Kurt Warner and fumbles.

Okay, so when you compare the number of times a quarterback fumbles to the number of times he was actually hit, Kurt Warner is in a class by himself, and not in a good way. The numbers are so high, in fact, that it's hard to imagine that they're simply the result of a player who is on the extreme end of the crappy side of the normal curve when it comes to hanging onto a football.

If you break down Warner's Fumble Rate by Hit (which I defined two posts below, and will call FRH from this point on) by each year he played, it goes:
1999 - 17.31
2000 - 10.53
2001 - 15.15
2002 - 27.59
2003 - 85.71
2004 - 23.08
2005 - 24.32
2006 - 37.04
2007 - 32.43
2008 - 25.00
2009 - 28.89

Wow. Warner had perfectly normal rates of fumbling, and then BAM. In 2002, his FRH jumps from normal, to higher than any other QB's career rate. That in itself isn't surprising; it could just be a high level of variance. But then, Warner never had another season that wasn't higher than everyone else's career standard. It's stark. Is there any possible explanation for why Warner would start fumbling the football more often? As it happens, there is. In 2002, Warner suffered two separate broken hands, one to the base of his hand late in the season, and a multiple fracture to his pinkie in week 4. He'd broken the pinkie once before, in the 2000 season, but the multiple fracture was considered significantly more severe.

It gets better, in fact. Although I realize that I'm descending even further into the land of small sample sizes, if you break down Warner's 2002 season into pre and post pinkie break segments, you find that he fumbled twice in 15 hits before the break... and 6 times in 14 hits afterward. The FRH rates are 13.33% and 42.86% respectively. Once again, that's stark.

So you have a player who exhibits no signs of fumbling much more often than other quarterbacks, who after a broken finger, starts fumbling like nobody has ever fumbled before. Correlation doesn't prove causation, but it sure seems like a pretty solid theory that Warner's week 4 2002 broken hand had an awful lot to do with his struggles to hang onto the football through the rest of his career.

Of Fumbles and Quarterbacks (Part 5)

With Kurt Warner's recent retirement, there have been any number of retrospectives on his career in the sports media, yet I don't think that I've heard the fumbling issue brought up specifically even once. It's something that has to be dealt with in any kind of accurate analysis of his career, but it doesn't even get mentioned? That's kind of hard to believe.

One interesting point, in the context of other historical developments with regard to fumbles... when I took 32 QBs out of the pool in order to include career sack statistics in the rates, I ran their numbers with just the rushes and receptions added in. (Fumbles/[attempts+rushes+receptions]) Then I did the same with the more recent passers in the main pool to compare.

An interesting historical shift over the last twenty seasons or so has been the decrease in the number of fumbles by running backs. Backs as recently as the 1980s fumbled at rates that would be unheard of today. For example, Tiki Barber was much maligned for much of his career for fumbling too much. Over the course of his career, he fumbled 1.89% of the time on his rushes and receptions. Adrian Peterson is in the same boat, and has coughed it up 2.00% of the time. Compare to Wendell Tyler, who fumbled 4.14% of the time on his rushes and receptions. Or Eric Dickerson, who fumbled 2.38% of the time. Or Earl Campbell at 1.86%. Or Walter Payton, for crying out loud, at 1.99%. Peterson is roundly criticized for fumbling at essentially the same rate as Walter Payton, who was rarely described as a fumbler. Running back fumbles have decreased over time, and dramatically so.

Quarterbacks, on the other hand, appear to be going the opposite direction. The 32 older QBs scored at an average of 1.60. Using the same formula, the more recent QBs scored an average of 1.71. Is this down to an expanded use of the shotgun? Bigger defensive players who hit harder? Defensive players going for the sack more often? A loss of fundamentals in protecting the football? I don't know, but the numbers, while not huge, appear to be real.

EDIT LONG AFTER THE FACT: It appears that this shift is the result of a bookkeeping change. Fumbles on handoffs used to be charged to running backs, and are now more sensibly charged to quarterbacks, who after all, were the last people to actually have control of the ball. I'm still trying to find out when this shift happened, though I'm betting sometime in the early to mid 1980s.

