Colorado Avalanche v San Jose Sharks

For the advanced stat community, this post is entirely entry-level thinking, but outside of that annoying niche group of hockey fans (ugh, they’re just the WORST, those guys), there’s a large misconception about how to best understand the simplest numbers we pile up over the course of a hockey game. And, there’s no reason to feel ashamed about that. I’ve played for some pretty decent coaches who weren’t exactly up to speed on things either, but times are changing, so maybe your thinking should too.

Below, I’ll take a look at a few categories of in-game statistics, and try to lay out where many of us have been misusing them.

***

The first thing we all need to agree upon before moving forward is that having the puck is a really neat-o thing, correct?  Most rational thinking humans would agree that having the puck, rather than not, gives you a better chance to score goals. I think we’re on the same page so far. That “having” of the puck is simply known as possession. You know this. I’m not trying to be patronizing, just making sure we’re on the same page. In general, the team who maintains possession more over the course of the game will have the best chance to win. No chart, no graph, that just makes logical sense.

Let’s continue down the yellow stat road.

Score effects

“Score effects,” or “game states” as it’s known in European Kick Hockey (that’s how we refer to soccer around theScore offices) is the idea that when a team is up a goal or two or three, they are more likely to give up more shots and chances. That’s the very general idea, anyway. There is plenty of data to back this up, but I’m just trying to lay this out logically, because it does make absolute sense.

In some cases, hockey teams in the lead will switch from their 2-1-2 foreceheck to a 1-2-2 to make sure they don’t give up any glaring chances (because they’re trying to protect the lead). The dialed back forecheck leads to less possession (but again, hopefully less clean chances against too), which means the losing team has more possession, which means they get more shots. From an individual standpoint, if you’re on a team in the lead you don’t want to be the guy to give up the glaring chance, so maybe you play more conservatively, leading your team to get hemmed in more. And if the lead is big enough, guys will still play more positionally conservative, but knowing they have the lead, they may do so with less fervor than they would in a one-goal game, and suddenly it looks like the team in the lead is getting smoked, and there go your shot stats.

There is data to back this up the idea of score effects, but again: logically, the team in the lead is going to retreat and protect the net more which leads to more shots against, whether it’s a tactical decision or a psychological one. And the more that team leads by, the looser their play gets, and the more shots they give up.

This is why, on occasion, fans think “boy, our boys sure rallied and almost completed the comeback!”

No they didn’t.

Or, “boy, we got robbed, we outshot them and still lost by a couple!”

There’s a reason for that.

I don’t know if the Leafs changed tactics up 4-1 in the third period of Game 7 in playoffs, but the team definitely sagged while Boston pushed, which resulted in a memorable final 10 minutes.

So yeah, “score effects.” If a team is down 3-0 and being outshot 15-3 after the first, you’re safe to bet your buddy that the losing team outshoots the leading team on the way in. They likely will.

Other stats are skewed by the score and possession too. Don’t get caught misusing them. OF NOTE:

Blocked Shots

One coach I played for used to lay out nine goals that we had to meet every game. If we didn’t, we had to bag skate for five minutes per missed category during the following practice.  I don’t remember them all (I have it written somewhere), but I remember the basics, which were shots (30 or higher), hits (75 or higher – we had a maniac team and loose hit definition), blocked shots (I believe it was something around 7-9 or higher) and on and on and on.

The problem, you see, is that when you utterly own your opponent all night and have the puck for the entirety of the game, you don’t get many chances to block shots.

THIS IS WHY when Don Cherry or some Good Canadian Analyst tells you that a team isn’t willing to sacrifice because they don’t block many shots, there’s a realistic possibility that they’re just a really good hockey club who dominates possession. Among the league’s bottom-feeders in blocked shots during last season? Chicago (19th), St. Louis (20th), Ottawa (21st), Boston (25th), Detroit (26th), Vancouver (27th), and Los Angeles (29th). Not a bad cluster at the bottom of the league.

In sum, a strategy of “just block a ton of shots” in your own zone means a ton of pucks are being directed at your net, meaning your giving up a lot of chances for pucks to go in. We’re on the fringe of talking about an advanced stat here, so let’s rein it in.

Hits

Hits works on the same premise. Teams that are trailing are told, implicitly, that they need to get out there and lay the body to change the flow of the game. Teams that never have the puck have endless opportunities to hit people, because y’know, when you have the damn thing you don’t hit anyone. The team without it is always the one doing the hitting.

Sooo teams that lead the league in hits… not necessarily the toughest to play. Of the teams who finished in the top 10 in hits last season, only two earned home ice in the playoffs, and they barely sneaked into the top-10. They were Pittsburgh who finished 9th in hits, and Boston who finished 10th.

This doesn’t mean that being physical is bad. Being physical is GREAT. You just have to note that if being physical results in turning more pucks over and having it more, a good physical team may finish the game with less hits than the less physical team, because they earned the puck. You with me on that?

