After the first 2 games, I’m sure there are OIlers fans disappointed with the lack of goal production (and wins), but I want to assure the Oilers faithful that–so far–this is not the same Oilers team of previous season (or the season before that, and so forth). The team made much-needed changes to management and coaching, as well as making acquisitions such as Connor McDavid, Top-4 defenseman, Andrej Sekera, and the New York Ranger’s former backup goalie, Cam Talbot. Coach Todd McLellan is developing the team from the ground up and there will be growing pains. I believe that based on his success in San Jose and internationally, that he’s clearly demonstrated that he is a successful coach and for that reason, he’s ready to take on the task of molding a competitive team. Rebuild 3.0 (or whatever it is; I lost count) is a work in progress and despite the losses, I’m liking what I’m seeing from an analytics perspective. I also like what I’m seeing when watching the game, besides the giveaways that lead to goals.
Randomness Influences Goals & Wins
Because of randomness is a major statistical factor in NHL goal-scoring, assessing a team’s progress using goals plus shot metrics (a.k.a. Weighted Shots) is a more reliable measure of future performance. There is also significant randomness in winning any particular NHL game; with 38% being one of the widely accepted estimates. Hard to believe for the average fan, but that’s the nature of statistics. The numbers are not always intuitive, but they are what they are. I think there would be less hair-pulling among fans if they accept the reality of statistical facts then to deny or dismiss them. First fact: There is a lot of luck in winning a hockey game! Second fact: Weighted Shot (and shot attempt; a.k.a. Corsi) differentials at even-strength (5v5) are the strongest predictors of a team’s performance in the regular season, as well as the playoffs.
Of course, despite the luck factor in any single game, a team can improve their chances at winning over the long-term. What’s the best way of doing that? Simply put, generating more shots and suppressing more shots, especially quality shots. If a team out-shoots their opponents more often than not (assuming league-average goal-tending), over time, they will win more games than they lose. The Oilers have been unable to accomplish this positive shot differential over the last 9 seasons. This season, though, we may see this shot tide turning.
Predictive Strength of Shot Metrics
Split-season analysis (i.e., using half of a season to predict the remainder of the season) has shown that shot-attempt differentials (SAT%) and Weighted Shot differentials (WghSh%) are better predictors than goal differentials. Goals are important, but because of randomness, just because a team is scoring more goals does not mean they are generating more shots than their opponents. The team could simply be getting lucky with unrealistically high shooting percentages (e.g., Calgary Flames’ shooting% last season and the Colorado Avalanche’s shooting% the season previous). Any time a team is shooting way above the mean (i.e., 8% shooting efficiency at 5v5) in any season, we can expect their shooting% to regress toward the mean the next season. We sat that with Colorado last season and Toronto the season before that. We can also expect to see that regression to the mean with Calgary this season.
With that preamble out of the way, this leads me to how I will be evaluating the Oilers progress at this early point. In this post-game analytics, I will comparing key shot metrics from the games against St. Louis (Oct. 7/15) and Nashville (Oct. 10/15) to their respective 2014/15 season series numbers. Specifically, I will compare Weighted Shots (WghSh%; 1 point for goals, 0.2 points for shot attempts); Shot Attempts (SAT%; blocked, missed, and shots on goal), Scoring Chances (SC%; as defined by war-on-ice), and High-Danger Scoring Chances (HSC%; i.e., shots from the slot area). As usual, all my data originates from war-on-ice.
[table id=17 /]
Progress?
If we look at offense (Weighted Shots – For) from both games, and compare each game to their respective 2014/15 series, we see little difference, as indicated by the progress indices. (Improvement is indicated on the Progress Index with a plus sign, both for offense and defense. Needless to say, worsening is shown with a negative sign.) On offense, it looks like nothing has changed. The defensive metric, though, shows some improvement. But this improvement is hard to grasp because the Weighted Shot figure incorporates shot attempts and goals; wherein goals are weighted 5 times more heavily than shot attempts. What does +1.4 mean? Frankly, I couldn’t tell you.
St. Louis
To help give this improvement some clarity, we need to unpack shots from goals. Thus, examining shot attempts, scoring chances, and high-danger scoring chances becomes useful. We see that offense has improved as measured by shot attempts (SAT-For), relative to the season series, by 6.3 shot attempts/60 min. Moreover, their defense also improved by allowing 3.6 fewer shot attempts per 60 minutes. Combining these figures, we see an improvement of +4.8% in their SAT differential. That is a very good sign. If it was around 2%, let’s say, I wouldn’t be doing a happy dance unless this improvement was consistent across 20 games. But not all shots are equal. How about the quality of shots?
