Special thanks to Doug Drinen and Clayton Gray for the statistical groundwork they provided for this study.

Most fantasy rankings and projections begin by looking at the previous year's stats, a simple and sound starting point. While those raw numbers display how well a quarterback produced, they don't consider the strength of his opponents. We all know the Pittsburgh Steelers were a considerably tougher opponent than the Kansas City Chiefs, and a quarterback should get more credit for producing well against tougher defenses. The table below shows how many fantasy points per game each team's defense allowed to opposing quarterbacks:

By examining how effective each quarterback is against a defense, we can help to see which QBs truly were difference makers. Below are two stat lines from the 2004 season:

Vinny Testaverde, week 1: 29/50, 355 yards, 1 TD/0 INT; 1 rush, 0 yards; 21.8 FPs
Joey Harrington, week 11: 15/27, 196 yards, 1 TD/1 INT; 6 rushes, 19 yards; 14.7 FPs

Testaverde appeared to have the much better game, since he passed for 159 more yards, and scored 48% more fantasy points than Harrington. But Testaverde scored 21.8 FPs against the Vikings, a defense that allowed 21.8 FP/G on average to opposing QBs. Harrington faced Arizona, a much tougher opponent that let up just 14.7 FP/G to QBs. Both quarterbacks performed like the league average against their opponents. After using this analysis for each quarterback and every game last year (with two caveats, seen at the bottom of the page), we can see which QBs were really the best.

For starters, let's look at which quarterbacks faced the easiest schedules. "SOS FP" is simply the average FP/G allowed by each opponent the quarterback faced. "SOS FP/G" is "SOS FP" divided by the games played by each QB. So when Ben Roethlisberger's first four games were against the Ravens (14.1 FP/G allowed), Dolphins (14.0), and Bengals (16.6) and the Browns (14.7), his SOS FP total was 59.4, and his SOS FP/G was 14.9 through four games. Think of those numbers as what the average quarterback would score against those defenses. "FP" and "FP/G" show how many actual fantasy points and fantasy points per game each quarterback scored in 2004. The higher the "SOS FP/G", the easier the strength of schedule.

The AFC South QBs took the top four spots on the chart! Teams in that division had four games against the bad defenses of the NFC South and four more against the similarly inept AFC West. The bottom three QBs on the chart all play in the AFC North, as they faced much tougher competition in the form of the AFC and NFC East. Based on this data, we can downgrade the great performances by Billy Volek and Peyton Manning, and upgrade the less than stellar seasons from the state of Ohio. So how exactly do we quantify it? Let's use a pair of former number one picks as examples:

Drew Bledsoe averaged just 13.4 FP/G, ranking him as the 29th best quarterback. He faced tough defenses nearly every week though, as his average opponent allowed 16.1 FP/G. Since the league average defense allowed 17.3 FP/G we add the difference - 1.2 FP/G - to Bledsoe's total. As a result, his "Adjusted FP/G" score becomes 14.6 FP/G.

David Carr appeared to make big strides last year, improving his FP/G production from 12.6 FP/G in 2003 to 16.0 FP/G last season. He faced a very easy schedule however, where his average opponent allowed 18.8 FP/G. Compared to the league average, he was facing a defense that was 1.5 FP/G softer. So after subtracting that amount from Carr's total, his "Adj FP/G" was the same 14.6 FP/G that Bledsoe averaged. (The numbers don't exactly add up due to rounding differences). Initially Carr ranked as the 18th best quarterback and Bledsoe the 29th; after adjusting for strength of schedule, Bledsoe moved up to 24th while Carr moved down to 25th.

The following table shows each quarterback's "Adj. FP/G" score, FP/G, where each QB ranked in both categories, and the difference between each rank.

"Dif/Gm" shows much better than the league average (17.3 FP/G) each quarterback was last season. This shows which fantasy QBs were real difference makers, and which were simply products of their schedule. By definition, "Dif/Gm" = "Adj. FP/G" - 17.3.

David Carr moved down the most ranks (seven), while Tom Brady moved up the most (eight spots). Drew Brees, Byron Leftwich and Kerry Collins each moved down five spots after adjusting for their cake schedules; Jeff Garcia, AJ Feeley and Drew Bledsoe each moved up five spots due to very difficult opponents.

This chart shows which quarterbacks were the best ones last season, perhaps making it a better starting point for your QB projections than last year's raw stats. If you project based on 2004 and state that Carson Palmer didn't do well last year and has a tough schedule this year, you'll be erroneously penalizing him twice for tough schedules. It's better to say that he was actually pretty good last year (13th in FP/G), but then downgrade him if you think his 2005 schedule will be difficult. If you don't believe much in the predictive value of strength of schedule, then there will be no need to downgrade him at all this year.