Having the puck equals chances to score. Not having it equals chances to hit. Therefore a higher hit count might not mean that one team is more physical than the other, just that they had the puck less.

***

Blocking a shot can be a huge, gutsy play for a team. Teams should block shots.

You should not say that a team is good because they block a lot of shots. It might be the opposite.

Hitting is a crucial part of the game. Teams should hit at every reasonable opportunity.

You should not say that a team is good because they have a lot of hits.

And with score effects, human nature and coaching combine so that leading teams are going to back off, which can skew a game’s stats.

So while stats are important in hockey, so are their context. Make sure you understand what the numbers are telling you. Sometimes that involves thinking a sliver deeper than you used to.

Comments (28)

  1. Another way I like to look at blocked shots is the inability of a team or player to generate shots in quality areas or on goal. Taking many shots from the perimeter is pretty useless you have a player that can get the pucks through the maze of humanity.

  2. Here’s my dumb score effects question:

    Is holding back after building a lead a good idea? Is it an effective strategy at denying quality shot attempts?

    If it is effective, then wouldn’t a team that trapped constantly have a pretty low Corsi, despite it being an effective way to win games?

    • @rw970 That’s why things like FenClose exist. We can divide the game into segments such that score effects won’t be having an impact on our results.

      Also, over the length of a season, or multiple seasons for even better effect, these types of things are more likely to come out in the wash. Even very good teams spend some time chasing the game, so over 82 games the numbers become less and less skewed by 5-10 minute segments of an individual game.

      I highly, highly recommend playing around with behindthenet.ca. It’s honestly just fun as a hockey fan.

      @Mike M:

      So if I understand what you’re saying correctly, a team that has a high percentage of its’ own shots blocked would probably be a low scoring team?

      • Right. What I’m wondering about, and apologies if I’m being thick here, is something like the old New Jersey Devils teams of the late 90s early 2000s. My recollection was that they trapped a lot, even in game tied or close situations – they were always in 1-2-2, as it were. (This might be wrong.)

        If this is true, it’s possible that they were getting outshot on a frequent basis, but legitimately keeping actual scoring chances down, even in FenClose situations, which would give them low Corsi and Fenwick values, all while they piled up wins.

        OTOH, my recollection is also that those NJD teams were very good – top 10 in goals scored as well as top 5 in GA, and I remember when the Leafs lost to them in the playoffs two years in a row, the Leafs were outshot.

        • Always worth looking things up. The 1998 New Jersey Devils finished second overall in the standings, and led the league in shot differential – second in both shots for and fewest shots against. They dominated the shot clock.

          There’s a difference between the Lemaire ‘trap’ system and falling back into a defensive shell.

      • pretty much. A team getting a larger than normal % of their shots blocked are probably shooting from distance or taking too much time to get the shots off. I would think without crunching the numbers that would indicate a lower scoring ability.

        • I’d be curious about this too. My gut tells me that it will be random, and not actually a very good predictor of good offensive teams or bad ones.

          My reasoning:

          Team A has 65 shot attempts, of which 20 are blocked and 20 miss leaving us with 25 shots on goal.

          Team B has 55 shot attempts, 10 are blocked and 20 miss the net, leaving us again with 25 shots.

          Team A has had a much bigger percentage of their shots blocked, but I’d be willing to bet that those 10 extra blocked shots meant more possession. That may not be the difference in a given game, but stretched over an entire season it’s going to be your sheer volume of attempts (relative to your opponent of course) that’s going to determine your success.

          • I am taking ‘real-time’ stats to mean 1 game. If a team is getting a much larger percentage of shots blocked than normal then their offense will go down in that 1 game.

    • It seems like it may be. David Johnson did a good post on this, which you can read here:

      http://hockeyanalysis.com/2013/05/21/the-theory-behind-the-defensive-shell-game/

      In essence, going into a defensive shell is a losing strategy over a long period of time, since you WILL be outplayed during that period of time. That said, it should in theory reduce scoring for BOTH teams, so if you have a lead, it’s worth doing at a certain point in the game.

      • Thanks. So a team can’t expect to trap constantly (e.g. even in the first period or when the game is tied) and win consistently?

  3. Not to mention that the hits stat is entirely subjective and varies wildly from arena to arena. I once saw a particular player deliver no less than three solid hits in the first period alone of a game I attended, and the official box score had him listed with zero.

    • Absolutely.

      Minnestoa would credit Clutterbuck a credit for a hit if he breathed on anyone. Hit tracking is horrific in some rinks.

    • Best bit is when you look at the season stats and see a team lay 50% more hits at home than on the road… or the rare case where those numbers are inverted and they have a boost on the road.

  4. “all the sudden”

    You’re one of those people.

    *************************************************************
    Blocking a shot can be a huge, gutsy play for a team. Teams should block shots.

    You should not say that a team is good because they block a lot of shots. It might be the opposite.