Of Fumbles and Quarterbacks (Part 4)

So, David Carr fumbled a lot. Not exactly shocking. As his apologists probably said any number of times during the years he was starting in Houston, of course he fumbled a lot, he was getting killed back there. And so Peyton Manning doesn't fumble much? Again, big surprise. Peyton gets hit only slightly more often than his backups. It's all well and good to use pass attempts as a way of getting a bead on playing time, but when you're looking at fumbles, 99.9% of pass attempts are by definition not potential chances for the quarterback to fumble. We can get a pretty good idea of how often a quarterback actually fumbled in games with the numbers we looked at before. Can we get an idea of how likely these guys are to hold onto the football when they actually do get hit? We can try.

Quarterback fumbles, as far as I can figure, break down into the following categories:
1) Botched center/quarterback exchanges, always charged as fumbles to the quarterback.
2) Botched handoffs, charged as fumbles to the quarterback.
3) Rushing attempts by the QB.
4) Sacks. (Including defensive strips of the QB, which are also sacks by definition.)
5) Plays where the QB is credited with a reception, either by design, or a batted pass from a shotgun formation.
6) Plays where the QB first recovers a fumble, and then fumbles himself.
7) Garo Yepremian style "what the hell was that?" debacles.

The only ones of these which could possibly result in a QB being credited with a pass attempt on a play where he also fumbled are numbers 5 and 6, and these are rare enough that it's far better to ignore them than try to account for them by using pass attempts. So, thankfully, are Garo-impersonations.

We have numbers for rushes, receptions and sacks. These plays are easy to account for. We'll ignore fumble recoveries and GYDs. That leaves us with center exchanges and handoffs. No QB fumbles on a high percentage of these or he wouldn't have had a job in high school, let alone the NFL, though I don't doubt that some do drop the ball more than others. But because these plays are so common, it's not exactly rare for fumbles to happen in these situations. Unless we go back and look at every individual fumble and sort out where it happened, we're kind of stuck. Rushes+sacks+receptions is going to give us a good estimation of how often a passer gets hit with the ball in his hands, but as a de facto estimation of "opportunities to fumble", it leaves a lot to be desired. It's going to be really friendly towards running quarterbacks, for one thing, because their exchange fumbles are going to get swallowed up by their high numbers of rushes. Still, it should yield some interesting results. So, Fumbles/(Receptions+Rushes+Sacks) = Fumble Rate by Hit.

Some of the interesting results... Kordell Stewart, one of the heaviest runners on the list, goes from below average before sacks and rushes are added, to above average after, to #3 on the list of 94 guys when you judge him per hit. As I said, this method is going to be kind to the running QBs, but it's pretty obvious that Kordell actually did a pretty good job of hanging onto the ball when he got hit. Dan Marino goes the opposite way. Not among the leaders in the other methods, he nevertheless had good scores. However, because he rarely ran and rarely got sacked, the fact that he put the ball on the ground a pretty high percentage of the time when he did get hit.

However, there are two players who really stand out on this list. Bert Jones has held the top spot no matter how we choose to compile the list. His 16 career fumbles are just ridiculous. He comes in at 3.34% on this list, almost two full percentage points ahead of the #2 player, Jim McMahon, and just over one quarter of the average of 12.37%.

The other standout, at the other end of the spectrum, is Kurt Warner. This won't come as a surprise to anyone who has followed his career, but Warner's quick release and avoidance of scrambling have disguised his numbers a bit until now. He isn't just last on the list, he's last by a difference higher than Bert Jones's score. He comes in at 23.96%, while Mark Rypien comes in at 20.09%. Yes, these numbers may be unfair to pocket passers, and yes, most of the guys at the bottom of the list could be described as "statues", but Warner's in another league when it comes to dropping the football.

Of Fumbles and Quarterbacks (part 3)

To be more fair to the running quarterbacks, we can add rush attempts to the pass attempts to get a better read on playing time, but while we're doing that, we might as well throw in two other categories: pass receptions and sacks. The first is going to be of minute importance in most cases, but in a few others, as with Billy Kilmer or Kordell Stewart, they're an important part of the overall picture. Sacks, well, they're pretty important too, particularly since so many quarterback fumbles come on sacks. More on this later. This does create a problem, though. Since the NFL only started officially keeping sack stats in 1969, any passers whose careers took place before 1969 are excluded from our pool. This drops us to 94 quarterbacks being studied.

