It's not MY goal, it's OUR goal.

[Over the course of drinks last night, my esteemed colleague Mr. Cam Charron and I had a bit of a discussion about the eccentric concept of 'the assist', and he suggested that, as he was writing on that very topic this very day, it might be edifying and synergistic if I did so as well. So I am.]

Another one of Lester Patrick’s magnificent PCHA hockey innovations, assists were invented to give credit for playmaking. As a defenseman, Patrick had long been frustrated with the fact that hockey credited only the scorer despite the fact that so much of the work of moving the puck up-ice was often done by the guy who passed to the scorer. So, in 1913, when the PCHA introduced the forward pass, they also introduced the assist. Like shifts and lines, forward passes and assists were one of several PCHA concepts that pushed the theory and practice of hockey slowly away from a game based on individual puck-hoggery and towards a team-first, share-the-biscuit ethic. By giving credit for playmaking, players were encouraged to try to set up a teammate to score rather than simply trying to score themselves.

The NHL didn’t start crediting assists until 1918, and when they did, they did so according to the conservative PCHA standard. Because forward passing was still illegal, and later very heavily restricted, through most of the1920s assists were awarded at a very low rate of between .4 and .6 per goal. However, in the late 20s, the League began eliminating restrictions on passing, and accordingly began a series of experiments in assist-counting that would continue for nearly a decade.

Our modern concept of the assist evolved over the course of the 1930s, but it was not a direct evolution.  Many of the provisional redefinitions of the assist seem to have been designed to limit the awarding of assists to direct passing. For example, from 1930 to 1936, the NHL allowed the awarding of up to three assists per goal, but didn’t allow passing between zones, which meant that on the rare occasions that all three assists were awarded they were the result of a successful continuous breakout or cycle. In 1936-37, they reduced the maximum number of assists to two, and also added the new provision that assists could only be credited for passes made in the offensive zone, meaning an assist in 1937 was significantly harder to earn than it is today.

Correspondingly, throughout the 30s and early 40s, assist-counting rates and practices fluctuate dramatically with changes in definition. In 1942-3, the NHL finally settled on the two-assists-from-anywhere-on-the-ice-system, but the standard for crediting them was still higher and left up to the discretion of the scorer- assists weren’t automatically given for touches. In the mid-40s, an average of 1.132 assists were credited per goal, suggesting that scorers were operating under a more restrictive definition, probably (although I don’t know this for sure) one that demanded a more direct playmaking contribution to a goal.

Over the next fifteen years, though, the definition of an NHL assist slowly softened. Assists-per-goal rise a little bit every year from 1945 to 1957. This seems to reflect the slow evolution and application of the modern definition, wherein the last two players to touch the puck are credited no matter how minor, incidental, or accidental the contact. From 1957 to the present day, the NHL’s assists-per-goal rate has remained remarkably level, hovering between about 1.65 and 1.75 every year. The system is, if nothing else, consistent.

However, as my esteemed colleague Mr. Charron points out, it’s also arbitrary. As Dellow noted in his breakdown of Karlsson’s assists, or as this fine fellow describes, a lot of assists are granted for actions that happened long before the goal and made no direct or intentional contribution to it. Given the fairly consistent application of the standard, it is probable that most players benefit more or less equivalently from the awarding of assists for non-playmaking actions- that is, Karlsson is probably not unique in getting points for a lot of plays that we might not consider, in and of themselves, all that remarkable. What Dellow identified (in the absence of the second part of his post, which may or may not have made the case more clearly) is likely a systemic problem with assists themselves rather than with a particular player. But it raises a valid and important question: why is hockey giving out points for non-playmaking gestures?

The obvious answer is that, most of the time, the non-playmaking gesture was still a good gesture. If Crosby makes a nice pass to the corner of the crease, and that pass is the foundation of sustained possession between two other players that eventually leads to a goal, he may not really have contributed to that goal, but, well, he still made a nice pass. If Karlsson makes a good clear out of the D-Zone that his forwards eventually process into a goal, well, it was still a good clear. Both players have made the right decision, executed it well, and thereby maintained their team’s possession.  Possibly, in doing so, they’ve helped to create the space necessary for their teammates’ offensive creativity. The two-touches system of awarding assists seems specifically designed to capture and reward these kinds of small but laudable actions.