    Hitting is a crucial part of the game. Teams should hit at every reasonable opportunity.

    You should not say that a team is good because they have a lot of hits.
    *************************************************************

    It amazes me the number of people who cannot grasp a concept like the above.

  5. It’s also funny how subjective “hits” can be based on which arena. I am curious to see if Cal Clutterbuck is still going to be at the top of the league in hits anymore?

    • Equally subjective are takeaways and giveaways. I immediately dismiss any argument that uses any of these three statistics as their primary evidence.

    • Cal’s new team has the top hitter of the last three years so I think his hit stats won’t suffer at NVMC.

  6. I’m also convinced that a big part of block shots is where it happens on the ice. Shots from the point blocked up high by forwards can ricochet out to neutral and create rushes. Shots blocked in the slot can lead to redirections that handcuff your own goalie.

  7. As somebody who is constantly frustrated with some corsi apologists who attack any data that threatens the story they want to tell, I can say that this is a very reasonable, yet overly basic explanation of some fundamental truth that I can stomach.
    I think you are beginning to see teams focus more on quality of possession and shots over raw possession and shot numbers (both offensively and defensively). If a team wants to build their personnel on the over simplified notion that possession=wins, more and more smarter teams will take advantage and grant you your shot and possession totals whilst making them pay in the pdo numbers.

    • “beginning to see”? I don’t think they ever stopped. GMs continually hand out ridiculous contracts on the basis of points which can be the result of one season’s lucky bounces bringing up a player’s sh%.

      The whole point of PDO is that sv% and sh% are less sustainable than the possession metrics. Of course having Lundvist or Crosby on your team will make other teams pay, but…there are 20 other spots on the roster and this is a salary-capped league. When you’re a GM looking at a couple of third-line guys with similar points and league-average career sh%, you might want to look at the one whose possession scores are better to see if he’s doing something that contributes to that.

    • You can lead the horse to water but you can’t force your opponent to shoot. The concept that teams can consistently force opponents to take bad shots is just silly. Your opponent chooses when they want to shoot the puck, not you.

      So if you play good defense, you can keep your opponent from taking a shot and force your opponent to attempt to pass for a chance of a better shot. Which either works or they turn over the puck. Which means you can limit your opponents shots but NOT dictate the quality of their shots.

      Regardless of which metric you use to determine quality, over any extended period of time, the ratio of shots for and against will eventually be about the same ratio as the ratio of your so called quality shots for and against. That is why people care about Corsi.

      People have investigated scoring chances over the last four or five years and have come to this conclusion because they have actually looked at the actual data. So if you disagree, then show the data. Because the only people I have found that still disagree are the ones that haven’t actually looked at the data.

  8. Do you not believe that some teams have more specific criteria provided by the coaching staff as to where the shots (offensive and defensive) should be coming from and coach to that? resulting in save and shooting percentages that run counter to league wide data?
    Also, what i’m saying is that teams did begin to place emphasis on raw possession via shot data but are “beginning to see” that more advanced criteria for scoring chances are necessary.
    The whole point of PDO is determined by how it’s user leverages the data. I believe that some teams can and do attempt to elevate their PDO and exploit teams relying too much on ideas like corsi.
    Make sense?

    • Did you read the shot quality post? That’s what Toronto think they’re doing. History suggests that it’s not. The reason they made the playoffs probably had a lot to do with Reimer being non-concussed and posting a .924.

      I don’t think enough teams are using possession metrics to make it worthwhile coming up with schemes to get around them. Good teams aren’t good because their Corsi is high, their Corsi is high because they’re good. Lots of things that contribute to high shot differential are for-all-you-young-hockey-players stuff. Would you rather have a line cycling around in the ozone looking for the perfect shot, or one that gets off a shot from farther out while someone else drives to the net to bang in a rebound? A defence that backchecks and clears the zone or one that focuses on never letting a forward into the slot but gives up slappers from outside the circles?

      • Of course I read the shot quality post. It’s frustrating because I’m simply stunned how many people fail to acknowledge that Reimer’s save percentage should be higher under the exact system I’m describing.
        I would prefer the patient team that is focussed creating quality shots on the net that are high percentage over the team simply directing the puck at the net and looking for rebounds.
        You are oversimplying the defensive aspects because backchecking and keeping the slot clear are not mutually exclusive at all. You cannot define shot quality by distance since many of the lowest percentage shots are the shots teams were forced to take at the leafs net from about 3 ft out but from the side of the net. I’ve made my opinion clear about the system I prefer.

  9. ahh yes, advanced stat nerds, you only started to become hockey fans because you weren’t advanced enough to be baseball fans, why dont you all take down your Billy Beane posters and get the fuck outta hockey?

  10. You also have to take into consideration that the team without possession are more likely to take penalties which result in the other team having power plays, more shots etc

Leave a Reply

Your email address will not be published. Required fields are marked *