Blog 9: LS-GAA and percentile – Is hockey a strong or weak link sport?

In the midst of writing a blog about player projections, I came to realize a much smarter way to graphically show my points from the previous blog. Instead of tiers, I’m simply drawing LS-GAA as a function of percentile. I’m using the same data as I did in the last blog, so I’m looking at a 3 year time frame from the season 2017/2018 to the season 2019/2020.

Forward LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>900):

This graph shows pretty much what I concluded in blog 8. The top end players (ca. top 5%) and the bottom players (ca. bottom 5%) have relatively greater impacts on the game than the other 90% of the forwards. So the difference between the best player and the 5th best player is much greater than the difference between the 101st best player and the 105th best player.

We can also look at the total contributions instead of just looking at per 60 data.

Forward LS-GAA_total from 2017/2018 to 2019/2020 (TOI>900):

Now we see an even steeper impact by the top players, but a much smaller negative impact by the worst players. This is of course due to ice time. Your top players play top line minutes and special teams, whereas the worst players hopefully play in a very limited 4th line role.

With that said, let’s now look at defensemen. Here’s the per 60 data.

Defender LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>1200):

Surprisingly, you see some differences between the forwards and the defenders. There seem to be a smaller percentage of elite defenders (ca. 3%) and a larger percentage of bad defensemen – the slope starts at around 85%.

And here’s the total LS-GAA as a function of ranking.

Defender LS-GAA_total from 2017/2018 to 2019/2020 (TOI>1200):

For the most part this looks like the graph for forwards. Most great defenders play a lot of minutes, so the impact of the best players increases, while most bad defenders play a limited role, so the negative impact is limited as much as possible. The worst defender (Jack Johnson) has an equal negative impact to the positive impact of the best defender (Jared Spurgeon). This is interesting, when it comes to roster construction.

Let’s finally take a look at the goaltenders. First we got the per 60 data.

Goaltender LS-GAA_60 from 2017/2018 to 2019/2020 (TOI>1200):

The first thing you see, is that the extremities are less extreme. It’s tougher to conclude exactly when a goaltender is elite, as the distribution is more linear throughout the graph. The second thing you see, is that 60% of the goaltenders are above average. So the bad goalies are worse than the good goalies are good. Bad goaltending can really kill a team.

And here’s the total contributions from the goaltenders over the last 3 seasons.

Goaltender LS-GAA_total from 2017/2018 to 2019/2020 (TOI>1200):

It’s pretty rare these days to have a true starter performing at an elite level like John Gibson is. Most of the teams with great goaltending today employs more of an 1A/1B tandem, with both goalies being in the top 25%.

Strong link or weak link game?                    

You often see discussions on whether hockey is a star driven sport (strong link sport) or a depth driven sport (weak link sport). I think most people lean toward a strong link game similar to Basketball. But perhaps not to the same extent.

With the graphs above we can better discuss this subject. I think the forward graphs show that having top end forwards is definitely more important than having forward depth. When it comes to defenders the picture is much more grey though. The top end defensemen are less dominant than the top end forwards, and the worst defenders have a greater negative impact than the worst forwards. I think you could argue having D-depth is at least as important as having an elite defenseman.


  • I think, the data indicates that having a top heavy forward group is better than having great forward depth with no star players. For defenders the discussion is much less black and white though. Here you could argue D-depth is more important than top end quality. I think a bottom pairing defenseman plays a bigger role than a 4th line forward. The impact of a depth defender is therefore greater than the impact of a depth forward. It’s easier to hide a bad forward than a bad defender.

Stay safe and remember to be kind

All raw data from

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: