Let’s continue the conversation about quality versus quantity of shots.
Here is a graph showing the winning percentages of teams who held a shot differential advantage so far this year:
So for example, teams that finished the game with one or two more shots than their opponent have won 54.42% of their games this year, teams that led by 10 or 11 shots have won 66.28%, and teams that led by more than 20 shots have won 61.22%.
Does having more shots equate to a higher win percentage?
What one would fully expect is for the win percentage to rise steadily as the shot differential grows, essentially it doesn’t at all. Having over 20 more shots results in winning 61% of games, but 18 to 19 more shots only results in 50%? There have been 38 games this year where the differential was 18 or 19 so the data sample is not huge, but it’s not tiny either. There doesn’t appear to any noticeable upwards trend.
I honestly couldn’t believe this, I initially expected that teams with over a 10-shot differential would be winning 75% of the games. Instead, even with a 20-shot surplus, you can only expect to win just over 60% of your games.
Now 60% of your games is pretty solid. A .600 winning percentage will essentially guarantee you a playoff spot, but not much more than that. Of the Western Conference playoff teams only the wild card teams are under .600.
So it would be very hard to say that taking a lot of shots by itself is what will seal the deal.
In the previous article on shot quality versus quantity I showed a link between high shot volume and high danger shots. Basically, the more shots you take, the more likely you are to get quality shots. Yet here we see that simply having a lot more shots doesn’t yield playoff guarantee results.
How Fenwick might affect the outcome
Another stat I would like to look at is Fenwick, or unblocked shot attempts as it is known by the NHL. These are exactly what they sound like—shot attempts that aren’t blocked. This includes shots that don’t hit the net. This and Corsi (which is the same stat but includes blocked shot attempts) are often seen as metrics for how well a team controlled the game under the assumption that the more you have the puck the more shot attempts you will have.
Here is the same graph as above but for Fenwick. There are fewer groupings and the upper limit is raised to 25 because Fenwick is far more common in a game than shots.
Up until a surplus of 15 unblocked shot attempts, teams that lead in Fenwick barely win more often at all. After that the 15 to 19 category and 25+ are decent bumps, but the 20 to 25 category is by far the lowest at 38%. It’s hard to understand why this is the case as each category has a decent sample size of games, but I have to believe that this would increase over a larger sample size.
Yet even here we see that having over 25 more unblocked shot attempts than the opponent doesn’t even lead to winning over 60% of games.
The expected goals view
Before I get into my point, I’d just like to make sure we’re all on the same page with the term expected goals. The name can be somewhat deceiving. In a hockey game, each unblocked shot attempt is assigned a probability, between zero and one, of being a goal. Many different considerations are taken into account, like the distance from the net, the angle of the shot, rebounds, and other events that happen before the shot. In that way, a rebound shot that happened immediately after and from the other side of the net of the original shot will be assigned a number close to one. A shot from the red line will be near 0.
So even though it may sound like a measure of goals it is really a measure of shot quality and, to a lesser degree, quantity.
Just in case you hadn’t seen enough, here’s the same graph again, but showing expected goals.
This starts to look more like what we would expect. The general trend is upwards though there are some large dips. As you get to the top the sample sizes become too small to put a lot of stock in, but I believe that trend would be likely to conitinue with more data.
If you end up with about 1.5 or more xG in a game you can expect to win around 70% of the time or better. The difference between a .600 team and .700 is approximately the difference between the second and 12th team in the league. Only Washington and Winnipeg are above .700 so far this season.
This paints a picture showing expected goals and quality shots as the key to winning in the NHL. Again, I have stated in a previous article that the data seems to indicate that taking more shots is the best way to get more quality shots. So now we’re going to look at whether teams fare well even when they don’t neccesarily get a lot of shots.
Here is the same graph but in CANADA GOLD! Thanks for the easy tune up match USA.
This graph shows the win percentage of teams who led in xG/shot this year. This is a great illustration of how to win a hockey game: take quality shots, however you can get them.
With this graph we can see that even when teams aren’t taking a lot of shots, they are still winning so long as they are good quality shots.
This is good insight from a theoretical standpoint but is difficult to usefully implement in practice. Essentially it says, “take lots of shots, but make sure they’re all high quality”, which is easy to say and hard to do. It is still important to understand this though in order to inform strategy.
The challenge is to try and devise a plan that maximizes shots without sacrificing shots too much. In that way it may be that getting a medium amount of medium danger shots is better than few high danger and many low danger shots.
stats from NHL.com , moneypuck.com, and naturalstatrick.com
One Comment