When I first compared Horachek to Carlyle in an article I wrote about a month ago, I argued that the Leafs did not have the right fit of players to successfully implement a Horachek style system. It was admirable what Horachek was trying to accomplish with this team defensively, but in doing so he actually nullified many of the strengths this team had, which were speed, creativity and finesse. Essentially, by eliminating high-risk plays from the game plan Horachek improved the team defensively, but as a consequence also hindered the team offensively. It was a classic risk/reward tradeoff. In fact, I deduced that despite a change in systems that had seen a significant improvement in the Leafs’ shot metrics, the Leafs would actually lose more often and more consistently over time as a result.
Many argued that these conclusions were premature consequently pointing to a Leafs PDO that was quite low as evidence that the sample size was too small and that the Leafs were just suffering from ‘bad luck’.
“Give Horachek 20 games and you’ll see! The Leafs will be way better than they were under Carlyle!” (tweeted one person)
“This is very similar to how Trotz transitioned to the Capitals. Give it 20 games and this team will be winning more often than not.” (tweeted another)
In the article’s defense, it was noted that the analysis was not based on statistical data alone and that a thorough scouting report of the Leafs accompanied these conclusions. A combination of both analytics and video scouting can prove to be a lot more insightful in small samples than either on its own. What’s really the point of game-by-game analysis, if we have to wait so long after the fact to make predictions about the future? In fact, based on the logic of ‘sample size’ arguments nothing in-season can be controlled by a team, its players or its coaches. Yes, anything that is in the short term should be taken with caution, but to discount it entirely is just plain neglect.
Again, the idea of ‘luck’ had very little to do with the Leafs losing this time around and ample evidence was provided to prove it. I’m not sure about other teams and for the time being I don’t really care, but for the Leafs, their low PDO had very little to do with luck.
Well the Leafs have played just about 20 games (18 to be exact) under Horachek and have gone 2-14-2 during that span (let that sink in for a bit). Yes, they are doing slightly worse than I initially expected, but losing begets losing so really it’s not at all that surprising. Let’s look at some team stats to see if anything has changed.
A look at the Numbers
Below are some tables that break down Team stats for both Horachek and Carlyle based on score situations. The results are very interesting.
All score situations = total overall stats
When tied = includes stats only when Leafs were tied in score with opponent
…Leading = includes stats when Leafs were leading in score vs. opponent
…Trailing = includes stats when Leafs were trailing in score vs. opponent
PDO = team save% + team shooting%
When comparing both tables what really stands out is % of TOI when leafs are tied, leading and trailing. It’s a rare occurrence that Horachek’s Leafs are ever in the lead. In fact, out of 1090 minutes of hockey played only 17.5% of that saw the Leafs with the lead. Seriously, that is absolutely, disgustingly, ridiculously atrocious. Further to that, more than half the time, 55.3% to be precise, saw the Leafs behind during that span. That’s a paltry 44.7% of the time the Leafs were either in the lead or tied during that span. Compare to that of Carlyle’s Leafs, and we see a team that spent a robust 72% of the time either in the lead or tied in a game. To say it’s a landslide victory in Carlyle’s favor is almost an understatement. However, before I go blowing a head gasket on these initial findings let’s look a little deeper to see if anything has gone overlooked.
Based on the above PDO results and if PDO is indeed truly a measure of luck, then luck must be a situational occurrence, because it appears the Leafs under Horachek have ONLY been ‘lucky’ when leading. PDO is abysmal at 91.9% when tied and improves significantly to 95.3% (still well below 100%) when they are trailing, but the most significant change occurs when they are leading at 103.2%. So based on a ‘certain’ group’s definition of PDO, the Leafs were very unlucky when they were tied, just slightly unlucky when they were trailing and quite lucky when they were leading. Wow, who would’ve thought score effects would have an influence on a team’s PDO! *Sarcasm alert*
On a serious note, let’s look at this PDO situation realistically. When a team is trailing they tend to loosen up their defensive systems and become more offensively aggressive to promote O-zone time in the hopes of scoring. As such, they are prone to counter attacks that will produce scoring chances of a higher quality from opponents. And when I mean scoring chances I’m talking REAL quality scoring chances and not some weighted shot location statistic known as SCF. Adversely, teams in the lead will tend to adopt a defensive structure electing for example to have 4 guys behind the red line and 1 fore checker in deep. This, if implemented successfully, suppresses an opponent’s quality chances for, but also limits the team’s ability to generate offense. Therefore, basic logic dictates that teams in the lead will have a better shooting % than teams that are trailing.
Turing to Carlyle, PDO was consistently and significantly better in each and every scoring situation when compared to Horachek, but the most noticeable and important difference in results were when the teams were tied. The utmost of even and unbiased analysis in hockey occurs when teams are tied as it shows an untethered strategy from both teams and determines who is actually better at implementing strategy to aggress, threaten and take control of the game. At 99.8% PDO, Carlyle’s Leafs had a far better chance of taking the lead when tied compared to that of Horachek’s Leafs who had a 91.9% PDO when tied.
Further to that, Horachek’s CF% is obviously tainted by the fact that his team is often trailing and as such playing teams that have adopted a defensive system of hockey. It’s no wonder then that shot metrics skew in his favor and in no way means that the Leafs are actually playing ‘better’ hockey under Horachek via shot metrics data. For example, Horachek has a CF% of 44.4% when leading, but he’s only leading 17.5% of the time. Meanwhile, Carlyle has a CF% of 40.9% when leading, but he is also leading significantly more often at 35.23% of the time. Hence, it’s fairly obvious Carlyle’s overall CF% will be further negatively affected compared to that of Horachek’s since his team is in the lead way more often!
The difference in PDO between Horachek and Carlyle is the result of coaching systems, pure and simple. It has very little if anything to do with luck. If PDO is measuring anything at all, it’s measuring this roster’s inability to play a structured system of defensive hockey implemented by Horachek. Carlyle worked with the tools he had, speed and finesse, to manufacture a system that would engineer wins for this team. Based on a comparison of both systems he did a damn fine job of it too. I stated in my previous article that I liked what Horachek has tried to do defensively with this team, but this roster just won’t allow him to do it.
Sure, Carlyle has nearly double Horachek’s TOI in this study, but I don’t think another 20 games is going to make any meaningful difference particularly now that this team is about to be torn apart. Statistically speaking, this team would have to improve its shooting % to nearly 20% for the remainder of the season to even come close. Not gonna happen.
One last thing, before you go. Carlyle was in a playoff spot when he got fired, granted the team was in decline and possibly may have never recovered, but in a playoff spot nevertheless. Horachek’s tenure has put this team in the lottery for McDavid. In the end, Horachek probably did more for this team than Carlyle ever could have, but for reasons no one would have ever dreamed or thought of when he first got hired.
P.S. If you haven’t filled out our survey on who should stay and who should go on this team, we would love to hear from you. The survey is on until Feb 24th whereby at that time the corresponding results will be compiled and posted in an article soon after. Compare your picks to that of the rest of Leafs Nation and see where you lie with respect to the direction of this team going forward. So far, the results are quite interesting.