Filed under:

# The Thermometer Problem & applying analytics in coaching

A few additional thoughts on how NHL coaches can turn data into results

You are babysitting a particularly clever eight-year-old child. He complains that he is feeling feverish. It's your responsibility to take care of him while his parents are out, and you have no reasons to question his motives, so you find a thermometer in the bathroom and are planning to take his body temperature.

Do you:

a) Hand him the thermometer, show him how to use it, and tell him to come see you in 10 minutes?

or b) Stick the thermometer under his armpit and wait a while for the temperature reading.

Assuming that the child is smart enough not to split the thermometer open and suck up a lethal dose of red mercury (because it kinda looks like candy), you might be indifferent to those two options. And you'd be very wrong.

Leaving things to chance (or worse)

As history has shown, older humans tend to underestimate how smart and resourceful younger humans can be. Middle-aged writers have been complaining about teens and young adults in op-ed pieces ever since anyone's put pen to paper. Kids don't think like we do (it feels weird for me to say this because I'm not particularly old). This confuses and frightens us, so we put up barriers.

If I give that thermometer to that precocious eight year old I mentioned earlier, I'd have no idea what measured body temperature he'd come back to me with.

Maybe he has a test tomorrow and didn't want to study, so he'll stick the thermometer under the heater until it shows 39 degrees centigrade (a fever).

Maybe he wants a hot chocolate with a big marshmallow on top before bedtime, even though his mother told me not to give him any sweets. In that case, he might put the thermometer on his windowsill, then hand it to me while feigning hypothermia symptoms.

Maybe he just wants to mess with me. So he puts the thermometer in the freezer and returns with a displayed body temperature of minus seven.

Or maybe he just doesn't understand how to use the thermometer properly, and comes back with some other distorted reading.

In any case, leaving this kid alone is leaving things to chance. So I'd grit my teeth, sit down with him for 10 minutes, and make sure everything is on the level.

Numbers out of context

I mentioned earlier this week that it's essentially human nature for people to try to manipulate their performance metrics, so clearly this isn't a problem reserved to babysitters and their charges. The issue is amplified when proxies such as body temperature (to estimate general health), steps (to estimate my dad's exercise volume) or Corsi (to estimate a skater's puck possession) are used.

The first contact I've had with my own player stats came in high school. One of our assistant coaches would track very basic things such as goals, assists, plus-minus and penalty minutes. Even at that level, it was interesting to see how individuals reacted to having their on-ice accomplishments boiled down to a row of numbers in a spreadsheet. In my case, it was "not well."

Pierre Dagenais. Photo credit: Getty Images

Back then, I was maybe a middle-six winger who played quite a bit on the powerplay and who didn't have much to offer other than a good finishing skills from the slot. The only reason why I made Varsity after not skating at all and getting kind of fat between the ages of 9 and 13 was because I shot 200 pucks a day during the summer. I didn't have wheels, I didn't have hands, and I certainly didn't have much hockey sense. I was perhaps the only kid ever to watch Pierre Dagenais play for the Canadiens, and think "this is something I should aspire to." (To this day, he remains my favorite Hab).

The coaches basically passed the updated stat sheet around before the first practice of the week, and never really did anything to explain to us what things meant on a deeper level. My scoring numbers were decent given my usage and my low-volume, high-percentage style of shooting, but for most of the season I had the lowest plus-minus rating on the entire team and was never really called into question about what my numbers said about me and how I could go about improving them. That year, I didn't play a lot down the stretch, leading to an especially memorable episode in which a teammate and I sneaked out for Big Macs during a tournament game and returned to eat them two rows up behind our bench.

Looking back on that year, I felt our coaches missed an opportunity to use data to help us become better players and better students of the game. We were one of the strongest schools in the country academically and maybe that would've helped us gain an edge on the ice. The more I learn about hockey, the more I feel like there is to discover, so I rue those types of missed opportunities more than any others.

Of course, the adults we were surrounded with back then were not well-equipped to look at things from a statistical point of view, but at least having an occasional discussion about the numbers and patterns of the game might have corrected some of the "wrong ideas" about hockey that us 15 and 16 year olds held at the time (and we definitely had many of those).

New stats, big implications

As I alluded to last time, I don't believe it's necessary for NHLers to know much, or anything, about advanced stats. But whether they like it or not, that information is freely available to them on databases such as War-on-Ice or even NHL.com, so there's nothing stopping them from taking a look, whether it is out of pure, harmless curiosity, or ahead of a contentious showdown with the coaching staff over ice time and usage.

With that being the case, it becomes absolutely critical for each NHL head coach to be able to sit down with a player and have an informed, positive conversation about what their advanced stats mean, how it relates to the system, and what changes (if any) need to be made. A coach can't afford to be outmaneuvered by his player statistically, just as how the baby-sitter should not think of handing the thermometer over.

For a progressive organization to move forward at all levels, how the data is understood must not be left to chance, and that applies to everyone from the manager down to the healthy scratches. Facts and figures needs to be put into context by a member of the team who already has the trust and respect of the players: the coach.

Hiring backroom analysts or paying for statistical services won't be productive unless they have a direct impact on how coaches speak and teach the game. Only coaches can help manage expectations for a breakout player due to regress, and only coaches can call a slumping player into his office and tell him: "The hell with the media. I know you're doing a terrific job and the numbers prove it."

Jack Han is the Video & Analytics Coordinator for the McGill Martlet Hockey team. He also writes occasionally about the NHL for Habs Eyes on the Prize. You can find him on Twitter or on the ice at McConnell Arena.