History was made when the Green Bay Packers defeated the New York Giants last Sunday. With this victory — the 149th of Brett Favre’s career — Favre set the all-time record for wins by a starting NFL quarterback.
Players are generally taught that there is “no I in team” and we “win or lose together.” Still, in baseball every win and loss is credited to a specific pitcher. And it is a milestone when a pitcher “wins” 300 games in his career or “loses” 20 games in a single season.
In football this is a less common practice. Favre setting the all-time record for wins was not widely noted. One would also have to do a fair amount of research to learn what the won-loss record was for each quarterback in the NFL. In fact, one could argue – which I am about to do — that assigning wins and losses to a quarterback is not entirely warranted.
Brett Favre’s Wins Produced
Quarterbacks do quite a bit in the course of a football game. Quarterbacks gain ground via the pass and run. But they also lose ground from sacks. And as Favre closes in on the all-time interception record, we see that signal callers also turn the ball over via interceptions and fumbles.
Given all that quarterbacks do, perhaps it’s not surprising that we believe these players are the difference between winning and losing. But when we look at the numbers, a different story is told.
In The Wages of Wins we noted the value of yards, plays (passing attempts + rushing attempts + sacks), and turnovers in terms of net points and wins. Table One below repeats what is said in Table 9.3 of the book (found on page 181 of the paperback edition).
Table One: Table 9.3 of The Wages of Wins
From this table you can see that each yard gained is worth 0.08 net points and 0.002 wins. An interception thrown, though, costs a team 2.7 points and 0.078 wins.
With value in hands, let’s turn to the career of Brett Favre. Favre entered the league in 1991. If we look at Yahoo.com we can find – with but one exception — all the statistics one needs to calculate Favre’s QB Score (explained HERE), Net Points, and Wins Produced. The one exception is fumble lost. According to Yahoo.com, Favre did not fumble until 1994. Actually if you look at all players in history at Yahoo.com you will see that no player fumbled before 1994. Since I am pretty sure I remember fumbles happening before 1994, this means that either no one bothered to either keep track of this stat or make this information available to Yahoo.com. Either way, we have a hole to fill in Favre’s record.
From 1994 to 2006, Favre averaged 0.64 fumbles lost per 100 plays. Given this record, we can assume that Favre lost 7.4 fumbles his first three seasons (by the way, if we didn’t make this effort the story being told here would be the same). And with this data in hand, we can now calculate how many net points and wins Favre has produced in his career.
After 243 career games (through last week), Favre has gained 56,929 yards. These yards were gained on 9,262 plays, of which 334.4 (remember, we are estimating a few fumbles) resulted in turnovers.
Put this all together, and we see that Favre has produced 1,620.1 net points in his career. And these points are worth 41.5 Wins Produced. Again, his team has won 149 games in his career. So the analysis of wins suggests that Favre is responsible for 27.9% of his team’s wins.
On the one hand, one NFL player is worth nearly 30% of a team’s wins. So this suggests Favre is pretty important. Of course, on the other hand (and economists always have another hand), Favre – the player we are crediting with each win – is not responsible for 72% of his team’s wins.
Crediting Statistics
And the 28% we are crediting to Favre assumes that he should receive full credit for all his statistics. And that is one assumption I am not comfortable making.
To understand my hesitation, consider baseball for a moment. Hitters bat entirely by themselves, and hence we can feel comfortable assigning responsibility for whatever happens during each at-bat to the hitter. When we look at pitchers – the players we are assigning wins and losses – it’s a different story. Pitchers – as JC Bradbury notes in a forthcoming article in The Journal of Sports Economics (which I need to devote an entire post to in the future) – are not very consistent. Such inconsistency suggests that the stats we assign to pitchers (ERA, for example) are not just about the pitcher.
Like pitchers, except even more so, quarterbacks in the NFL are very inconsistent. And this is because a quarterback does virtually nothing by himself. Eleven players take the field on offense. Eleven players take the field on defense. The outcome of each play often depends upon the performance of most, if not all, of the 22 players on the field. Yes, we track how many passing yards Favre has thrown in his career. And we note his interceptions. But none of these statistics are created by Favre alone. Each completed pass is completed to another player. For that completion to happen, receivers must run their routes and line-man must block. The play itself must be designed and taught by the coaching staff. Even on interceptions, one can look at the performances of line-man and receivers (and perhaps coaching) when assigning blame.
The importance of teammates becomes quite clear when we look at the consistency of quarterbacks across time. Again, I turn to the point I made in the post detailing the New QB Score.
The page detailing the final QB Score rankings for 2006 begins with the following disclaimer: “As we note in the book, statistics in the NFL do not necessarily represent a quarterback’s ability. So these evaluations should not been seen as conclusive evidence that one quarterback is “better” than another.”
Quarterbacks (and running backs) are quite inconsistent across time (unlike basketball players who are quite consistent across time). This point about quarterbacks was made in the following posts:
Consistent Inconsistency in Football
Football Outsiders and QB Score
The Value of Player Statistics in the NFL
In The Wages of Wins we argue that the inconsistency on the gridiron is a reflection of the impact teammates and coaches have on player performance in the NFL. A quarterback might have “good” statistics because he is “good”, his teammates are “good”, and/or he plays in a “good” system. The statistics cannot tell us who is ultimately responsible for the outcome we observe. Hence the primary purpose of tracking statistics – it allows us to assign responsibility for outcomes – is not achieved in football. This observation should always be kept in mind when we look at statistical evaluations of football players.
By the way, of the three sports (baseball, football, and basketball), basketball players post the most consistent statistics across time. And that is why I am quite comfortable assigning credit to individual NBA players for their statistics. Perhaps another post on this topic would be a good idea in the future.