I think that’s why they were so important in hockey from the fifties through the seventies, and why back then the two-touches theory was perfectly reasonable. Remember, for these generations hockey games were not always recorded and seldom broadcast, and when they were, the quality was low and replays difficult. Hockey, without the benefit of digital resolution and slow-motion, is a complex swamp of collaboration between players. Everyone knows, then as now, that there are lots of small, solid, unglamorous plays that go into the creation of a scoring chance, but for most of hockey’s history, the accurate identification and counting of such plays would have been excruciatingly difficult and impossible to verify. In that limited milieu, the best possible way of rewarding the tiny contributions that create “a climate where scoring chances are more likely to happen” (quoth Charron quothing Dellow) is the two-touch assist system.

But that was then and this is now. Now everyone has access to high resolution digital recordings of games that we can speed up or slow down at will. One of the best things about being born in this generation of hockey is that we- all of us, not just the coaches, not just the broadcasters- can see the game better than ever before. Which means we shouldn’t need to use assists as a sort of vague proxy for both playmaking and some random percentage of good gestures. We don’t have to count only the last two things we saw from memory alone. We can go to the tape. We can count all the things.

The small good gestures that sometimes turn into non-playmaking assists happen hundreds of times in a game, and most of them don’t turn into goals. The player who makes that simple perfect play makes dozens of them, and every now and then, by pure luck or the whims of the hockey gods, one becomes a goal. Sure, second assists cover some of them, but we’d know and value his contribution more fully and more accurately if we counted all the simple perfect plays regardless of whether they end in goals. And, in fact, there are analysts trying to do just that- breaking down players according to puck battles won and lost, passes completed and intercepted, zone entries successful and unsuccessful, shot attempts and scoring chances generated. That some journalists reject these kinds of advanced statistical projects while still considering non-playmaking assists legitimate is a rarely noticed but indefensible form of hypocrisy.

So I present the question, speculatively but sincerely: would it be good for hockey if the NHL went back to crediting assists only for passes that directly set up the goal, while looking to other stats to capture valuable non-playmaking contributions? Assists, then, would become a more precise metric that measured a specific and highly essential offensive skill: the ability to set up a shooter in scoring position. There would be fewer, true, but they would be more meaningful, and might give the non-specialist fan a better read on the difference between an elite set-up man and an intelligent, diligent support player- between a driver of scoring and a driver of possession. Both, of course, are essential to a good team, but they’re different skill sets. Right now, our system of assist-counting blends them together in a stew of accident and chaos that, ultimately, reflects neither ability clearly. We have the technology to separate them. Maybe we should try it.

Info on the history of assists was taken from The Patricks: Hockey’s Royal Family, by Eric Whitehead, The Annotated Rules of Hockey, by James Duplacey, and QuantHockey.

Comments (26)

  1. I think we need to keep in mind that the NHL is not the only place where stats are kept. As someone who has reffed kids and beer league, I think that we need to have stats that are consistent across all levels ( I also take the time to award them as best I can) . Certainly, there is a place for advanced stats at the higher levels of the game, but you also need these “flawed” stats that can be applied to other levels, for comparisons sake and, let’s face it, for fun!

    • Oh man, consistent awarding of assists at beer league/kids levels is a whole other can of worms. A tighter standard that credited direct passes only might make it actually easier to be consistent at lower levels (although harder to give charity points to the less talented, I guess- wait a minute… that’s me! OPINION CHANGED.).

  2. Good to at least have a standard – cursory examination of the KHL suggests that standards are less stringent there. Last season, St. Petersburg was at 1.5 assists per goal, Omsk 1.44, Chelyabinsk at 1.36, Cherepovets at 1.66, Kazan at 1.6.