Random Thoughts

QBBC

This gives a lot of support to those advocates of quarterback-by-committee. Only Manning, Culpepper, McNabb and Trent Green were three fantasy points better per game than the average QB. Most of the stars from last year had very easy schedules that made them great value picks. In last year's QBBC article I advocated drafting Jake Plummer (good), Jake Delhomme (good) and Brad Johnson (not so good). I had ranked Denver's schedule as the easiest one in the NFL last year, and that helped Plummer have a breakout season. If you can't grab one of the top QBs, I certainly recommend following a QBBC approach with a strong eye on strength of schedule.

Plummer vs. Hasselbeck

Most fantasy gamers would explain that Plummer was a great value last year, while Hasselbeck was a big disappointment. While it's true that Plummer ranked fifth in total fantasy points and Hasselbeck was only thirteenth, the difference may not reflect their abilities as much as you might imagine.

Consider that Plummer played two more games, and he only ranked 7th in FP/G while Hasselbeck was still 13th in FP/G. The big difference however, lies in each QB's schedule: The league average QB would have scored 299.9 FPs last season playing the Broncos schedule, while Plummer scored 318.7 - so Plummer only scored 1.17 FP/G more than an average quarterback. Hasselbeck's schedule was considerably tougher and based on his fourteen-game schedule the league average QB would have scored 241.1 FPs, while the Seahawks star had 257.1 FPs. That difference of 16 FPs over 14 games shows that Hasselbeck was just as skilled as Plummer last year: he scored 1.14 more FP/G than the league average.

Best and Worst

Another observation can be made by checking each team's defense and see which QB scored the most points on them, and which QB they shut down the most. (Note: Only the 33 QBs in this study were used. So if hypothetically Quincy Carter had a six touchdown performance against the Patriots, that wouldn't show up.)

Six different QBs scored the most points against more than one defense:

• Brady (vs. Buffalo and Pittsburgh)
• Bulger (vs. Seattle and New York Jets)
• Green (vs. New England and Tampa Bay)
• Vick (vs. Carolina and Denver)
• Manning (vs. Detroit, Jacksonville and Kansas City)
• Culpepper (vs. Chicago, Philadelphia, Houston and New Orleans)
• McNabb (vs. Washington, Dallas, Cleveland, Green Bay and New York Giants)

McNabb was scored more points than any other QB last year against FIVE different teams. The former Orangeman star's feat is even more impressive considering that he only faced eleven unique teams last year.

On the other side, nine QBs were the low scorer against more than one defense:

• Boller (vs. Washington, Cleveland, Buffalo and Cincinnati) led the group with four.
• Surprisingly, Leftwich (vs. Houston, Tennessee and Oakland) was second with three cellar performances.
• Each other QB had two: Collins, Dorsey, Harrington, McCown, Testaverde, Vick and Volek. Mike Vick was the only QB to appear on the "multiple best games" and "multiple worst games" list. Of course to anyone that followed fantasy football last year, reporting Mike Vick to be inconsistent is as newsworthy as a Drew Rosenhaus holdout.

Best and worst individual games

The three best and three worst individual performances based on adjusted fantasy points:

Best

1. Volek vs. Oakland, week 15: 492 yards, 4 TDs, 1 rushing TD = 45.7 FPs. Even when you adjust for the poor defense (21.8 FP/G allowed), Volek still scored 23.9 FPs above average.
2. Culpepper vs. Houston, week 5: 396 yards, 5 TDs, 30 rushing yards = 42.8 FPs. After adjusting for SOS (Houston allowed 19.5 FP/G), Culpepper was 23.3 FPs better than the norm.
3. McNabb vs. Green Bay, week 13: 464 yards, 5 TDs = 43.2 FPs. The Packers allowed 21.0 FP/G on average, making McNabb 22.2 FPs better than average. Don't forget however, that McNabb accomplished most of this in the first half.

Worst

1. Brunell vs. Cincinnati, week 10: 1/8, 6 passing yards, 1 interception, 3 rushing yards = -0.4 FPs. After adjusting for defense, Brunell scored 17 FPs lower than average.
2. Leftwich vs. Houston, week 16: 6/14, 35 passing yards, 1 interception, 19 rushing yards = 2.7 FPs. This game would knock the Jaguars out of the playoffs, and Leftwich suffered a mild concussion. He was a miserable 16.9 FPs (and 190 passing yards) below average against a bad Texans defense.
3. Brunell vs. Detroit, week 9: 6/17, 58 passing yards = 2.9 FPs. Against a weak Lions defense (19.3 FP/G allowed on average), he was 16.4 FPs below average.