The numbers don't actually change that much, actually. Dave Krieg replaces Don Meredith in the bottom 5 because Meredith is an older player, and Jim McMahon replaces Joe Namath in the top group for the same reason. The distribution is a bit more skewed now, with a longer right tail ending with Bert Jones. The running quarterbacks do benefit, of course, the most notable being Kordell Stewart, who moves from slightly below average to well above. And of course, the numbers are centered around a lower mean, 1.60% instead of 1.81.

What kind of difference are we talking about here? Well, based on a season with 500 attempts, a figure that is often but not always reached in a 16 game season, the guys at the bottom of the list would fumble 10-12 times. The guys at the top would fumble 3-5 times. Cut that in half for fumbles recovered by the offense, and you've got several extra turnovers a year for the guys at the bottom of the list compared to the guys at the top. Significant, but not overwhelming. Three INTs would equate to 2.5 points of passer rating over 500 attempts, so that's approximately the kind of difference we're talking about with these guys' careers. And that's the difference between the very best and the very worst. Compare the extremes to the middle, and the difference might be put into better context.

Of course, there's another side to look at as well. Looking at the players at both extremes, you might begin to suspect something about the reason these particular quarterbacks find themselves pushed out to the edges of the graphs. More on that in part 4.

Of Fumbles and Quarterbacks (part 2)

What I did to start was take every passer in NFL history who threw more than 2000 passes in his career. This gave me 126 quarterbacks, the guys who have (arguably) played the position the most. Obviously, you can't simply look at their raw numbers of fumbles... 64 fumbles by Jay Schroeder isn't the same as 64 fumbles by Jim Everett, who played roughly twice as much in his career. So, using pass attempts as an analog for playing time, I divided fumbles by attempts to get a working rate.

This resulted in a fairly normal distribution (a bell curve) around 1.81%. The average quarterback in the pool fumbled 1.81 times for every 100 attempts. The standouts? There were five quarterbacks under 1.00: Joey Harrington, Joe Montana, Joe Namath, Peyton Manning, and Bert Jones. There were also five above 3.00: Rodney Peete, Don Meredith, Tony Banks, David Carr, and Daunte Culpepper.

The top group looks like we're playing a game of "which one of these things is not like the others", doesn't it? (Some younger fans might not remember Bert Jones all that well, but he did win an MVP award.) More importantly, there's a huge difference between the top and bottom of the list... Daunte Culpepper fumbles once per 31.36 attempts; Bert Jones dropped the ball once every 174.88 attempts. That's an enormous gap, and clearly meaningful.

However, this isn't precisely fair, is it? Running quarterbacks are getting the shaft, since they will as a group have far more rushes than pocket passers, and this quick and dirty method looks pretty dirty as a result. So we'll keep working.

Of Fumbles and Quarterbacks

The evaluation of quarterbacks begins with passing. That's as it should be. Too often, though, it also ends there. That's not as it should be at all.

Sure, when discussing a quarterback known for his running, that will be brought up, and leadership will be discussed sometimes, though that's a tricky one, since it's hard to agree on how to measure an intangible.

Fumbles? They don't come up all that often. Occasionally, you'll hear it mentioned about a Quarterback with a particular reputation for fumbling, like Dave Krieg. But how often have you heard it said about a QB, "He was good at avoiding fumbles"? I'm willing to bet on never.

Thing is, though, we all know turnovers are bad. You'd never evaluate any quarterback's career numbers without looking at his interceptions, right? But I'll guarantee that people regularly miss the fumbles. Looking at Krieg, (admittedly an extreme example) he threw 199 career INTs. He fumbled 153 times. Obviously, not all those fumbles were lost; Krieg himself recovered 47 fumbles, most of them likely his own. Still, assuming something like half of those fumbles were lost, that's an awful lot of turnovers to ignore, and if you look at his passer rating, that's exactly what you're doing.

I believe most of the lack of attention to fumbles comes from two factors. First, that the numbers haven't been readily available. A few years ago, I was still resorting to print materials to find fumble data, and currently, fumbles lost still aren't easily available for retired players. Second, from an assumption that fumbles tend to even out over time, that they're more the result of external factors than any innate ability on the part of a quarterback. The first issue isn't a good excuse to avoid addressing the issue, and the second... well, it only works if it's true.

I looked at the NFL's top passers to see just how true or false it is. Results will be reported in the following posts.

Jay Cutler is the most talented quarterback I have ever worked with.

How long before Mike Martz is quoted as saying this?

I'm betting some version of it will be trotted out during or immediately after the Bears' first mini-camp of the offseason.