clock menu more-arrow no yes

Filed under:

Another Quick Thought From Kid Ish

These are things I thought that I am sharing.

At the end of this month, will be adding new (advanced) statistics to its website. According to a Bob McKenzie report, the league will add 35 new stats, including visualization tools similar to what's out there on, etc. In a Sean Gentille Sporting News post this came out:

"One issue according to McKenzie - and confirmed, whatever that's worth, by Sporting News - has been what, exactly, to call them. Are we going with Corsi, the internet-borne name for 5-on-5 shot attempts, or will it just be "5-on-5 shot attempts?" Fenwick, or unblocked 5-on-5 shot attempts?"

The new stats on the site will have new names, because while their naming is indeed an issue, it seems the older names have already been looked over. Let's say I'm 99% certain on that one, too. Obviously anything can happen, but given the chance to create its own stat names and abbreviations, it makes business sense for the league to do so. For what it's worth, the new names will certainly be more descriptive of what each stat really is - including PDO, which will actually have a meaning.

None of this is new to many of you, but it is as good a time as any to remind readers why these stats are valuable, even in the simplest sense. I'm not going to get into all the millions of words put into this space before, justifying why any of this matters. Clearly, it all matters, since the NHL is adding them as stats to its site.

But that doesn't mean we can't continue to make them simple for people who just don't want to get into numbers and predictive outlooks and stuff like that. Sports are fun, let's have fun.


With the NHL moving to microchips on jerseys and pucks, the idea that there will be accurate zone time numbers in the future is looking up! That's great news. But aside from the kinks in working out what all that means - is all time in the offensive zone considered time on attack, do we account all touches after the first (regardless of zone) as positive possession, etc. - the main issue with this going forward is that the NHL may not reveal this data to fans. At least at first.

But this is ok, since the league will show advanced stats on the site. As I've said before, shot attempts are an excellent proxy of time on attack (positive possession). So while we wait to see how tracking data is released to the public, we can still approximate things based on shot attempts should we not get anything.

But let's say we do get those numbers from the chips. Will shot attempts matter then? I think so, yes.

On a team level, there's still nuance to tease out between shots attempted and shots on. When added with shot location data, which is already available without chips, the "threat level" of each attempt can be looked at. And those can begin any examination of a team's systems. Indeed, much of the irregularities in the Ducks attack (middling possession-by-attempts but generally higher chance counts) have pointed me to notice what the team does to end up with its results. You've all read me breaking that down here, so clearly there's meaning there.

On an individual level, looking at shot attempts for and against numbers, say broken up by periods of time (per 20 minutes or per 60 minutes), can reveal which players are driving the possession time we track. This again can focus our viewing toward examining why players post the numbers they do. If two players generate similar shot attempt numbers but only player A suppresses them well (defends), there's a valuable difference between the two players, regardless of end results. We can get into those differences and the value player A adds over player B, etc.

When I think about adding shot attempt data to the tracking (via microchips) the NHL may one day release to the public, I get even more excited. The same insights into the sport are there, but the ability to visually chart how it works, whether it is heat maps of a player's movement on the ice or the propensity of the team to start breakouts from the left side goal line, makes figuring it out and showing others an easier task.

Think about our scoring plays stuff in my posts this season. We all love the touch charts and player movements they show. Now think about those for every play, and instead of just the route the puck (or player with the puck) takes, all players can be charted and displayed. Each shot attempt is shown, making the existing location data even easier to translate. "Oh that's how the attempt from the point ended up being a scoring chance two seconds later."


This is the one stat that won't go anywhere, regardless of what future visual (or otherwise) data we can get. This is also the easiest one to utilize in terms of judging a team beyond just wins or losses. I mean, come on, name aside: if your team's shooters scored by shooting higher than the league average, figure it to drop eventually. If your team's goaltender made more saves than the league average, figure it to drop in time. And this is also true going the other way, with percentages that are under the average.

Adding individual player percentages on top of that can add a lot of nuance and meaning, but that can continue being left to people who want to work with all the numbers. For instance, we know that the league shooting average isn't certain elite players' shooting average. The same is true for elite goaltenders, who can regularly post higher-than-league-average save percentages. Working that individual data into team on ice data is more complex, certainly, but not beyond what some websites are doing for you now.

So let's say this stat is renamed as on ice shooting and save percentage, or OISSP. (It won't be that, and I know what it will be with like 99% certainty, but this is hypothetical.) Last year, Player A scored 19 goals on 151 shots at even strength, or a 12.6% shooting percentage. That is higher than average by about four percents-plus. Predictably, Player A has four goals on 74 shots, or 5.4% shooting, this year. The last two years, therefore, Player A is shooting at 9%, which is much closer to league average, but has very wide-ranging year totals to get there.

Now, Player A is not as good as 12.6% but not as bad as 5.4%. He's definitely closer to that 9% average, but teams can't control when that percentage will break for them. Looking at Player A's OISSP number - combining his shooting percentage with the save percentage he faced - reveals that indeed, he's just become unlucky in the second year. No noticeable change in the average SV% faced is evident, but his own fluctuating S% is giving him a minus OISSP - let's say a -3 (or 97 PDO if using current standards). Knowing this, a team doesn't have to hound a player for the 5.4% spells or praise the player too much for the 12.6% ones - there's level ground in recognizing when play is breaking for a guy, when he's "hot," so to speak.

So we can definitely point to a -3 OISSP and say "Player A is just getting so unlucky right now. But he's better than this and can expect some correction to his average gradually." Which is called regression, of course, and should never be taken as an insult but as a fact of reality. (Conversely, when he was shooting 12.6% and was +4 OISSP, we could have said he's not that good and shouldn't be expected to continue scoring at the pace he did.)

This one got a little longer than I thought, but the relative simplicity of the stat and what it tells us about team performance on a given night (or fortnight if tracking it over a time) or about player performance during the same time will benefit most from having a more descriptive, meaningful tag. This is because it will finally have a set name that will make sense, and hopefully that will expand its use. It isn't PDO or just "puck luck," but an actual statistic defined by something more similar to the OISSP I made up above.


Did you know that the stats on the NHL's website are able to be used in contract negotiations as well as arbitrations? This means that, as general managers and player agents begin understanding and incorporating them better, players like Clayton Stoner won't be offered idiot money [as often] unless they provide a very specific upside that's not measurable, like leadershipping charactergrit.

Ok, jokes aside: a smarter league is one that will begin seeing more efficiencies going forward, both in terms of the on ice product and the presentation of it that we're treated with. A more efficient, "professional" sport benefits fans because it takes it out of the hands of the few, who think only they know the sport, and gives it to all. Player B's plus/minus stat will eventually be forgotten because we'll look at (and be treated to) his possession plus/minus, which will better determine his impact on offense/defense. Indeed, even adopting an on ice GF% over plus/minus is better than the original.

In other words, we're getting there. Teams will get smarter. Contracts will get smarter. Broadcasts and print coverage will get smarter. All of this because these numbers will now be accepted and in use from the league itself. While some teams (like the Ducks) won't jump on and use them correctly (or at all) right away, in time every team in the league will have to, because the stats will become the language currency by which we know the game.

And I say this as someone who knows, with absolute surety that, up until this very day, general manager Bob Murray is not a believer in stats. I hope this changes, it would benefit the organization greatly!


...although of course sacrifice bunting still happens in baseball so maybe we're all doomed and I'm excited for nothing! COULD BE.