– DJ
Update: Noah noted in the comments that NFL.com has game logs, which do tell us Favre’s fumbles lost for 1991-1993. I estimated that Favre lost 7.4 fumbles in these years. NFL.com tells us that he actually lost 10. With this data incorporated, Favre’s career wins fall from 41.5 to 41.3. Thanks to Noah for alerting me to this data source.
I should also note that Brian, from bbnflstats.com, offers a good comment as well. Hopefully my response is as good.
Wages of Wins Journal 
noah
September 23, 2007
Just an fyi NFL.com has fumbles for favre’s entire career.
http://www.nfl.com/players/brettfavre/gamelogs?id=FAV540222&season=1992
Brian
September 23, 2007
There is a debate about fumbles lost vs. all fumbles for rating teams and players.
Fumble recoveries are almost completely random. Teams and players have never shown an enduring/repeatable ability to recover fumbles more or less often than average.
A player that fumbles basically puts his team at risk of losing possession, and recovery of the fumble is effectively beyond his control. Plus, when a QB fumbles it is almost certainly at least a busted play with the equivalent consequences of a sack. In fact, I believe official NFL records count forced QB fumbles as a sack.
If you’re trying to expain what happened (like past wins), using fumbles lost probably makes sense for the Wins Produced stat. But if you’re really trying to guage a player’s skill or trying to predict future performance, it may make more sense to use all fumbles, lost or not.
Regarding the consistency of NFL QBs, I’ve looked at trying to improve QB passer ratings. Yards After Catch (YAC) are a big part of passing yards, which is usually part the primary metric for grading QBs. But YAC correlates far more strongly to the receiver from year to year, than to the QB. I’ve suggested the use of a new stat called Air Yards, that subtracts the YAC from the QB’s passing yards.
Similar to the DIPS concept in baseball pitching, Air Yards sacrifices some information for data that better isolates a player’s individual contribution and is accordingly more predictive of future performance.
Farve, in particular, benefitted in 2006 from an unusually large amount of YAC from his receivers. I would suggest Brett should not get credit for the 2-yd screen passes to Ahman Green who scrambles 50+ yds for a TD.
dberri
September 23, 2007
Thanks Noah for the tip.
Brian,
My analysis of football suggests that turnovers in general are fairly random. This is certainly true for interceptions. So my first question is, how consistent are fumbles per play across time for quarterbacks?
And my second question…how much data do you have on YAC? It would be interesting to incorporate that into the analysis. I have thought it would be good to credit the QB with the throw and the receiver with everything after that. This way there is no double counting of yards. But the stat sites I look at don’t give such data.
Thanks for the comments. Your recent analysis of field goal kickers looks pretty interesting.
Brian
September 24, 2007
YAC Data is really hard to find, especially for QBs. Stats Inc keeps it, but does not publish it on its own.
I found receiver data from several recent years at the SI/CNN website, and QB YAC for 2006 on the myway.com NFL stats page. But now that the 2007 season has begun, ’07 data has superceded ’06 data.
A lot of people believe the QB contributes to YAC by leading the receiver, or by chosing to throw to the receiver who he thinks can get the most YAC. But QB YAC does not correlate at all with other QB metrics. Also, YAC correlates far stronger to receivers (about 0.8) than to the QB (about 0.3-0.4). I have lots more on this at my site. Click around the Air Yards and Passer Rating articles.
Rod
December 19, 2007
The very way QBs are “graded” is very dependant on their recievers. You can have the best QB in the world, as far as arm strength, accuracy, poise in the pocket, but if his recievers drop the ball every time or worse yet bat the ball up for an easy INT, it is reflected on the QB.
I’ll propose this, put Randy Moss on the Packers this year and see how many more TDs and less INTs Farve would have. I’m guessing it would be more than a 5 pt swing(5 more TDs and 5 fewer INTs)
Bottom line is you can’t rank QBs by the stats currently used.
It’s no mystery that a QB is the most valued player on the field and are responsible for ~25% of wether you win or lose. They get the ball in their hand every offensive snap. If teams get behind they rely almost 100% on the passing game to get back into a game.
YAC is another big area of fuzziness for QBs. I mean is it really the QBs fault if someone grabs a 2 yard shovel pass and takes it for a long TD? I’ll say yes and no. yes, it is a forward pass and a reception was made. no, it was basically a forward hand off and the runner made the play happen. Either yes or no, the stat goes to the QB and makes him look good, eventhough it was a forward hand off instead of a backwards hand off.
Passing stats would be much more accurate if it was a 23 yard completion and a 44 yard rush after the catch, but let’s be real about it, the stat would be much harder to calculate. Passing stats could look like this at the end of games
12/21 45 yards
Now lets get to INTs. How many times have you seen a reciever run the wrong route and a QB throws the ball where he was suppose to and the only thing waiting for the ball is a safety? I’d say a lot of the time. Who is at fault? The QB or the WR? Well the QB gets the blame, but it really was the WR. Then you have things like WRs trying to get balls and end up batting the ball up in the air to a defender. But again those INTs are put into the QBs stats.
Turn overs are never a good thing, but some turn overs really don’t hurt you as bad as others. Take the pick 6, those really hurt. Take the INT in the end zone, those really hurt, cause at the very least you lost out on a FG attempt. Take a 3rd and 10 at mid field and someone picks you off at the 20(long pass) and tackled immediately. Those don’t really hurt you, cause a punt would have done about the same thing. Of course granted that your defense is at least competent.
Football is a team sport and pouring accolades on individuals does the game more dis-service than recognized. When players realize they made it to the pro-bowl cause of their teammates and are completely self-less, that is a thing of beauty.