Analytics and the Ottawa Senators: Does "Money Puck" Work?
Ever since Brad Pitt and Jonah Hill starred in the 2011 hit movie, Moneyball, analytics have become even more prevalent in professional sports usage and daily banter.
The NHL is no different. The new school of thought believes you can predict future success for a player or team by enhancing the time-honoured statistic of plus-minus at 5 on 5 by factoring in statistics like shots, shot attempts and shot blocks. By factoring in those newer statistics, it is believed that you can now predict plus-minus more effectively. There is truly enough data available online to drown a fish. However, I am going to look at the Canadian division at the team level and see if the data holds any validity.
Let’s dumb it down for a moment. We cannot talk about every analytic. I will focus on two.
The most common analytic or advanced statistical rating for players and teams is the CORSI FOR or CF rating or %. This simply adds up the total number of shots on goal, missed shots and shots blocked, both on offence and defence, while a player is on the ice, then divides the offensive number by the sum of the offensive and defensive numbers. You simply take the global totals for the team score. Anything over 50% is considered positive. It ignores situational play such as 4 on 4, 3 on 3, OT, power play and penalty killing for the simple reason that nearly 80% of games are played at 5 on 5. As well, some players do not play certain situations and this stat is meant to compare everyone relative to their peers.
Statistics, like everything else in the world, have continued to evolve over the years and newer analytics have been developed.
We will also be looking at the Expected Goals For and Against and compare to the actuals. The Expected Goals For and Against is calculated by using the shots taken by the team and assigning values to them based on where the shots were taken from. From there, you can calculate an expected Goals For (or xGF%) and anything greater than 1 is positive. For the same reasons as CORSI, the focus of this statistic is 5 on 5. It ignores all situational play.
In short, I am going to compare this:
To not only see if they make sense but to predict who will come out of the Canadian Division after round 2 of the playoffs. Of course, thanks to COVID-19, only 2 Canadian teams have completed the regular season. However, we know the four playoff teams and the matchup so the comparison is not too far afield.
Based on their CF% stats for the season, it makes sense that 3 of the 4 teams made the playoffs (Leafs, Oilers & Habs). Oddly enough, only 2 of the 4 teams making the playoffs (Habs and Leafs) have a score higher than 50%. One of the teams that is not making the playoffs (Flames) did score the 2nd highest CORSI in the division.
Based on their Expected Goals For or xGF%, again, it makes sense that 3 of the 4 teams made the playoffs (Leafs, Jets & Habs). Notice that it does not have the same teams. In this case, the Jets bump the Oilers. Once again, the Flames are the anomaly having the 2nd best xGF% in the division and not being able to finish any higher than 5th.
What sense can we make of this?
Our beloved Senators ranked 5th in the division for CF%. This appears to support where they'll finish the season. The xGF% does not flatter them nearly as much putting them dead last. Given that the Flames have 4 games left and the Canucks 5 (4 against the Flames), it is possible that the Senators could fall to 6th or even 7th if there are a lot of 3 point games. The likely finishing place will be 6th and that seems to make sense if they are 5th in CF% and 7th in xGF%. I would love to see what their scores were at the 15 game mark and the final 41 games.
The Flames appear to be the anomaly of this exercise. They are 2nd in both CF% and xGF% in the division yet they are not going to make playoffs. Why might this be? They have been to the extra frame 6 times so far this season posting 2 wins, 2 losses and one each in the shootout. Their special teams are respectable with the 3rd best power play and 2nd best penalty kill in the division and yet they will finish 5th at best. Based on the stats, this should be a playoff team. It would seem numbers cannot explain everything.
CORSI and xGF focus on 5 on 5 play. The Habs have the highest CF% and 2nd highest xGF% in the division. The Habs were abysmal in overtime, losing 9 games and another 2 in the shootout. Their ability to get the loser’s point in games is why they are in the playoffs, unlike the Flames, and likely why the scores in those categories are more flattering. It is also likely why they will go no further than that. Having shown no ability to win in the extra frame, the lowest ranked power play and 3rd ranked penalty kill of the 4 teams making the playoffs and having no shootouts in the playoffs, look for the Habs, despite their analytics, to exit quickly.
