September 22, 2014

Outside Foxborough – The Injury Effect, Part 2

logoby Bill Barnwell
[email protected]
 

Note: Last week, I mentioned that I would be discussing the offensive and defensive relationships to the injury data I gathered in this week’s column on Patriots Daily. The day my column went up, Willie Parker went down for the year with a broken leg, and it seemed natural to write a short followup piece on Football Outsiders discussing what the broken leg might do to the Steelers’ chances. In short, I mentioned that running back injuries have a very weak relationship to a decrease in wins or DVOA, and while they have somewhat of a relationship to a team’s rushing DVOA, an injury to a quarterback seems to have a similar relationship to the team’s rushing attack. You can find the full article here.

In the second part of my research into injury data, I looked at how injuries at different positions affected their teams’ performances, to see if it revealed hotspots for potential injury or tipoffs towards future success.

My findings were pretty distinct. I’ll summate them in order of importance.

1. Offensive injuries affect a team much more than defensive injuries.

This is a fascinating thing, to me at least. Of course, you’ll always hear more about an offensive injury because skill position players receive significantly more glory than most, if not all defensive players, but I was surprised to see that actually match up with their real importance as well. A team’s offensive injury rate has a -.37 correlation with their difference in year-to-year wins, while their defensive injury rate only has a -.23 correlation. Offensive injuries have a -.21 correlation with DVOA within a season, and a -.38 correlation with the difference in year-to-year DVOA, while defensive injuries are at -.06 and -.25, respectively. Essentially, what the data says is that offensive players are harder to replace than defensive ones.

2. Offensive line injuries are the most traumatic to a team.

Across all positions, injuries to the offensive line had the strongest correlation (-.30) against year-to-year wins as well as DVOA (-.33). That’s also likely to become stronger this year, when you consider the success of the Cleveland Browns following the return to health of their offensive line, as well as the impact that offensive line injuries have had on teams like the Rams and 49ers.

3. Quarterbacks are the second-strongest correlated.

Quarterback injuries have the second-strongest negative correlation with year-to-year wins (-.24) and DVOA (-.26). Does that mean that quarterbacks are less important than offensive linemen? It’s hard to say. I think there’s something to be said for the positional scarcity of quarterbacks, as well as the ability to hide a subpar offensive lineman with an extra blocking back or a tight end (which does, on the other hand, ignore the consequences that has on the offense as a whole, which come out in these correlations to a team’s overall performance).

It should also be noted, strangely enough, that while offensive line injuries correlate strongest with a drop in overall team DVOA, quarterback injuries (-.31) have a stronger inverse relationship with a drop in offensive DVOA than offensive line injuries (-.19). Whether that points to noise in the data or an effect of a team’s salary cap construction, I can’t say.

4. Linebacker injuries are the most difficult to recover from defensively.

Across the board, defensive back injuries are the least-correlated with success (perhaps owing to the Patriots’ success despite their mammoth infirmary behind the linebackers), and also across the board, linebackers have both the strongest relationship with year-to-year changes in wins (-.21) and DVOA (-.25).

5. Receivers are the most replaceable players in football.

I know, I know. This all sounds very silly when you consider the ’06 and ’07 Patriot offenses. But the data shows that wide receiver injuries have virtually no effect on offensive DVOA (-.02 difference year-to-year). Does that mean that the Patriots would do just as well without Randy Moss? Of course not. No statement is a hard rule. But all-in-all, wideout injuries are probably overspoken. If you think about guys who got hurt this year at wide receiver, have any crushed their team? Terry Glenn’s injury certainly hasn’t. Has Marvin Harrison slowed down the Colts’ offense? A little, but not noticeably. The Broncos passing attack was just as good without Javon Walker, Seattle without Deion Branch, and even arguably Houston without Andre Johnson.

6. Age has very little bearing on injury.

On offense, the average age of a team’s starters has virtually no relationship (+.02) with their propensity for getting injured. On defense, there’s a slight relationship (+.18). Overall, there’s virtually no relationship between the 22 starters’ age (-.04) and the team’s injury rate.

Of course, this research needs more data to munch on before we can assign it more reliability. Six years is a good start, but by this time next year, there will hopefully be twice as many injury reports for it to compare with success. If these findings hold up, the result could be some different guidelines for how to ideally construct a roster and use salary cap space, focusing more on offensive depth than defensive, specifically at offensive line and quarterback. We might be able to more accurately define replacement-level if we understand the propensity of players at a certain position to get hurt in a given season, which could have an economic effect on how those players are valued. For example, if the findings above were true, teams would likely be overvaluing a good portion of their wide receivers, whose absence from the team would have little effect on their team’s offensive performance.

Again, more data’s needed, but these findings are a very interesting first look into the effects on injuries across a whole team, and how success sometimes has more to do with random luck than some teams or media members would like to admit.

Comments

  1. What happens if you put in a squared term on the injury score and run a regression accordingly? I can imagine that the relationship between injuries and performance (wins or DVOA) is stronger for teams at the extremes of injuries.

Leave a Reply