Football Frontier

The sophomore slump is a term we’re all familiar with. Rookies who do well, or even just decently, are doomed to regress in their second season in the NFL. But does this phenomenon exist for quarterbacks?

To take a look at this concept, I examined the records of 102 quarterbacks who started for their teams in their rookie year or their second year, and then started the following year.

Among rookies who started their first two years, I found an average increase in performance of four points in my quarterback rating. Forty of 55 quarterbacks improved in their second season, thirteen regressed. The two most notable cases in the last 39 years, by the way, were Colt McCoy and Matt Ryan, both dropping 11 points in their second season. Ryan rebounded quickly. McCoy lost his job.

The average score for the rookies who started was 44. For those who were better than average as rookies, their average rating was one point higher as sophomores. Those who were worse than average saw an average rating increase of seven points. Of the six rookies who scored better than 51 in that first season, two definitely improved (Ben Roethlisberger and Dan Marino), Ryan dropped considerably, and the other three were within one point of their rookie score in their second season.

Looking at those who started for the first time as sophomores, and then started in their third year, the average improvement was two points. Of that group of 47 quarterbacks, 27 improved in that third season and 17 regressed. Of the twelve who were above average in their second season, the average drop was 5 points in their third season. Those who were below average saw an average increase of four points.

Seven of the 12 who were above average in year two regressed significantly. These seven include Tom Brady, Brett Favre, Steve McNair and Daunte Culpepper. Four remained about the same, and one, Joe Montana, improved considerably in his third season.

Summarizing: I was unable to find any indication that a sophomore slump is the norm amongst NFL quarterbacks. In fact, it’s rather rare for quarterbacks who do not excel in their first season as a starter. For those who do excel, some regression to the mean is possible, but not the average result. The only group where I saw what might be the expectation of a second-year slump was in those quarterbacks who earned their first starting job in their second season and excelled. Their third seasons were likely to be not quite as good. Though the sample size is very small there, it includes some famous names who turned it back around in the following season.

Much is made of the fact that Aaron Rodgers, one of the top three quarterbacks in the NFL today by many measures, was in his fourth season before he was handed the starting job for the Green Bay Packers. He stepped into the role and was effective immediately. Within a year, Packers fans loved him almost as much as they loved Brett Favre (OK, that might be a bit of an exaggeration).

But did Rodgers benefit from that apprenticeship?

To examine this question, I looked at the 60 quarterbacks who were drafted in the first round from 1974-2012 and have been starters. For the most part, you don’t draft a quarterback in the first round if you’re not hoping he takes the starting job right away. These are supposed to be franchise quarterbacks. So 38 of the 60 started as rookies, and the other 22 started in seasons 2, 3 or 4. The average draft position of those who started as rookies was 7th and the average for the others was 13th. So a direct comparison may not be as revealing as it should be.

How did the immediate starters fare? Their average game score was a 43, significantly below the league average of 51. Keep in mind these are supposed to be franchise quarterbacks, so fans probably were a bit disappointed. Only four of the 38 were above average as rookies. The four are very familiar names – Ben Roethlisberger, Matt Ryan, Robert Griffin III and Cam Newton.

The 22 who waited averaged a 48 in their first season as a starter. Clearly, apprenticeship has value, especially since these 22 who waited were not drafted as high as the rookie starters. Seven of these 22 – including Carson Palmer, Jay Cutler, Philip Rivers and Rodgers – were above average in their first seasons.

But that’s not the entire story. There’s a cost to having your first-round pick ride the bench, especially since teams that pick quarterbacks in the first round often do so from need. So, how did those 38 fare in their second season? Four – Ryan Leaf, Tommy Maddox, Matthew Stafford (injuries derailed most of his season) and Matt Leinart – did not start the following season. And four others, of course, have yet to have a sophomore season because 2012 was their debut. So of the 30 who did start in that second season, the average game score was 47. In year three, with 24 data points, the average rose to 48.

Meanwhile, the second season of those who did not start as rookies, with 17 data points, was 49 – and year three rose to 54 (with 13 data points).

I extended this study to second-rounders, of which 24 fit the parameters of this study. Nine started as rookies, and 15 in years 2-4. The rookie starters scored an average of 44, 50 and 46 in their first three years. The ones who didn’t start as rookies scored 48, 49 and 51.

Keeping in mind that the sample size is small, it does seem that there’s a benefit to giving your franchise quarterback a year to learn the game before throwing him in the pocket. But circumstances could be very different. The ones who have an extra year could be receiving better mentorship, or perhaps they’re on better teams.

I can’t form any definite conclusions here, but for those GMs and coaches who believe a year of apprenticeship has a significant value, there’s certainly justification for that belief even though this year’s crop of rookies have been quite impressive. So when we see a team draft Geno Smith this April, it’s entirely possible that team would benefit from bringing in a veteran quarterback as well.

Many who have played the game say learning the quarterback position is the toughest job in the world of sports. There are so many skills to learn, not the least of which is deciding, within a split second, where to throw the ball against a defense that is determined to trick you into making a bad throw. And perfection is mandatory. If you’re indecisive, you’ll end up taking a drive-killing sack. If you make a poor decision and throw an interception, well, that’s about the only result in football that’s worse for an offense than a sack.

A team that plays an unprepared quarterback is asking for disaster. But how does a rookie learn in the first place?

To address this question, I put each quarterback’s games (with eight or more throws) in my 1974-2012 study in order. Then broke down averages for each game number from 1 to 324 (Brett Favre accounted for the only games numbered 263 or higher). At about game 190-195, the sample size became so small that individual games had too much impact. But what I found was surprising.

For games 1 and 2, the average game score was 44. The average didn’t drop below 45 again until game 204, where the sample size was just 6 (as opposed to 431 for game 2). Games 3 and 4 saw an average of 46, then the average remained in the 46-49 range until game 16. To try and smooth out the data, I calculated rolling averages, and found the following:

Games 5-7: 47, Games 8-12: 48, Games 13-17: 49, Games 18-31: 50, Games 32-57: 51 or 52 (the NFL average game score is 51), Games 58-117: 53 or 54, Games 118-132: 55, Games 133-175: 52-54, Games 176-195: 54-56.