The Jets were 6th in team CORSI % and 4th in team xGF% yet they finished 3rd in the division. How is this? Again, looking at the stats that are not considered, the Jets won 6 games in OT while only losing 2 and losing 1 in a shootout. Unlike the Habs, they showed an ability or flare for the dramatic and with no shootouts to lose in, this gives them an edge. They are also 2nd in power play % and 3rd in penalty kill among the 4 teams making the playoffs. Given their level of experience, they are poised to make a run in the Canadian division despite not having bolstered their lineup at the trade deadline.
The Oilers CF% was 4th in the division and the xGF% was 6th. This might seem like an anomaly compared to where they finished in the division, however, looking at their special teams, it is easy to explain the delta. Having the league’s best power play and being 2nd among Canadian teams in the playoffs on penalty kill explains how they have managed to get by. The Oilers only went to overtime 6 times this year and won 4 of them having never gone to a shootout. Does this mean that they are poised to for a long playoff run? I’m afraid not. The Oilers are a Pig in a Prom Dress. Beware the bottom feeder. How can a team who takes 18 points from an opponent giving none back not finish in 1st place in their division? It is simple, they fed on the Senators, in particular early on, and this inflates them much like the Habs ability to get the loser’s point. They had a respectable record in overtime. However, overtime is not played at 3 on 3 and their speed will not be able to exploit the open ice. Look for the Jets' experience and better goaltending to trump the analytics and give the Oilers a 1st round exit.
The Leafs are an interesting case. They were 3rd in the division in CF% and were landslide winners in the xGF% and 5 on 5 differential. This would point to potential success if not for what these analytics do not cover. The Leafs went to the extra frame 13 times, losing 7 of them and winning 1 in a shootout. This is not alarming but not comforting either, given neither goalie has enjoyed playoff success. The Leafs were 3rd in power play and 4th in penalty kill among the 4 teams qualifying for the playoffs. This will ultimately be their undoing. They will survive the Habs but no further.
That’s right. I am calling the J-E-T-S JETS! JETS JETS! to win the Canadian division based on everything analytics in CORSI and xGF does not tell us. They also have the best goaltender and best coach.
As for analytics and their role in predictive analysis, I think this example shows that CORSI and xGF can be used, at best, as a guideline for predicting success. If 80% of the games are played at 5 on 5, clearly the other 20% of the game decides a lot higher % of the games. I think the playoffs will show that the stats that matter most are special teams and the ability to deliver in the clutch. Basically, when CORSI and xGF are not on the clock, that is when the real winners emerge.
As a side note, one of the criticisms of the old plus-minus statistic is that it's misleading and not always accurate. Looking at the formulas that go into calculating the advanced statistics being reported, I am led to wonder how any of it can be done reliably. If we cannot reliably know who is on for a goal against or for and if it is even strength, power play or shorthanded, what possible hope can we have in calculating shots taken, shots missed and shots blocked for a team or player at the pace of today’s game? I keep imagining Despicable GRU and his minions in underground laboratory churning out these statistics.
To me, statistics has evolved to the point where we are drowning a fish as I eluded to earlier. The two stats I brought forward are among the most commonly relied upon today. They are not exclusive. Are they irrelevant? No? Should the Senators invest more beyond Tom Gillis, Director of Data & Analytics, to get to the next level? Absolutely not. At most, the data should be used to compared with what the eyes and ears have seen and heard. There is no substitute for hard work and the work being done around analytics strikes me as an illusion, which diminishes the value of pro and amateur scouting.
BOOTS ON THE GROUND! THERE IS NO SUBSTITUTE.
By Pat Maguire | Sens Nation Hockey