Brunell and Ramsey, Manning and Culpepper

Brunell had just one game out of nine where he scored more than the league average, and that was against the Cowboys. Patrick Ramsey wasn't any better, with his week 13 game against the Giants being his only performance out of nine where he was above average. On the other side, Peyton Manning had just one game out of fifteen where he scored below the league average, with his 198/3 performance against Oakland just 0.9 FP lower than average.

Culpepper had three games out of sixteen where he scored below the league average against a particular defense. They occurred consecutively against the Titans, Giants and Colts in weeks seven, eight and nine - all with Randy Moss on the sidelines.

Random craziness

• The Green Bay Packers faced the thirty-three QBs thirteen times last year. Only twice did the Pack hold the opposing QB to less than ten fantasy points. And both times that guy was Joey Harrington. In fact, his 14.5 combined fantasy points in two games would have still ranked at the bottom of this list. How bad is Harrington against Green Bay? Here is his stat line from his four games the past two seasons against them:
  
  64/132 (48%) for 572 passing yards, 1 TD/6 INT and a 47.3 QB rating
  
  
• Week 15: The week of the surprising quarterback. Billy Volek (45.7 FPs against the Raiders), Kerry Collins (38.6 FPs against the Titans) and Josh McCown (37.7 FPs against the Rams) were the top three scorers in week fifteen. The only other week where three QBs topped the 35 FP mark occurred in week 8 in the AFC West: Manning (42.6 FPs against KC), Plummer (38.5 against Atlanta) and Brees (36.2 against the Raiders).
  
  
• Which divisions were the best and worst for opposing QBs? The chart below shows the average FP/G allowed to opposing QBs by each division.

• You know the old adage that says "If you have two starting quarterbacks you don't really have any?" Well that idea rang true in this study, as two teams had two QBs on the list and one had zero. Only three QBs averaged less than 13 adjusted fantasy points per game: Mark Brunell (Was), Ken Dorsey (SF) and Patrick Ramsey (Was). Tim Rattay (SF) was actually just barely below league average, but that was in large part due to his high number of passing attempts - only Kerry Collins averaged more passes per game. Chad Hutchinson, Craig Krenzel, Rex Grossman and Jonathon Quinn combined for 173 FPs for the Bears last year. Based on Chicago's very easy schedule (their average opponent allowed 18.7 FP/G to QBs), their quarterbacks were a whopping 7.9 FP/G below league average, easily the worst mark in the league.

As always, please send any comments and questions to stuart@footballguys.com.

Article Notes

The phrase "all quarterbacks" really means "all quarterbacks that have thrown enough pass attempts to qualify for inclusion in this study." Certain NFL statistics such as quarterback rating require quarterbacks to have thrown fourteen passes per team game (224 in a season) to qualify. The thirty-three quarterbacks that attempted at least 224 passes last season were the ones included.

In the interest of making the study as meaningful as possible, there were certain games that I decided to remove. I eliminated the following fourteen "partial games" from each quarterback, as I deemed them inadequate to determine overall effectiveness. Assigning Marc Bulger a low score for a game when he was hurt after just four pass attempts inaccurately reflects his ability. Here are the fourteen: (Note: Certain partial games, like Mark Brunell's week ten performance against the Bengals were not excluded from the study. Despite not throwing many passes, he was benched due to ineffectiveness, and that should be noted.)

• Bulger vs. 49ers, week 13: Injured after four passes
• Dorsey vs. Falcons, week 1: Replaced Tim Rattay in 2nd quarter, hurt in 3rd quarter
• Dorsey vs. Bills, week 16: Injured in first half
• Favre vs. Lions, week 17: Played just one quarter and one series
• Feeley vs. Patriots, week 5: Two pass attempts
• Feeley vs. Cardinals, week 9: Replaced Jay Fiedler late in 3rd quarter
• Garcia vs. Bills, week 14: One pass attempt
• Manning vs. Broncos, week 17: Played one series
• McCown vs. Jets, week 12: Played late in second half
• McNabb vs. Rams, week 16: Played one series
• Testaverde vs. Bears, week 12: Played only second half
• Vick vs. Seahawks, week 17: Played one series
• Volek vs. Jaguars, week 3: One pass attempt
• Warner vs. Ravens, week 14: Replaced Eli Manning late in second half