What does that mean? The development curve is very steep during a quarterback’s first season, and his first two games can be completely overwhelming. After that, he steadily develops at a slower pace until he has about four seasons as a starter, at which point he is likely mature as a starting quarterback. After that, he will develop at a much slower rate until he has about nine seasons as a starter. He will then decline, slowly, but those who have 12 or more years as a starter tend to be the elite quarterbacks who started from day one, so that skews the data up a little until they hit the age (36-37) when most quarterbacks start aging quickly.

For my own purposes developing Front Office Football, I think this is quite interesting, and gives me ideas on how I can better age players within the game. What I have now is fairly accurate, but it could use fine-tuning, and this study shows me how to do that.

One question I had when looking at this data is whether that 44 average from games 1 and 2 was skewed by the quarterbacks who never get a chance to do more than spot-start. So I separated the quarterbacks with less than eight career games with eight or more throws. They averaged just 38 in games 1 and 2, and didn’t improve. That accounted for 99 of the 468 quarterbacks studied here. But the remaining 369 averaged a 46 in game 1, 45 in game 2 and 46 in game 3. So even for the quarterbacks who “make it,” those first few starts are usually quite rough.

You can’t underscore the importance of coaches who know a quarterback’s potential. Look at Pete Carroll in Seattle. The Seahawks have spent quite a bit bringing in promising quarterbacks with a little bit of experience – Charlie Whitehurst a few years ago and Matt Flynn last year. But Carroll gave the job to third-round pick Russell Wilson – all 5-11 of him. Wilson posted game scores of 24-71-52-44 and the Seahawks were 2-2 after four games (including a rather controversial win over Green Bay). That’s a little better than average for a rookie, but the season was headed south at that point. Carroll stuck with the rookie, and he averaged an amazing 68 the rest of the way. Seattle went from averaging 16.6 points per game in its first seven games to 31.6 in its last eleven. By the time Seattle reached the playoffs, some analysts were saying this was a Super-Bowl quality team. Wilson had a great game against Atlanta in the divisional round, but came up a bit short.

Of those quarterbacks from 1974-2012 who started as a rookie, Wilson’s 61 average game score for the season is second only to Ben Roethlisberger, who guided the Steelers into the playoffs as a rookie in the 2004 season. He won his first Super Bowl the following season. I don’t think I’m standing too far out on a limb when I say Seattle should have the lowest odds of any team to win next year’s Super Bowl.

For my final look at the quarterback project in this series of posts, I will examine the value of letting a quarterback learn by watching, rather than starting in his rookie season.

The average quarterback career lasts about 4 1/2 years. But what does that mean?

Since players come into the NFL at age 22-23, generally, does this mean they’re gone at age 27 as often as not? Not at all. Those who can play the game tend to stick around a long time, as long as they’re not injured.

We’ve all heard about running backs, who have the shortest careers. Analysts commonly say 30 is a magic age that no running back can pass and still hold up to hits. I believe that, but I’d like to see more numbers. The hits add up for running backs.

But they don’t necessarily add up for quarterbacks. Anyone who was watching football back in 1985 remembers the Lawrence Taylor hit on Joe Theismann that ended his career at age 36. Theismann was probably close to retirement anyway. His performance had declined steeply the last two seasons. And that’s more how it ends for quarterbacks. At some point, not necessarily at 36, decline begins, and they simply don’t retain a starting role.

The quarterback position is far too important in football to allow players to age gracefully and fade into the sunset. There comes a training camp when the ball just isn’t going where it needs to go, and all of a sudden someone else is starting. Or maybe the performance declines during the season and the coach and quarterback agree that next year isn’t going to happen. It’s very rare seeing a career like Brett Favre’s, where he looked like he was in decline at 36, kept his starting job after the worst season of his career at age 37, then somehow had a pretty good season at 38, then lost his job, was serviceable with the Jets at 39, then went to the Vikings at 40 and had a career year before going back to subpar at 41 and then retiring. Some of that was probably injury, and a good coach can tell whether a player has lost it or is hurting. So there’s undoubtedly an explanation for Favre’s career track. It’s unusual, though.

In general, young quarterbacks who are going to be long-time starters are starting at age 23-24. Average performance rises steeply from 21-25. Then it rises smoothly from 25-36. There’s some attrition along the line – only about 1 in 8 quarterbacks who starts in the NFL is still playing at 36. But those who are still playing can play fairly well. Average performance then declines slowly afterward. And only two quarterbacks – Vinny Testaverde and Warren Moon – had significant playing time after age 41. They weren’t that good by then.

Here’s a quick look at the starting quarterbacks in today’s NFL, where they are in their career arcs compared to past performers. The average quarterback performance is a 51. Included is every quarterback who threw at least eight passes in six or more games.

Buffalo Bills: Ryan Fitzpatrick is 30 and has been significantly below the NFL average at every age, though he has improved every year and the gap is reduced. He had a career-best 49 in 2012. It would be reasonable to expect him to remain below average, though he’s better than replacement level. The Bills will be looking to develop a quarterback.

Miami Dolphins: Ryan Tannehill is 24 and had an average season for a 24-year-old quarterback (46). The average first season for a quarterback is 43, so the Dolphins are probably pleased and are hoping Tannehill continues to develop.

New England Patriots: Tom Brady is 35, and his score of 61 this season is in line with his performances the last nine years. There’s no reason to believe he doesn’t have a season or three left in him at the elite level.

New York Jets: Mark Sanchez is 26. He had a good rookie season in 2009 (47), but followed that with 46, 46 and now a completely unacceptable 39. Unless he was playing hurt, I can’t see the Jets giving him the ball again. There are only a couple of long-time career arcs that look like Sanchez’s at the beginning. The most similar is Kerry Collins, who was never great, but seemed to get chances. Sanchez is probably going to be a backup the rest of his career, if he does keep a job.

