was fifth on short passes with a DVOA of 29.7%, but Charlie Frye was near the bottom of the league with a DVOA of 3.0% on short passes. Nearly every quarterback has a positive DVOA on short passes because only measuring passes where we can record a pass length automatically leaves out sacks, throw aways, and passes tipped at the line.)

Wednesday, I'll look at which quarterbacks had the biggest accuracy problems (underthrown and overthrown passes) and which quarterbacks suffered the most due to the pass rush (thrown away, tipped at line, and hit in motion passes).