Does Excessive Hitting Exacerbate NHL Player Performance Decline?

One popular belief among hockey stat nerds is that individual players throwing a higher number of hits can exacerbate statistical decline as they get older. As the body takes more punishment, it’s more likely to break down, which is sound logic. I decided to examine if there was evidence to support this theory, and if there was, then by how much? Players generally don’t start to decline on average until either age 26 or 27. There are clearly a long list of exceptions, but 27 has long been acknowledged as the age of peak performance for most athletes (with exceptions by sport).

For this investigation, I looked at players who had played at least 20 games in one season, and at least 20 in the next season since 2006, and measured their change in average time on ice and points scored per 60 minutes played. For players under the age of 30, there was 0% correlation between number of hits thrown over the previous 3 seasons and rate of statistical decline. Not even a weak correlation exists between hitting and rate stats. Perhaps part of the problem is that my data is lacking hits thrown at the AHL level, so we’re not counting all pro hockey hits, just the ones thrown in the NHL. That data might be needed to precisely quantify the causal relationship between total number of collisions and performance erosion.

When dealing with players over the age of 30, there was some evidence (albeit it a very weak correlation) that hits can affect average time on ice, but it’s only by a handful of seconds per 100 hits. The charts are below. There was surprisingly a positive relationship between total hits and PTS per 60 minutes for players over the age of 32. Perhaps some of that is due to players like Brad Marchand, Alex Ovechkin, Dustin Brown, etc who are the best examples to cite if you’re seeking to debunk the idea that collision totals increase scoring decline.

 

The most plausible explanation as to why my own investigation did not uncover that causal relationship is simply looking at NHL collisions over the previous years. If I’m going to embark on another investigation, it will include total professional hits, both thrown and received, then see if that affects scoring decline. I do not currently have access to that data. I’m still leaving the door open that there is an effect, I was just not able to uncover it in this particular investigation. 

It's also worth pointing out that this is measuring impact on rate stats. Larger collision numbers could have its most dramatic impact on total games played, not just performance per game. If collisions lead to more injuries, then that’s where we should be looking for causation. That does pose some challenges (hence why I did not choose to investigate that parameter in this particular study), especially when it comes to retirement. Many players in their 30s are forced to stop playing because they choose to retire or are simply no longer good enough, yet are perfectly healthy with no injury history. That will be a large subsample of the population that’s difficult to separate from those who are unable to play due to an injury. It will cloud the results.

I’ll often site “hard miles on the odometer” in my summarizing why some contracts do or might go bad. There are many examples where it logically fits (see Jamie Benn, David Clarkson, Milan Lucic, etc). But I’ve also been burned in the past by prematurely forecasting the demise of Brad Marchand and Alex Ovechkin, so I’ve got conflicting emotions on this issue. I’m not closing the door completely on this notion, but was not able to find supporting evidence in this particular investigation.

Since I do not have a complete database of injuries, it's hard to quantify how much a high collision count actually increases the probability of getting injured. We certainly know this to be true in the case of concussions (see Michael Ferland), which is among the most common injuries to emerge from bone crunching hits. Does a higher probability of injury hurt a player's market value? If a manager wants to acquire physical players, he can't possibly do so without taking on some injury risk. Hiring employees to "live by the sword" will inevitably lead to casualties "dying by the sword". It's not optional if you want your team to play that style, which can be effective in a 7-game playoff series.

Injuries are not the end of the world, and can even be good news if it allows you to hide a bad contract on LTIR. Sending David Clarkson to “Robidas Island” was actually advantageous for Toronto, as it helped them hide a mistake. Where you really want to be careful is not giving out too much term to the bangers. That’s where managers can run into the biggest problems. Taking on a player with a higher probability of injury has less risk if it’s on a 1-year term. The players who can get reach high totals in both hits and points are a rare commodity that’s in high demand, so you won’t be able to attain that asset without an enticing offer. It might be smarter to treat your bangers as cheap disposable commodities on short-term deals that you can plug into your line-up. It sounds terrible to say from a humanitarian perspective (they’re human beings after all), but it’s just smart asset management.