Baltimore Ravens: Joe Flacco just turned 28. After a bit of a regression last year, he was back to being just a bit above average (53). That, and his amazing ability to win playoff games on the road, means he’s on the fringe of the top ten quarterbacks in the league.

Cincinnati Bengals: Andy Dalton is 25 and has been pretty much average for his age his first two seasons (45, 50). That makes things a little difficult for the Bengals, because he might become a franchise quarterback, or he might not. This coming season will be the test.

Cleveland Browns: Brandon Weeden, at 29, had an average season for a rookie, but a terrible season for a 29-year-old quarterback (44). There’s no precedent to look at to guess whether he will improve next year. I’d imagine he will, but the Browns undoubtedly, with new management, will look for a young quarterback to develop.

Pittsburgh Steelers: Ben Roethlisberger is coming off a slightly below-average season (55). He’s still one of the best in the league. He came into the league as a 22-year-old, which is unusual for a starter, and posted the best score for a 22-year-old (62) in the study (1974-2012).

Houston Texans: Matt Schaub is 31 and coming off a score of 56. He didn’t start until his fourth year, but when he did, he was already much better than average. His six seasons range from 56 to 67. Since he was a third-rounder with no hype in 2004, it’s easy to level criticism. But he is definitely a top-ten quarterback.

Indianapolis Colts: Andrew Luck is 23 and posted a 43 in his rookie season. While that’s pretty bad, reports indicate he was asked to do a lot, and since the Colts won 11 games, there’s reason for optimism. Though I wouldn’t be surprised if the Colts dropped to 7-9 wins in 2013. Peyton Manning had a 48 score coming into the league at age 22, though the Colts only won three times.

Jacksonville Jaguars: Blaine Gabbert is 23, and had a score of 45 following a disastrous 32 in his rookie season. Chad Henne scored a 42 at age 27. Gabbert probably gets another shot at this, but the quarterback situation in Jacksonville is bad enough that I’m sure the Jaguars would take one with the second pick if they felt there was a quarterback worthy of the pick.

Tennessee Titans: Jake Locker is 24 and posted the average score of 46 for a 24-year-old. He will get at least another year to show development. Matt Hasselbeck is 37 and has posted only one season above 50 since taking the Seahawks to the Super Bowl in the 2004 season. He’s really a backup now, at best.

Denver Broncos: Peyton Manning is 36 and coming off one of his best seasons (67). He has not been below 57 since his rookie season. If he’s healthy, there’s no reason to doubt he can play at least another season or two at an outstanding level. He is undoubtedly one of the top two or three quarterbacks ever to play the game.

Kansas City Chiefs: Matt Cassel is 30 and coming off a career-worst 38 score. He posted a 46 the previous season. There’s no reason to believe he will be the quarterback next season, especially with the Chiefs picking first in the upcoming draft. Brady Quinn scored a career-best 41 in relief. He is 28, and hasn’t given people reason to hope for much better.

Oakland Raiders: Carson Palmer just turned 33 and is coming off a 50 score. He was dominating in 2005, with a 65 score, until a serious knee injury in the playoffs. He came back from that injury fairly strong, still well above average. But he missed much of the 2008 season with an elbow injury, and he has been a little below average since. The Raiders could be in worse shape, because Palmer certainly is better than nothing. But they have to want to develop a young quarterback at this point.

San Diego Chargers: Philip Rivers is 31 and coming off a 53 score, his lowest since his second season as a starter. Before this season, he had posted four straight seasons as a top-ten quarterback. There’s no reason to believe he won’t return to that level.

Dallas Cowboys: Tony Romo is 32 and just posted a 59 score. He has six seasons of full-time experience, since he had to be “discovered,” going undrafted out of Eastern Illinois. He has never scored below a 55 and is definitely an elite quarterback when healthy.

New York Giants: Eli Manning just turned 32. He scored a 52 this season. In his nine seasons as a starter, he has only scored above a 53 twice – including 2011, when he scored a 60 and earned his second Super Bowl MVP. Performance-wise, he is maybe a little better than average. But you don’t consider benching a guy with two Super Bowl MVPs and a brother who could be the best all-time.

Philadelphia Eagles: Nick Foles just turned 24 and scored a 44 as a rookie. He will probably get the starting job if he has a decent training camp. Michael Vick is 32. It seems like he should be a lot older as the first pick in the draft 12 years ago. But he started as a 21-year-old rookie and posted the best passing season of any 21-year old in the study (46). Then you add what he did with the running game and you can see why people were so excited. However, 50 was as high as he got before his sentence for mistreatment of dogs. When he came back, he posted a 59 in 2010 and a 53 in 2011, but dropped to a 46 this season. The hits have added up for Vick, and I’m not sure he will be on a roster next season.

Washington Redskins. Robert Griffin III will turn 23 in a couple of weeks and posted a 58 as a rookie. Then you add in his record-setting rushing, and you see why Redskins fans are ecstatic. Hopefully, his knee injury isn’t as serious as it looks.

Chicago Bears: Jay Cutler is 29 and hasn’t lived up to the promise he showed early in his career. He scored a career-worst 48 this season. If a new coach in Chicago can’t turn him around, there’s cause for concern.

Detroit Lions: Matthew Stafford is 24 and coming off a 48 score. That’s a little low, but he had a 61 in 2011. As a 21-year-old rookie in 2009, he scored only a 36, then missed most of 2010 with an injury. Because of what he did in 2011, his starting position is secure. But another year like this one and perhaps there will be some worries in Detroit.

Green Bay Packers: Aaron Rodgers is 29, and one of the elite quarterbacks in the league. After a record-setting season in 2011 in which he scored a 73 (second-best all time), he dropped to a merely-excellent 64 – his fourth career season above 60. As long as he’s healthy, there’s no reason to believe he isn’t going remain at this level, if not get a little better.

Minnesota Vikings: Christian Ponder is 24, and scored a 46, which is average for his age. As a rookie in 2011, he scored a below-average 40. He needs to continue to improve to justify his hold on the starting spot.