    I’m for keeping the assist as is – it would radically alter how I think about basic hockey stats to change it. I tend to give it far less weight than goals when evaluating a forward.

    • Those are some pretty small variances. You sure that some of the teams with lower assists per goal don’t have more guys that carry the puck in themselves or make good defensive plays in the neutral zone?

  3. I’m sort of on the fence with this one. I think you present some reasonable arguments, but… I still like the idea of crediting the additional guy. Sure, it’s not a perfect system. Some guys get assists they shouldn’t have. Some guys should get some they didn’t. And yeah, we COULD go back to the recording and make more precise decisions. But even then we’re still a bit reliant on opinion for the 2nd assist. You might say it’s the same play that guy made a dozen times earlier that didn’t result in a goal. Maybe I think it was different and subtly more deserving of credit.

    I also think eliminating the 2nd assist would likely effect defensemen disproportionately as (and I have no data to support this claim) they are probably the group that receives the higher number of 2nd assist points, since they are the ones usually starting the outlet from the zone.

    Also… we want more points!!

    • Re: first paragraph: I’m a bit skeptical of the idea that the little things that eventually, later on, end up in goals are subtly different from ones that don’t, but my view of hockey tends to be more “it’s all chaos and madness” than “everything happens for a reason”.

      Re: second paragraph: I think we need to look beyond goals and assists for evaluating defense and defensemen generally, so I wouldn’t necessarily see them getting fewer assists as a bad thing if we compensated for it by valuing other measures.

      Re: third paragraph: The NHL should go back to three assists and let guys be credited with assists on their own goals. Now THAT would make for some exciting scoring races.

      • I always wanted to be able to assist on my own goals even when i was very young, like 6 or 7 years old.

      • I have absolutely, unequivocally, been responsible for setting up a few of my own goals.

        Regarding defensemen, I agree that we cannot just look at points for evaluating defensemen. But, some of them a really good at escaping pressure and setting up their offensemen for a good quick counterattack. I still think they should receive some credit for giving birth to the rush that lead to the goal. It’s total hyperbole, but there are definitely times when the Dman creates the play. He escapes pressure. Alludes a check. Makes the great outlet pass. Only to be erased from the points sheet by that one forward who touch passes his outlet to the streaking scorer?

  4. Right now, as the rules go, if a goalie makes a save and there’s a goal scored ‘on the rebound’, the player who took the original shot gets an assist. WHY? He has no control over where the rebound goes. I think first assists should be for passes that result in goals, and second assists for passes that create scoring opportunities that result in goals, such as 2-on-1s.

  5. Actually goalies do have control over where the rebound goes… and one assist per goal? tryin to shake up the hockey pool system are ya?

    • They have partial control. Great shots can be harder to control properly; JB has talked in prior posts, for example, about shots to the back low side that are specifically designed to cause second shots – they’re very unlikely to beat an NHL keeper, but they’re also tricky to steer to a harmless spot on the ice.

    • The player who took the original shot has no control over where the rebound goes, or if there even IS a rebound, so why should he get credit for the goalie’s actions?

      • While technically true, i’d say that’s only half true. Many, MANY guys shoot the puck in a place to leave it in a spot for a rebound goal. Goalies take the shooter. Defensemen take the extra attacker. If you can’t make a pass, and the goalie is good position to stop the puck, it’s not a bad idea to shoot one low, hard, and sort of into the middle/weak side of the goalie. He’ll stop it for sure, but likely won’t have the time to react and deflect it out of the play. It will give the additional attacker a chance at a rebound. It happens more often than maybe you would think. It is definitely an intended play and a conscious decision by the puck carrier. It’s deserving of an assist when it works.

  6. “would it be good for hockey if the NHL went back to crediting assists only for passes that directly set up the goal, while looking to other stats to capture valuable non-playmaking contributions?”

    I’m all for trying to find new ways to describe “good” and “bad” in hockey. But all your suggestion would really do is add *more* scorer bias into the scoring system. There are going to be some assists that are junk and not really reflective of the “goodness” of the play. And there are going to be some goals that are junk and not really reflective of the “goodness” of the play.

