clock menu more-arrow no yes mobile

Filed under:

Stu Hackel Week on Arctic Ice Hockey: Bringing it all together

Getty Images

I just wanted to summarize and point you to all of the pieces in our ten-part series on potential NHL statistics...

Stu Hackel hit on two of the most important items in the NHL Advanced Statistics toolbox - possession and faceoff zones. Of course, these stats are already recorded at numerous sites, so it's not clear to me how we would benefit if the NHL published the same numbers. Indeed, the NHL publishes numerous official statistics that are quite obviously wrong (e.g - turnovers, shot data at Madison Square Garden), so the official imprimatur might not be worth anything if the league didn't care about the accuracy of the data.

Go to the front page of, or and take a look at the scant statistical offerings - I think the lesson from other leagues is that we should not look to the NHL head office for top-down analytical and statistical innovation. And why should we? The purpose of all these fancy stats is to better predict future team and player performance. The NHL itself is not involved in any such enterprise - its only true goal is to make sure that the correct number of goals are counted for each team in every game. The recording of everything else - faceoffs, penalties, shots, turnovers and their locations - is left to the whims of individual arena crews, which are even further removed from the need for accurate data to drive scouting decisions.

In other sports, private organizations independent of the league have driven the development of new statistics in no small part because the league itself is ill-equipped to decide what statistics represent true talents. You can see this problem in Hackel's requests: he'd like to see 5- and 10-game special teams stats, hit posts, individual shot attempts blocked and head-to-head faceoff matchups, and he's convinced that there are NHL teams that consistently win hockey games despite failing to dominate puck possession. There's no evidence that any of these items represent repeatable talents; as a result, they haven't caught on among analysts. Not to get all starry-eyed like it's 1998, but this is the great strength of the internet analysis community - good ideas bubble up to the top while bad ones get rejected because they don't bring any value to the table (shot quality being a notable exception.)

Overall, I'd focus Hackel's list down to one item: tracking zone and possession time. While it may not ultimately turn out to be more valuable than shot counts, I'm guessing that it is, especially if we could get zone entries and exits tracked as play-by-play events. The NHL is reasonably good at recording on-ice events, and we'll get the most benefit once the league starts tracking every touch and every player's location at all times.