Atlanta Falcons: Matt Ryan is 27 and coming off a career-best 65 score – second-best in the NFL. He has been over 55 in three of his five seasons and should be entering his prime.

Carolina Panthers: Cam Newton is only 23 and scored a 54 after putting up a 55 in 2011 as a rookie. He also adds a lot with his feet, and the Panthers have reason to believe they’ve found their franchise quarterback.

New Orleans Saints: Drew Brees turned 34 last week. He scored a 59 this season – his ninth straight with a 55 or better. He is definitely one of the elite quarterbacks in the NFL.

Tampa Bay Buccaneers: Josh Freeman is only 24, and has a remarkable 56 career starts for his age. He posted a 41 as a rookie in 2009, then jumped to a 57 in 2010 as the Buccaneers surprisingly narrowly missed the playoffs. But he regressed to a 45 in 2011 and a 48 this past season. While his job status is in no danger, another below-average season will put him on the hot seat. I would not be surprised to see Tampa Bay try and develop another young quarterback.

Arizona Cardinals: Kevin Kolb is 28 and has struggled with injuries and a difficult media situation. Yet he posted a 52 in limited play this season – far better than his competition. He did the same with a 50 in 2011. If he’s healthy, he should be given the starting job and a real opportunity to develop. John Skelton is 24 and scored a 33 this season, a 40 in 2011. Ryan Lindley is a 23-year-old rookie who posted an abysmal 27 this season (all three quarterbacks barely qualified for a season-score this season).

San Francisco 49ers: Colin Kaepernick is 25 and has led the 49ers to the Super Bowl, taking over the starting role after Alex Smith was injured. He scored a 62 this season, and has also done a lot running the ball. Smith is 28 despite having eight years in the league. He was a disaster early in his career, but regained the starting job in 2009, and has posted scores of 50, 51, 55 and 64 since. Kaepernick will undoubtedly be the starter in 2013 and Smith will probably be a free agent. Given Smith’s age and how well he’s done the last two seasons, he will get a big contract in free agency. I can see Jacksonville making this move. The Jets should make this move, but the media situation is so bad in New York that it really can’t happen.

Seattle Seahawks: Russell Wilson started as a 24-year-old rookie and posted a 61, also doing a lot of damage with the running game. I think he should have been the Rookie of the Year, though Griffin had a special season, too, so there’s no serious complaint.

St. Louis Rams: Sam Bradford is 25 and had a 48 score this season. He had a 44 as a rookie in 2010, but dropped to a 37 in his sophomore season. So far, it’s not looking like he’ll have a great career, but he did enough this season to hope for continued improvement.

As a test of my system for rating quarterback performance, I thought it would be a good idea to list the top ten seasons, since 1974, that my computer spit out without knowing anything about the names or the legends. For this list, I’ve averaged all the game scores, including playoffs, for each season. A quarterback must have thrown at least 8 attempts in at least 14 games to be eligible for this list.

1. Joe Montana, 1989, average score 75 (regular-season NFL passer rating of 112.4). Montana earned his third Super Bowl MVP this season, leading the 49ers to a 14-2 record.

2. Ken Stabler, 1976, 73 (103.4). Stabler led the Raiders to a 13-1 record and a Super Bowl victory.

2. Aaron Rodgers, 2011, 73 (122.5 – best all-time). Rodgers set plenty of passing records in leading the Packers to a 15-1 record, but lost to the Super-Bowl champion Giants in the divisional round of the playoffs.

4. Peyton Manning, 2004, 71 (121.1). Manning had his greatest season statistically as the Colts were 12-4 and lost to the Patriots in the divisional round of the playoffs.

4. Drew Brees, 2011, 71 (110.6). Brees led the Saints to a 13-3 record and a thrilling division-round loss to the 49ers in the playoffs, setting the single-season passing yardage record.

6. Steve Young, 1994, 70 (112.8). Young led the 49ers to a 13-3 record and was MVP of the Super Bowl.

6. Steve Young, 1992, 70 (107.0). Young was 14-2 for the 49ers during the regular season, but lost to the Cowboys in the Conference championship.

8. Bert Jones, 1976, 69 (102.5). Jones led the Colts to an 11-3 record, but they lost to the Steelers in the divisional round of the playoffs.

8. Troy Aikman, 1993, 69 (99.0). Aikman led the Cowboys to an 12-4 record and a Super Bowl victory.

8. Kurt Warner, 1999, 69 (109.2). Warner led the Rams and the Greatest Show on Turf to a 13-3 record. He was Super Bowl MVP.

8. Kurt Warner, 2001, 69 (101.4). Warner led the Rams to a 14-2 record, but they lost to the Patriots in the Super Bowl.

It’s interesting to see both Jones and Stabler on the list from 1976. Those two had the highest passer ratings (as the NFL calculates them) for a season in the NFL in the period from 1972-1983. But were those all-time great seasons? Perhaps a reason to exclude the 14-game seasons from 1974-77 from this data set.

The quarterback project is my name for a series of spreadsheets I’m using to learn more about quarterback performance in the NFL. The more I learn, the better I’ll be able to model career development. And, of course, provide a more accurate simulation engine.

In the first post about this project, I made some observations about quarterback height. I concluded that shorter quarterbacks who succeed in the NFL likely are coached or develop ways of compensating for that lack of height, whether it’s through mechanics or movement. While there is a correlation between height and performance, it has disappeared since 1990, and the only observation I can make is that shorter quarterbacks must prove themselves more quickly than taller quarterbacks.

Future posts will focus on career tracks and aging, as well as overall performance.

I wanted to describe the database a little. I started with the 1974 season, because that was the first year when the NFL truly addressed the problem of wide receivers getting mugged on every play. That opened up the passing game in the NFL. I could just as easily have started with 1978, when the five-yard contact rule was established and the NFL went to 16-game schedules. I’m not certain the intervening data (1974-1977) is all that valuable, but since Lynn Swann and Fred Biletnikoff were named Super Bowl MVPs during that time period, I think that’s evidence the wide receiver was coming of age and the data doesn’t detract from the quality of the database.

There have been 22,896 quarterback “appearances” since 1974, ending with yesterday’s conference championships. An appearance consists of one or more pass attempts from someone listed on the roster as a quarterback. In other words, fakes from punters or options from a running back are not included. I track the basic statistics – attempts, completions, yardage, touchdowns and interceptions. Since 2000, I track fumbles. At some point, I may go back and add sacks, if possible, as these are very important and sometimes the responsibility of a quarterback who held the ball too long. With the resurgence of running quarterbacks, I thought about adding running attempts. But it is next to impossible to separate designed runs and scrambles, and my goal is to assess passing, not tell you whether it’s better to have Russell Wilson run the draw or hand it to Marshawn Lynch to run the draw. I believe fumbles, over a longer period, at least mostly addresses the sack issue since almost 20% of all sacks result in a fumble.

Unlike ESPN, I do not try and add the leverage of a particular situation. Every pass is important, whether it happens to be part of a first drive that went nowhere or a fourth-quarter drive that determined the outcome of the game. If a quarterback is suddenly a different person in the fourth quarter, as some “analysts” claim, I don’t see why he belongs on the field in the first quarter. My analysis of fourth-quarter comebacks supports this conclusion. Magic “clutch” quarterbacks are fodder for Hollywood movies. If you want to maximize fourth-quarter comebacks, play for a ten-win team with a lousy defense. Your quarterback will get far more than his share.

The next step I took was to try and come up with a normalized method for giving each quarterback a game score from 0-100 that was equally relevant in 1974 and in 2012. This took some time. Quarterback rating is a neat, understandable formula, but it is not all that useful for comparing performance more than a few years apart. The game evolves. What I finally did was calculate the correlation between the key statistics and winning, then give each statistic a separate coefficient for each year. I based the coefficient on a 10-year rolling average. This is producing a nice, comparable rating. A “perfect” score in each category is based on reaching the 90th percentile for the season amongst quarterbacks earning a decision that year. A zero score is given for falling below the 10th percentile.

A score of 100 is possible, but it hasn’t been reached in a game since 1974. A score of zero is also possible, but I’d imagine a quarterback would be pulled before he’d get there. The best performances of 2012 were scores of 92 – by Alex Smith in a week 5 rout of Buffalo (18-24-303-3-0) and Peyton Manning in a week 17 rout of Kansas City (23-29-304-3-0). Colin Kaepernick’s 16-21-233-1-0 yesterday earned an 85 – the best performance of the playoffs so far. Matt Ryan scored a 75, Joe Flacco a 57 and Tom Brady a 36. An average score this year was 52. The worst score (by far) this season was turned in by Brandon Weeden in his NFL debut this September, a one-point loss to Philadelphia. He was 12-35-118-0-4 with a pair of fumbles and scored a 2 out of 100. This is the worst score in the entire 39-year database.

An earlier version of this metric didn’t using rolling averages to normalize the data. It also attempted to modify performance based on the defense. The problem with defensive modifications is that it creates a metric where the “best” performances are only possible against the best teams. History suggests this isn’t realistic – they might be the most impressive given our knowledge of the game, but you’re still helping your team considerably when you complete a 25-yard pass against a lousy team. Also, since the value of cross-year comparison is through using seasons’ worth of data, the NFL scheduling algorithm mostly eliminates defensive differences. At most, in a normal season, a quarterback will go up against 5% better defenses than another quarterback. In most cases, the difference is below 2%. So the value of addressing defensive performance, with the associated problem of over-attributing “greatness” against a good defense, doesn’t add much over the course of a season.

For 2012, the top ten quarterbacks in the NFL, based on average game score (6 games with 8 or more pass attempts minimum):

1. Peyton Manning 67
2. Matt Ryan 65
3. Aaron Rodgers 64
3. Alex Smith 64
5. Colin Kaepernick 62
6. Tom Brady 61
6. Russell Wilson 61
8. Drew Brees 59
8. Tony Romo 59
10. Robert Griffin III 58

Finally, I wanted to address an email I received about posting on this blog. Right now, it’s broken. My web host changed the directory structure of my web space about a year ago. I haven’t found the time to go through everything and address each place a change is necessary. Since this is a WordPress blog, I’m not certain what needs to be done – I’ve handled most of the cases on the rest of the site. I apologize to those of you who have taken the trouble to write a comment only to find the spam protector doesn’t work. I also found that WordPress now requires a higher PHP level than my web host is using. So I can’t upgrade the blog, which would allow me to use better spam protection that presumably works. If I turn off all the protection that exists, from the logs, I will have several hundred spam comments per day. So removing protection entirely would result in your comments not being seen amidst the spam. If you have something you’d like me to see, please send email to customer support (address in the FAQ). I will see it, unless you’ve been hit by a address-book virus that bombarded customer support in the past, in which case your address might be black-listed. I should probably reset that black list – I think there are only a dozen or so addresses in it anyway, and the address-book virus creators don’t seem to archive what they use.

I’ll confess. I don’t get a lot of hits on this blog, and part of me is curious to see if I can generate any traffic with my upcoming series of posts.

At this moment 15 years ago, I had just made a life-changing decision. I was a programmer working for a large corporation. The job paid well, but it wasn’t very rewarding. I liked most of the people around me, including my manager, but I absolutely hated working in a cubicle. I’m easily distracted, and need a quiet place to be productive. I was 32 years old, and on what I would term a crossroads. Do I continue to do mediocre, uninteresting work for a good paycheck the rest of my life? Or do I indulge myself and try and create something I love, with no security?

Something had happened at work a few months earlier. I was talking to an intern about the company politics. He had asked me why the senior team members were so apathetic toward the job. I explained that the tech boom was ending, and stock options weren’t the road to riches they were in the early ’90s. They were frustrated and had little motivation to work as hard as they once did. If you treat programming as a 9-to-5 job, you really can’t accomplish much. I was overheard by my team leader. Rather than discussing this with me, or even with our manager (who would have laughed it off), he went to the corporate representative who ran our division. I was called into the corner office and asked, “do you want to work here?”

