The Free Agent Frenzy of 2021 has brought some new names and faces to the Leafs’ roster, with some notable departures as well. Zach Hyman and Frederik Andersen are relatively long-serving Leafs that have moved on to other teams. To compensate, Kyle Dubas and company have brought in a few low-cost names to try to fill the gap Hyman left, and replaced Andersen with Petr Mrazek.

All signs point to the Leafs still working to add to this roster, and we could still see some players from last year moved out to make room, such as Pierre Engvall, Ilya Mikheyev or Travis Dermott. However, it still interests me to know how the additions and subtractions made so far have affected the team.

Before we start, I want to point you to the second part of my advanced stats primer, which gives what I think is a good description of where the “Wins Above Replacement” or “WAR” numbers we’re using below come from. In short, a couple folks much smarter than I have developed a model that boils down a lot of the things we know about advanced hockey stats into a few numbers that essentially show how much better or worse a player is than what a typical replacement player would be. This idea comes from a very similarly named model in the baseball world that you’ve probably heard of if you’re reading this article.

Because someone can be above or below replacement level in this model, not all players who leave will be subtractions, and not all players who come will be additions. So instead, the players are grouped by “departures” and “arrivals”, and the numbers will tell you which are subtractions by subtraction, additions by addition, or otherwise.

A note about the tables below, they all start with “WAR” which is wins above replacement, then “GAR” which is goals above replacement. The former is based on the latter. The “GAR” is based on each of the components that are in the following columns. The glossary from Evolving-Hockey will be of help if you’re not sure about any of the short forms used.

In the outgoing column of our ledger, we have a bunch of depth players, and then the aforementioned more sizeable departures of a top-6 LW and the one-time goaltender who had their job taken.

Looking at the data above, there’s really only one change that on its own will have a significant effect on the change in quality to Toronto’s roster. Zach Hyman is a significant loss that the Leafs will have to actively try to replace, or suffer a top-6 forward group significantly reduced in effectiveness. By himself, Hyman was worth 1.5 wins for the Leafs.

In fact, the other departures will net to a change of exactly 0 wins above replacement, which is neat.

As we led in with, there’s almost certainly more to come from Toronto, but so far, the only major change is bringing in Petr Mrazek to replace Frederik Andersen. Otherwise, the players brought in are, at best, bottom-six players. If all goes well, I’d be surprised if Alex Biega or Michael Amadio ever play for the Leafs. However, because I can’t predict the future, we’re still counting those guys in the totals here.\

Note that Ondrej Kase, who the Leafs signed last night, is not included, as he only had 25 minutes of ice time in 2021. That’s not enough of a sample to be predictive enough to know what we’re going to get. His injury history makes him a total wildcard, and so it’d be unfair to use his 2020 numbers and assume that’s what the Leafs are going to get.

We can see that as a major signing, Petr Mrazek appears to be a significant add for Toronto. He stole a little over 2 games last year, even though he only had 12 starts. Now, obviously there’s some small sample size issues here, so Mrazek probably won’t do the same again. It would be a tremendous feat if he could. But the fact that he performed so well for Carolina is certainly a good sign for Toronto.

On the other hand, it looks like David Kampf could be a problem for Toronto if he’s given anything more than a 4th line role. He did average 3rd line minutes last season in Chicago, on a team that frankly wasn’t very good, but the WAR model deos take Quality of Teammates into account, so this is a little bit worrying.

As mentioned above, Amadio and Biega probably won’t see much time in the big league, but their below-replacement level seasons last year will negatively contribute to how Toronto’s additions look.

To end on a positive note, the one significant signing here other than Mrazek is Michael Bunting, who seems to be a mold for a Zach Hyman replacement, and has good numbers to back it up.

It seems like a bit of a crime to take such a complicated statistical model and just do a bunch of addition and subtraction with it and call that my own analysis. However, that’s the beauty of a WAR model, as that’s exactly what it’s meant to do. So, adding up the Goals Above Replacement (GAR) and Wins Above Replacement (WAR) of the players Toronto brought in, and subtracting the same of the players who’ve left, we get the following.

*-with Kase this included this would be 1.8

A very slight, but definitely positive, change.

As mentioned, this is likely not the final list of either arrivals or departures for Toronto. It’s unfortunate that Toronto couldn’t hold on to Jared McCann and his 3.2 WAR last year, good for 2nd in the league for all forwards, as that would have made this look fair bit nicer.

There are still some quality players out there for Toronto to add, though; from Nick Ritchie, to Tomas Tatar, to Tyler Bertuzzi, to someone I haven’t even thought of. Whoever it is, the Leafs will need to make sure that they’re a positive addition, and not one that fails to improve upon last year’s roster.

Also, it’s important to note that boiling down piles of data to one number can be flawed, as it lacks nuance, and fails to account for whatever biases might exist in the model. But, this isn’t the place for research papers; WAR makes a quick look at whether the Leafs might have gotten better or worse very, very easy, and that’s why it’s what I chose to use for this.

All WAR data comes from Evolving-Hockey.com. This data is behind a paywall, but they have some more basic data available for free and it’s all in a really easy-to-use web format. They also have, for free, links to references and in depth articles on how their model was constructed. I do highly recommend supporting the site if you have any interest in pulling numbers like this for yourself in the future.