    Id rather the world spend more time trying to develop those other stats, which help offer a new perspective on scoring, rather than add a new subjective variable to an already imperfect stat.

    • There’s a lot of subjectivity in pretty much every advanced stat count, which is a problem that can be managed but will never entirely go away. But in the difference between counting something objectively that may or may not actually contribute significantly vs. counting something subjectively that definitely contributes significantly is a matter of preference rather than fact. It’s just shifting where the ambiguity goes, not mitigating it.

      That said, I agree with the general point that we need to develop/popularize more accurate stats overall.

      • Right now, there’s little ambiguity in assist counting. The last two touches get it. Yea, there are the examples of bounces off opposing players and goalies but Im guessing those are the minority. You’re concluding, based, it seems, solely on one (admittedly good) Dellow post, and one guy who had showed 2 extreme examples, that…

        “…it is probable that most players benefit more or less equivalently from the awarding of assists for non-playmaking actions- that is, Karlsson is probably not unique in getting points for a lot of plays that we might not consider, in and of themselves, all that remarkable.”

        There may be more Dellow-like analyses out there that second assists are “non-playmaking actions” or not “all that remarkable.” But at this moment, I dont have a desire to change assists from a primarily objective stat whose value is questionable, to a primarily subjective stat whose value is questionable.

  7. “Given the fairly consistent application of the standard, it is probable that most players benefit more or less equivalently from the awarding of assists for non-playmaking actions- that is, Karlsson is probably not unique in getting points for a lot of plays that we might not consider, in and of themselves, all that remarkable.”

    I disagree with this. If we imagine a pool of ‘non-playmaking assists’, we can imagine these being assigned uniformly at random to the passes made in a game. Players who make more passes and are primary puck-handlers for their team would have a larger subset of these assists then power forwards or defensive defensemen who simply don’t touch the puck as much to get a substantial portion of these assists. For a clear example, one expects that Karlsson benefits more from these assists than Craig Anderson.

  8. You can’t change the scoring system because you’ll impact history. It’s like going back in time and doing something that didn’t happen and returning to current time to find out that Don Cherry is now the Prime Minister of Canada. Think of all the history records. How would we compare and denigrate today’s players versus Gretzky and Kurri and Coffey and others who weren’t Oilers in the 80′s. Record books would be full of asterisks and explanations of why scoring records don’t jive between eras. Bloggers would have endless crazy theories about how to figure out which current players, with some complex playmaking assist thingy total, were the equal of past stars who averaged 2.7 points per games. And without that second useless assist I would have been constantly last in my team’s scoring.

    Don’t change a thing.

    • Scoring already doesn’t jive between eras. As goals per game varies, assist totals will vary. It’s laregely a function of new rules and how frequently penalties for infractions are called.

  9. In today’s computer age there are already tons of various stats that can be and are pulled up in seconds in ways of analyzing and assessing a players abilities and contributions. I don’t have a problem with the way assists are now. They will balance out. Besides, there seem to me to be far more important changes that need to be considered.

  10. On my beer league team this season I am leading the teams in goals but I am always the first to admit that I am the recipient of great plays at the back end on the break out and great plays on the set up. I like the way the assist system is set up right now. Sure its imperfect, and you will get assists awarded that probably are more a lucky break than a steller play but the result was the same.

    It would be like situations where a defender or goaltender coughs up a puck and accidentally knocks it in his own net. Its never scored and announced as “Goal for Team A scored by So and so from Team B.”

  11. Ever wonder how many of Gretzky’s assists (which, keep in mind, had he never scored a goal in his entire career, he would still be the all-time points leader, on assists alone) were secondary assists? How many of those assists could be considered as having directly set up a goal?

  12. This is a great read.

    Ive been looking through advanced stats websites and one stats I cant find is the amount of assists that are in actual fact missed shots that rebound to the goalscorer/primary assister.

Leave a Reply

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