Like most people in that situation, I lied, and sat through a half-hour lecture. I came out of it feeling embarrassed, angry toward my team leader (I immediately asked my manager for a new assignment, which I received when I explained why). And, perhaps this admission speaks badly of me since I continued to accept my paycheck, not very loyal to my company. I think the decision was set in motion that day. I knew that I would never be happy professionally if I didn’t run my own company.

I’ve been “inventing” games since I was a little kid. Deep in the nest of cardboard boxes containing remnants of my past life, there are dice-and-paper football, Olympics, basketball and baseball games. There’s even a cricket game inspired from our family’s frequent trips to London (I was born in England, while my father was on a sabbatical doing research for one of his books). They’re not terribly sophisticated games. No research went into them. They weren’t created for any audience. I played them myself, and created imaginary leagues. There wasn’t quite a Damon Rutherford (son of the incomparable Brock Rutherford, for those who recognize The Universal Baseball Association, Inc. reference). But it occupied a lot of my time, and a big part of who I am is rooted in these imaginary playing fields. I was never a good athlete, despite constant effort. This was my outlet.

So when (and I’ve written about this many times) I had the opportunity to review Clay Dreslough’s Baseball Mogul for Computer Gaming World, I realized that I wanted very much to create one of these games myself. And it was 15 years ago this week that I wrote a short letter to my manager, gave my four-weeks notice, and started preparing to make Solecismic Software my full-time work. On February 20, 1998, I was on my own. I have not sat down in a cubicle since.

Today, I stand at another crossroads. I have not done much in the last five years. My marriage ended in the spring of 2008, just after Solecismic Software turned 10, and money ceased to be a motivator for anything. Fortunately, I had saved most of what I made from Front Office Football, so I’ve been able to live off of those savings and the decreasing sales of my products by reducing any urge I have to buy “stuff” I don’t need. But five years of “finding myself” is really too much. It’s getting old, and I need to redefine what my life means. I’ve been able to work in decent bursts over the last five years. There have been a couple of free updates of Front Office Football – one that was almost the size of a new product in terms of work and value. And there was the card/board game the past few months. I think that was a good product, but not really right for a modern marketplace where the computer is far more efficient.

I’ve continued to research the game of professional football. In fact, I’ve created a pretty thorough database of quarterback performance over the last 40 years. I’ll be reporting some of my findings in this blog in the near future with a series of posts here. I hope some of you will find that entertaining and interesting. I wrote, 15 years ago, that when I started the Front Office Football project, I threw out everything I thought I knew about professional football. I let the numbers speak to me. Today, there’s so much more data out there. It’s time to throw everything out again. That’s what the quarterback project represents.

I still can’t say what the future holds. Now I’m 47, entering what people call middle age. I can’t make any more promises than I made when I was 32. I didn’t know what I could achieve then, and I don’t know any different today. I’ve been tempted over the last five years simply to put my professional and college games in “maintenance” mode and simply put out updates every year. But I don’t think that’s terribly stimulating from an intellectual perspective. It’s just a financial decision. If I do that, then I’ve become the “corporate” identity that was never satisfying for me when I was on the other side. Maybe some day, if I’ve truly created the games that capture everything I want to capture, I can do that, and use those updates to focus on one or two smaller features that enhance the experience. Certainly real prices have declined to the point where customers would feel that’s a fair decision financially and morally. But the games, as they stand now, are not “finished.”

I would like to finish them, or at least take them to the next level from a simulation perspective. The quarterback project is the beginning of that effort. Whether that results in a new Front Office Football this fall or not I can’t say right now. I might take this in an entirely new direction. Or I might build it into the existing framework and simply overhaul as much as I can. I’ve looked into the world of mobile gaming. While I believe this could be a good direction for my products, I don’t mobile-game myself. So I think, instead, I have to build something new and then maybe farm out that development if it’s successful.

I don’t have any solid answers, as I’ve written. I’m at that crossroads, and I’ve decided that I still love Solecismic Software and that there are many new challenges to face. I wish I could tell you what that means, but I can’t. Any more than I could have told you 15 years ago when none of you would have been very interested. I’m glad you care (enough to have read this far). It means a lot to me to realize there are people out there who believe in my vision. But part of why I’m capable of these things means I can’t realistically make promises. And for me to do effective work, I have to let the work come organically, from my love of sports and my love of numbers.

Will there be regular status updates? No. I made a business decision after I released The College Years that hype is bad. That was a good decision at the time, for many reasons. Now that I haven’t put out a new Front Office Football in six years, it’s not so much a business decision as a personal one. I’m simply a private person and I hate disappointing people. I don’t like a lot of attention, frankly. Nor do I want to limit myself or put any type of artificial constraint on my work. So I might comment from time to time. I will definitely say something if I feel for certain that a new Front Office Football is imminent. I can say that my direction right now is firmly behind the professional football game, and I will only return to the college game if I find new success with the main product. Other than that, as always, the future is uncertain and waiting to be discovered.

I like discovery. I hope you do, too.

Drew Brees is six feet tall, plus an eighth of an inch. He is also a quarterback who may very well end up with a bust in Canton some day.

Yet conventional wisdom says Brees is too short to play quarterback in the NFL.

Russell Wilson is 5-10 and five eighths. He was the 75th pick in the 2012 draft, and, despite his team trading for 6-2 Matt Flynn during the off-season, won Seattle’s starting job in training camp and very nearly took the Seahawks to the NFC Championship game. Wilson won’t win Rookie of the Year because Robert Griffin III had gaudier numbers and was drafted second overall. But he’d get my vote.

Players are measured fairly thoroughly at the annual Scouting Combine, so we know their heights. But in college and in the pros prior to the combine, quarterbacks were encouraged to lie a little because height apparently means everything. If you can’t see over tall linemen, you’re doomed to a life of batted-down passes and receivers you can’t see downfield.

