The four players named above have had some of the most eye-opening seasons this year on the Leafs, for different reasons. @Nick Ritchie’s season opens your eyes in the way that makes you question whether he’s one of Toronto’s 12 best forwards when everyone is healthy. @Timothy Liljegren, on the positive side, has opened our eyes to how confident and effective the young two-way defender can look on the bottom pair. @Justin Holl, obviously, earned himself a healthy scratch with seemingly slow and ineffective defending, losing his spot to Liljegren. While @Alexander Kerfoot didn’t get off to a great start, his play lately with Mitch Marner has looked electric.
It can’t be questioned that each player’s natural ability has lead them, to some degree, to this point. However, all three of these players have also been affected by things outside of their control. What I want to explore today is statistical regression around, specifically, PDO.
PDO isn’t actually an acronym, though it seems like it would be one. Someone once proposed that it would stand for “Percentage Driven Output”, but that was after the term was already popularized. The name is apt, though: what it’s all about is how percentages, specifically shooting percentage and save percentage, drive perceptions around you.
It’s a tale that keeps telling itself, every season, to the point where it’s almost boring. Some player starts to get a surprisingly bad reputation or a surprisingly hot run of play, and then, just as surprisingly, they’re back to their usual results. Except the return to normal isn’t surprising at all, it’s the natural order of things, it’s equilibrium, and that’s what PDO is here to show us.
How does it work? It’s simple: you add a player’s on-ice shooting percentage and their on-ice save percentage together. If it’s more than 100, they’re getting excessively positive results, and conversely, less than 100 means excessively negative results. Typically we look at this at 5-on-5 only.
In most cases, after a lot of ice time, PDO returns to something close to true talent. For most players, this true talent is at 100 PDO. For Toronto’s stars, they have been playing for so long with such a high shooting percentage, we can accept that their natural PDO sits higher at about 102. In general, Toronto sits a bit higher than 100 — no one with at least 1500 minutes of ice time as a Maple Leaf over the last 3 seasons has less than 99.9 in PDO.
What about this season? As I led into, the three players in the title have been especially affected by PDO. Here’s how Toronto’s roster has shown this season:
This table and all of the stats so far come from Natural Stat Trick.
We can see that Holl and Ritchie sit at the top of this, which is the lowest PDO, and Kerfoot and Liljegren at the top with the highest. That’s obviously why they’re highlighted at the front end of this post.
However, we can’t just look at PDO. It’s not the only thing in hockey that drives results. We know that Corsi For Percentage (CF%) can be predictive of future outcomes. So, together, each of these four players has a different story.
With @Nick Ritchie, his PDO has been atrocious, while his CF% has been pretty good. To me this is indicative of his season turning around for the positive. His on-ice shooting percentage is the lowest for any of the Leafs, an unfortunate result for him so far. It suggests that Ritchie just needs a little more time to see some pucks start going in the other net.
However, for @Justin Holl, his PDO is kind of bad, but his CF% is also pretty bad. His percentage results might improve somewhat, but CF% is predicting that his results will continue to be poor. Unsurprisingly, @Jake Muzzin’s story is the same, since Holl and Muzzin have been paired together whenever Holl has played this season. Notably, Muzzin and Brodie together are a 52.4% pairing in terms of Corsi, and sit just a shade above 100 in PDO, so hopefully that pairing can get re-united.
Looking on the high PDO end of the table, @Timothy Liljegren has the Leafs’ highest PDO, driven by a very high on-ice save percentage. It’s a luck-driven result. We have learned over the years that defenders don’t influence their goalies save percentage to any substantial amount. But, at the same time, Liljegren and his partner @Rasmus Sandin have been one of Toronto’s best CF% defensive pairs, albeit in soft competition. As such, I expect that as the season goes on, Liljegren may not look quite as good as he has so far, but his results will continue to be strong.
And, once again in a reverse of the previous story, @Alexander Kerfoot’s CF% has been just mediocre. It’s not bad, being above 50%, but not great either. It suggests that what’s been happening so far will probably continue. His PDO, though, is suggesting that what has been happening has been too good, that things will fall backwards a bit for him. This is concerning, as I don’t think he’s been great, other than a few good games, including last night’s game against Calgary.
There are a few other notable PDO values that I didn’t break down, because I liked the way the two players at the top and the bottom of the table each told different stories. One thing that is really impressive to me is @William Nylander, who has been on an incredible run of good play, but has a perfectly reasonable PDO.
Another thing that should be scary for the rest of the league is that @Auston Matthews’ on-ice shooting percentage so far sits a fair bit below what we would expect.
PDO is a wonky concept, I know. The human mind wants to credit what these players have done as being what they truly are. But these percentage-based results have been regressing back to the mean for as long as I’ve been paying attention to hockey stats. I don’t have any reservations about the predictions that PDO makes when they’re in small samples like the 15 games of this young season. Feel free to doubt; skepticism is healthy, but in this case, I’d say that the numbers don’t lie.