Football Injuries: The Metric That Matters

It’s football season in America again. Hallelujah.

As bad as most prognosticators are about most things, football prognosticators are really bad. Go back and look at just about any group of experts’ predictions for the coming season and you’ll see that their success rate is lower than that of the average monkey with a dartboard.

But for anyone looking to vindicate their terrible predictions, help has arrived: this article by Bill Barnwell of It is a great look at a long-overlooked topic: the importance of injury on a given football team’s win-loss record. Using a metric called History-Adjusted Games Lost, Barnwell gives us an empirical look at the impact of various injuries on a team and a variety of nuances. If you are at all interested in the topic, you should read the entire article, but here are a few of Barnwell’s important conclusions:

  • “Injuries, or their absence, have a drastic effect on a team’s success.”
  • “A superstar is worth about six to seven times as much as a reserve.”
  • “Offensive injuries are more significant than defensive ones.”
  • “Injuries to the starting halfback do not affect the running game.”

So how do these conclusions help the case of bad football prognosticators?

Because football injuries are mostly the result of bad luck. Yes, you could argue that better players get injured less because they’re better prepared, better conditioned, and less likely to be caught out of position. But my sense is that these arguments may hold truer in a sport like baseball or soccer than in football, which is a game of collisions. If you agree that injuries are mostly the result of luck, then you could argue that many predictions go astray because the predictors couldn’t foresee the bad luck coming — but that their picks are otherwise sound.

If you run a football team and buy Barnwell’s analysis, then you’d probably want to spend a lot more time thinking about how injuries affect your season. Statheads have long grumbled that football has been slow to embrace a Sabermetric approach to the game (see Bill James here, on baseball; see Mike Zarren here, on basketball). It seems that football executives and coaches haven’t yet been convinced that certain statistical analyses provide the appropriate value and insight. But I can see how the kind of insights on injury that Barnwell offers might start to turn that tide. Even if a particular injury is the result of nothing more than bad luck, there are still lots of questions to empirically explore, such as:

  • What is the optimal approach for a coaching and training staff to minimize injuries?
  • What is the proper tradeoff between having very good and more expensive backups (an insurance policy against key injuries) and splurging for superstar starters?
  • What is the best way to assess how a player’s previous injuries hurt his future abilities?

C’mon, NFL owners. There are a lot of newly minted statisticians out there. Surely a few of them are football fans who’d probably work cheap for a chance to bring some injury science to your business.


My sense is that NFL owners are attempting to take such an approach. One of my colleagues (I'm a faculty member in statistics department at a large university) was recently offered a lucrative job with a prominent NFL team. He turned the job down, but he did note that the NFL is (or at least has been) behind the curve in terms of data collection.

To make the sorts of assessments that have been made in baseball and basketball, you need individual-level data on every single play of every single game and not just what play was run, what type of defense was faced, how many yards for that play, etc. As you can imagine, this sort of data is very difficult to obtain (i.e., assessing how each player performed on each play). Moreover, far fewer games are played each season in football than in basketball or baseball, contributing to the dearth of data.

Teams have started to collect data, but I wonder if it will be awhile before we start to see dramatic effects from that data collection.



Actually injuries have remained a remarkably resistant problem in all sports. Assessing the injury questions you raise is something that all professional sports teams try to do. Since I follow the RS I know that they- (a) hired Bill James and other sabermetricians to assess these probabilities and make recommendations, (b) hired the trainer who had worked with Dr. James Andrews with extensive experience with baseball injuries and designed a specific arm strengthening program for each pitcher, and (c) frequently sign injured players to variable value deals so that the player and team share the costs and benefits of their recovery. Still the differences are so small and luck is so important that the years when they had the fewest players lost to injury- 2004 and 2007, they won the WS, while each time the year after there were key injuries (especially to the pitching staff) that prevented them from repeating.

I suspect that the answer is to get superstars at key positions then realize that if they stay healthy you can win a championship, and if they get injured you can't. Shoot for getting those superstars under contract for a few years together, surround them with average to above-average players and hope that one of those years everybody stays healthy. Assuming that is that the goal is to win a championship in the sport, which may not be the case for every owner.



I played football in high school and have watched it closely since then.

Injuries basically happen when people get tired, so I think that being in great cardiovascular condition would be the best way to stay injury free.

When you get tired..

1. you stop moving your feet (biggest cause of lower extremity injuries)

2. you may let yourself get out of position

3. you will be taking hits, not absorbing them

I wonder if my thesis could be tested by tracking if injuries increase in frequency as the game wears on.


I think the person who figures out how to prevent offensive line injuries will be the most valuable to an organization - but a lot of those injures seem to be the ones where someone rolls over a leg in a funny, awkward position.

the Gooch


Even at the college level, teams grade out the performance of every individual on every play. All that data stays internal to the team, and may not be even stored electronically.


As far as offensive line injuries go, note that every lineman in the college and pros now wears knee braces as prophylactics.

Mike M

Also, football is a very situational sport, which makes data difficult to interpret. For example, giving up 8 yards on first or second down is never a good thing, but on 3rd and 15 it is totally acceptable.

Statistics for individuals are also difficult to interpret. A star defensive tackle may be double and triple teamed at various points throughout a game, which could enable a less talented player to make more plays.

I do agree that most teams would benefit from a more statistical approach to the game, but I don't know how far you will get breaking it down to individual players and positions.


Has anyone done a study on the effects of drugs and or injuries to athletes and or birth defects in their children?

Could it be that steroids are making their mucles to large for their bone structure and thus causing injuries. Doesn't it seem that a large portion of professional atheltes have kids with issues. More so than the general population? Or is this just not there and it's the notoriety that make it seem that way???