Wilson is definitely the shortest quarterback starting today. Brees is the third-shortest, an eighth of an inch taller than Michael Vick (who probably won’t win a starting job next season, if he even makes a roster).

I’ve been taking a long look at quarterbacks lately. It relates to a database I’ve constructed consisting of quarterback performances going back to the 1974 season. Why 1974? Because that’s when the NFL began addressing the issue with defenders climbing all over receivers running passing routes, opening up the forward pass in the NFL. From 1974 through this season, 445 quarterbacks have either started a game or gained a “decision” in a game (replacing the starter before the deciding score takes place). Only one of those 445, Doug Flutie, is listed as 5-10. Five others, including Wilson, are listed at 5-11. The average is a little under 6-3.

Brees has been interviewed quite a bit about his height. Many teams won’t draft a quarterback under 6-3. Brees says he compensates for his size by having a very quick release and making sure he releases the ball at the top of his throwing motion. He also is careful to shift laterally within the pocket to develop sight lanes to his receivers. The results are fairly astonishing. Brees has completed nearly 66% of his pass attempts during his career, which is the best of any quarterback since 1974 who has more than one career decision. He has won more than 100 games, including playoff appearances. He has a Super Bowl MVP trophy. Wilson also has a high completion percentage (64%), but rather than careful lateral shifts he’s simply great at getting out of the pocket to find his receivers.

So does height matter for quarterbacks? Are Brees and Wilson simply anecdotal exceptions to a rule?

To answer this question, I tallied up the statistics produced by NFL quarterbacks, and separated performances by the listed height of the quarterback. Since I found that quarterback height has a .22 correlation with the year the quarterback entered the league, I divided the quarterbacks into two groups – those who entered the NFL before 1990 and those who entered the NFL since.

As a quick measure of performance, I used the NFL’s quarterback rating. While it’s not a perfect measure, it is a fair measure and relatively easy to understand. What I found was interesting. Prior to 1990, there seemed to be a slight advantage for those who 6-3 and taller – they had a quarterback rating of 75.3, compared to 72.4 for those 6-2 and under. The difference was basically about a 2% higher completion percentage and fewer interceptions.

However, this advantage has disappeared in the last 20 years. Those 6-1 and shorter have an 80.5 rating, 6-2: 81.5, 6-3: 77.3, 6-4: 79.6, 6-5 and taller: 81.2. The completion percentage and interception percentage advantages disappeared.

What do I make from this? Better coaching, perhaps. Brees knows what he has to do to compensate for his height. Perhaps shorter quarterbacks are given specific advice early in their careers. Maybe it wasn’t as obvious in the past.

Still, the bias exists. There’s one statistic that hasn’t changed over time. And that’s the average number of career starts per quarterback, by height. The average 6-1-or-shorter quarterback has about 30 career starts. Six-three and taller gets you about 50 career starts. The leash is “shorter” for shorter quarterbacks. Which may mean that there is a performance difference, as poor-performing shorter quarterbacks are removed from the league much more quickly.

I did find that the standard deviation of quarterback rating decreased with quarterbacks 6-3 and taller, so that supports the notion that the less talented shorter quarterbacks are not getting a long look.

So my conclusion is that height is not essential for an NFL quarterback. But shorter quarterbacks will have fewer opportunities, because some teams won’t draft them. And they have a smaller window of opportunity to prove they can run an NFL team effectively if they do get that chance.

Of course, the NFL is a league of copy-cats, so the enormous success of Wilson this season may well change how quarterbacks are perceived in the near future. Is this potentially good news for Jordan Rodgers – the little (6-1) brother of Green Bay’s all-Pro Aaron Rodgers, now a senior at Vanderbilt?

After this weekend’s explosion of scoring, the 2012 NFL season has featured 22.9 points per team per game over 264 games. This is an increase of .7 over last season, and it looks like we’ll have the second-highest scoring season of all-time, just short of the 1948 season, when there were 10 teams, 12 games per team, linemen could catch passes and substitutions were limited to three per game.

The game is definitely changing in favor of the offense. This could be dangerous for the NFL. Scoring has been remarkably consistent over the years. Since the receiver contact rules were revamped in 1978, scoring ranged from 18.9 to 22.2 from 1979-2011. Just four seasons have seen less than 20.3 points per team, and that led to adjustments in the chucking rule in 1994.

Twenty-two points has been reached four times now – all in the last five seasons. The reason could be more innovation with the passing game – the use of fast seam-splitting tight ends, insistence on finding a good quarterback (there’s always a perceived shortage these days). More than one GM has lamented that the days of shutdown cornerbacks appear to have ended. Or it could be that rules protecting “defenseless” quarterbacks and receivers have opened up strategies a little.

I can’t help but think the NFL is concerned. Records were shattered this year. Calvin Johnson nearly caught 2,000 yards in passes – for a 4-12 team. Adrian Peterson almost broke the rushing yardage record. The Patriots almost broke their own league scoring record. When every offensive record is threatened and scoring is so high, it’s simply a different product. Older fans may start tuning the game out because it isn’t the game they grew to love. In turn, they stop exposing their children to the game.

In the past, the NFL has acted quickly to keep scoring in the 20.5 – 21.5 comfort zone. While there’s no open talk of altering the rules, the Competition Committee has to be concerned. I would hope that changes are coming. Especially in light of recent research on concussions, they have to continue to expand protection against hits to the head. That keeps the middle of the field open more often. So there should be a balancing rule change helping the defense. Perhaps the contact rule could be relaxed to ten yards, or there could be a modification to the substitution rules that would limit the effectiveness of no-huddle – which seems like a gimmick in that its sole purpose seems to be forcing the defense to use personnel that’s not suited for the current play.

While the change we’ve seen in scoring over the last 6-7 years doesn’t seem that great on its surface, it is remarkable when you look at NFL history as a whole. We’ve seen what happened to baseball during the Steroid Age. Part of the reason it has caused so much damage to the sport is because offensive records were destroyed as a result. Now there’s obviously no cheating going on with Calvin Johnson and the Lions – he’s an amazing athlete and the Lions are a pass-first team that was playing catch-up ball about half of the time this season. But I think the NFL should take this change quite seriously.

As always (maybe a little late this year because everything is rather simple), I’m running some end-of-season playoff and draft possibilities.

AFC Playoffs

Houston (12-3): 64% 1st seed, 25% 2nd seed, 11% 3rd seed. Holds tiebreakers at 12-4 or 13-3 with Denver due to head-to-head. At Indianapolis next week.

Denver (12-3): 23% 1st seed, 51% 2nd seed, 26% 3rd seed. Hosts Kansas City next week.

New England (11-4): 13% 1st seed, 24% 2nd seed, 37% 3rd seed, 26% 4th seed. Holds all 12-4 tiebreakers with Denver, Houston or both due to head-to-head sweep. Hosts Miami next week.

Baltimore (10-5): 26% 3rd seed, 74% 4th seed. Holds tiebreaker at 10-6 over Cincinnati due to divisional record. Holds tiebreaker at 11-5 over New England due to head-to-head. At Cincinnati next week (a win sets up a potential rematch – this might be the worst game of the week, as Cincinnati has no reason to compete, either).

Indianapolis (10-5): 5th seed. Holds tiebreaker at 10-6 over Cincinnati due to common games. Hosts Houston next week.

Cincinnati (9-6): 6th seed. Hosts Baltimore next week.

All six playoff spots are clinched. The bye weeks are up in the air, with New England taking a bye away from either Houston or Denver with a win plus a loss from one of those teams. Houston has the edge for home field throughout the playoffs. This is definitely the most settled I’ve ever seen a playoff picture with one week to go.

Note that if we expanded to seven AFC playoff teams, we’d be looking at whether Miami could edge Pittsburgh out for the third wild card at 8-8 based on conference record, needing a win at New England. If both teams lost, could 6-9 San Diego sneak through? The 6-9 Jets would be eliminated based on today’s home loss to the Chargers.

So, Roger Goodell, if you are reading this blog (obviously, he isn’t), please think for a minute about whether you really want to expand the playoffs. This is potentially 100 times worse than the 2010 NFC West (Seattle beating out St. Louis in a 7-9 tiebreaker). Plus you’d have to take away bye weeks for the better teams, which means you’d have the good teams throwing out their bland game plans and resting starters as early as week 11. I sincerely hope this was a quickly-deflated trial balloon, because it is a terrible, terrible idea.

NFC Playoffs

Atlanta (13-2): 1st seed. Hosts Tampa Bay next week, has no reason to compete.

Green Bay (11-4): 64% 2nd seed, 36% 3rd seed. At Minnesota next week.

San Francisco (10-4-1): 23% 2nd seed, 51% 3rd seed, 26% 5th seed. Holds the 10-5-1 tiebreaker over Seattle based on common games. Holds the 11-4-1 or the 10-4-2/11-5 tiebreaker over Green Bay based on head-to-head. Hosts Arizona next week.

Seattle (10-5): 13% second seed, 13% 3rd seed, 74% 5th seed. Holds the 11-5 tiebreaker over Green Bay based on (sorry, replacement refs) head-to-head. Holds the 10-6 tiebreakers over both Minnesota and Chicago based on head-to-head. Hosts St. Louis next week.

Dallas (8-7): 40% 4th seed, otherwise out. Holds the 9-7 tiebreaker over Washington based on common games. At Washington next week.

Washington (9-6): 60% 4th seed, 13% 6th seed, otherwise out. Holds the 9-7 tiebreaker over New York Giants based on divisional record. Holds the 9-7 tiebreaker over Minnesota based on head-to-head. Holds the 9-7 tiebreaker over Chicago based on conference record. Hosts Dallas next week.

Minnesota (9-6): 56% 6th seed, otherwise out. Holds the 10-6 and 9-7 tiebreakers over Chicago based on head-to-head. Hosts Green Bay next week.

Chicago (9-6): 26% 6th seed, otherwise out. At Detroit next week.

New York Giants (8-7): 6% 6th seed, otherwise out. Holds the 9-7 tiebreakers over Chicago and Minnesota based on conference record. Loses the 9-7 tiebreaker with Dallas and Washington based on divisional record. Hosts Philadelphia next week.

The NFC is also fairly well settled, but, every team this side of Atlanta has something important on the line next week. Dallas and Washington will play for the NFC East title and the losing team is probably out (Washington has a chance, but would need losses from Minnesota and Chicago as well). The 6th seed is on the line. Minnesota controls it, but has to beat Green Bay, seeking to clinch a bye week, to get it. Otherwise, Minnesota needs losses from Chicago, the Giants and Dallas. Chicago is next, needing a win and a loss from Minnesota. Finally, the Giants would still be alive with a win next week, but would also need losses from Chicago, Minnesota and Dallas.

2013 NFL Draft Position (range, then expected pick in parenthesis).

1-2 (1). Kansas City (2-13). A loss clinches the first pick.
1-2 (2). Jacksonville (2-13). Cannot lose strength-of-schedule over Kansas City.
3-5 (3). Oakland (4-11).
3-7 (4). Philadelphia (4-11).
3-9 (5). Detroit (4-11).
4-10 (8). Buffalo (5-10).
4-10 (6). Cleveland (5-10).
5-12 (10). Tennessee (5-10).
5-13 (7). Arizona (5-10).
6-13 (12). San Diego (6-9).
8-15 (9). New York Jets (6-9).
8-15 (13). Tampa Bay (6-9).
9-16 (11). Carolina (6-9).
11-17 (16). Pittsburgh (7-8).
11-17 (14). Miami (7-8).
13-18 (17). New Orleans (7-8).
14-19 (15). St. Louis (7-7-1).
16-X (18). Dallas (8-7).
17-X (19). New York Giants (8-7).
18-X (X). Washington (8-7). This pick is owned by St. Louis.
19-X (X). Chicago (9-6).
19-X (20). Minnesota (9-6).