Let's not forget that in football, as compared to most sports other than boxing, injuies are frequently inflicted on purpose. The bigger the superstar, the more likely it is the opposing team wil be "out to get him".

Travis L

Will Carroll at has pretty much made it his mission in life to build an injury database in order to do this in baseball. You can't really have a conversation about injury analysis without mentioning him.


If your goal is to win championships, then you should never go for depth at the expense of superstars. If you goal is to simply have a good shot at a winning record and a decent shot at the playoffs every year, then perhaps a deep roster of average talent will get you there.


I'd guess that superstars play more minutes per game than average players, leading to a higher risk of injury. The more times you're tackled, the greater the chance of getting hurt. No one gets hurt sitting on the bench.

As #8 points out, superstars can make the difference in a game, so your opponents will concentrate on stopping you. A superstar might be tackled by three guys instead of one.

Likewise, the longer you play, the more fatigued you get, as #3 mentions. For example, you might plant your foot incorrectly as you tire, leading to greater injuries.

As #7 notes, performance enhancing drugs make muscles bigger & stronger. As far as I know, drugs don't make tendons & ligaments stronger, so they can tear or rupture. IOW, the muscles get too strong for the tendons & ligaments to handle, particularly the longer you play/the longer the season progresses.


While I agree with Dan's comment #10 in theory, (the principle that superstars are needed for championships), in principle an emphasis on getting star players will probably lead to financial ruin and put an underwhelming team on the field.

If you instead look for players that are undervalued or fit into your system, you may find some players who develop into superstars.

I think that most of the winningest teams in NFL history benefited from mature, smart, and restrained free agent pickups, and also relied on 2nd-7th round drafts picks and player development.

David Isenbergh

These are interesting and difficult questions.

One factor not mentioned in injury prevention is the factor of youth. Young players (everything else being equal) are significantly less prone to injuries than aging players (including aging super stars). So building a team around a core of young players often makes more sense than bidding for a load of established free-agents.

And let's not forget that other factors can be just as weighty as the sum of individual talents and durabilities: players, like members of orchestras or business associations, mesh well with some teammates and not others. Team chemistry is hugely important. The best teams are often greater than the sum of their parts (i.e., players).

Finally, the head coach and his staff cannot be overlooked: some coaches are more inspired than others when it comes to play-calling. And of course preparing the players physically and mentally for the next game, and injecting approprate confidence, is no small matter.

Thus teams like the Washington Redskins of the 70's and 80's, not egregiously blessed with either youth or talent, were perennial contenders. Likewise, the Pittsburgh Steelers of our day, while not conspicuous for talent or freedom from injuries, have been on balance, arguably, the best team of the decade.


Johnny E

People that got to the pro level probably are less prone to injury just because they had survived years of playing to get there. All the weaker bodies were selected out already. Their bodies must be inherently stronger besides the motivated conditioning they go through. If there is a way to indentify those players early you might be on to something.


To answer the questions posed at end of the article one only has to look at the New England Patriots.


First of all, we're all glad to have football season back so we can laze around on Sundays rather than doing any other more productive things. Of course we'll argue that this is the most productive thing we can be doing, and I assure anybody who doubts it, that it is true.

The evidence that a football player's injury affecting his team is just overwhelming. The chemistry of the team is not as great, and of course, skill is not as great. I agree that injuries are just a result of bad luck, and that the NFL can't really do anything else to prevent these. Anything avoidable has been patched up with the proper protection gear, and any other protection would just affect the players' movement too much. As to where should the point of skill between starter and reserve, is up to the team to decide. All teams have different budgets so all teams handle themselves as best as they consider right. The NFL's business is already great, I doubt any changes will bring to it any major economic advantage (or more profit) than what it currently is.


Tommy K

Ragrding Sabermetrics and the "Moneyball" approach, though he refuses to discuss it himself, it is known that Bill Belichick of the Patriots (otherwise known as the team of the decade) relies strongly on numerical analysis. This is according to people who have left the team.

It was this approach that led him to his most important realization - that this nobody second-string QB named Tom Brady was ten times the quarterback his highly-paid superstar Drew Bledsoe was.

James Roane

# 10 has it wrong. In a salary capped sport where injuries are the norm, you need a star at some key positions, QB being one on offense and a dominate game changer somewhere on defense. After that you need lots of skilled players but not necessarily "star" players. Look, NE, Pitts, and Indy (to a lesser degree) all have had pretty much the same formula. Star QB, a couple of game changers on defense, and a deep squad of skilled but fairly priced players. You've got to have the depth to keep players fresh and have adequate replacements for injuries. That said, an injury to the QB or the Def game changer and a championship is out the window.

Midway Monster

Not enough games in football to use statistics with any useful confidence level.

I suppose that if one brought the statistics down to the 'snap' granularity, then one might have something to work with. But game-by-game in football is too macro, and you, Phd Freakonomics, should know that. The average football player's career is what...3 seasons? That's fewer games than a basketball season. Even a seasoned veteran football player (10 seasons) is less than a single baseball year.

Oh, I weep for the future of μάθημα. (math, for those that don't read ancient greek).


The first statement in this post is intriguing, "As bad as most prognosticators are about most things, football prognosticators are really bad. Go back and look at just about any group of experts' predictions for the coming season and you'll see that their success rate is lower than that of the average monkey with a dartboard."

If it is true that prognosticators is statistically are worse than random (the monkey analogy) then I think I know how to make some money. It doesn't matter which direction the variation from random is for it to be useful.