War-on-ice analysts have developed a refined metric of shot quality, which they refer to as Scoring Chances. The definition is quite long, so I’ll refer you to this link if you want to understand what it entails. In short, Scoring Chances have a higher chance of going in the net than simply shot attempts. Looking the progress indices, here we see that the Oilers improved both in offense (+3.6/60) and slightly on defense (+1.5/60), which results in an improved Scoring Chance differential of +5.3%.
Finally, we look at the highest quality shots; those shots that are generated from the slot. Again, if you follow the war-on-ice link above, you’ll see a diagram of the slot area. The scoring rate from the high-danger zone, if we just factor in shots on goal (excluding missed and blocked shots), is 20%! On average, a team scores on 8% of their shots on goal. This helps explain how Tampa Bay was the highest scoring team last season. They easily lead the league in shots from the slot with 25% above league-average. In generating offense, the Oilers performed slightly worse relative to the season series (-2.5 high-danger scoring chances/60). In contrast, defensively, the Oilers allowed 5 fewer high-danger scoring chances per 60 minutes.
Overall, then, the Oilers’ offense improved both in terms of shot volume and to a degree, in shot quality, except for high-danger scoring chances. Defensively, they showed improvement in all 3 shot metrics. True, in the game itself, the Blues outperformed the Oilers on all these metrics, but the upshot is that relative to their former selves, the Oilers peformed better. That’s something. Isn’t that what we ought to aspire to in life? To better our current selves relative to our past selves?
Nashville
How about the Nashville game? Again, the Weighted Shot metrics show little change to offense with a slight improvement on defense. But if we examine shot metrics, we see a different and, dare I say, optimistic story. In generating offense, the Oilers outperformed their season series on every metric: shot attempts (+4.6/60), scoring chances (+5.3/60), and high-danger scoring chances (+3/60). That’s what I expected with McLellan. In his time with the Sharks, they were not only one of the best power-play teams, they were one of the best offensive teams. (Of course, it helps to have an elite playmaker like Joe Thornton.)
How about defensively? Again, relative to last season, the Oilers improved by allowing fewer shot attempts (+6.0/60) and scoring chances (+5.6/60). However, the Oilers did allow more high-danger scoring chances (-3.7/60).
Overall, the Oilers shot attempt differential (SAT%) improved by 4.9%, their Scoring Chance differential (SC%) improved by a whopping 11.6%, and their High-Danger Scoring Chance differential (HSC%) remained about the same (+0.3%). As a bonus, the Oilers actually won two of the shot metric battles in the game: SAT% = 52.5% and SC% = 54.3%. 52.5% is actually the playoff “magic” number. Teams who average a 52.5% shot-attempt differential have a 90% chance of making the playoffs. (Not to mean I believe the OIlers are a playoff team, but I thought this was an interesting tidbit to share.)
Summary
Despite the losses, then, I think these metrics show that the Oilers are on the right track. The everyday fan wants to see wins, which is understandable, but the solipsistic focus on wins can blind us to what is happening on the ice. What we’re seeing is that the team has performed better defensively, and importantly, independently of goaltending (more on goaltending below). They appear to be doing a better job at suppressing shot attempts and scoring chances, although they are still weak when it comes to allowing shots from the slot.
Offensively, the Oilers showed signs of improvement in both games, especially the Nashville game. McLellan seems to have a system in place that is helping them generate more offensive opportunities. If this continues, we will see the goals coming. Our bad puck luck cannot continue forever. Looking at the player stats of the Nashville game, Hall was on fire with 7 scoring chances! (5 is a high, Ovechkin-like number, so 7 is exceptional.) The McDavid/Yakupov tandem also looked good and during their brief time together (TOI=6:23), their line had a shot-attempt differential of 66.7% (8 for; 4 against). I’m curious to see if McLellan keeps them together. I’ll be at the game in Dallas tomorrow (Oct. 13/15) and I’ll be rocking my Hall jersey. Hope to see McDavid pot his first one. That would be memorable.
Goaltending?
Analytically, goal tending is almost impossible to evaluate over a few games. Still, by the eye, Talbot has looked solid. One metric that I will be tracking is high-danger zone saves. Analysis by Stephen Burtch (unpublished; so I’m taking his word on Twitter) has shown that this save percentages from this zone are the most reliable over time. Thus, it appears to be a good indicator of goaltender ability. Against St. Louis, Talbot saved 5 of 6 high-danger shots and against Nashville, he stopped 7 of 9. Combining both games, he stopped 12 of 15 for high-danger zone save% of 80%. How did our goaltenders do last year against these teams? Scrivens and Fasth had a combined high-danger zone save% of 78.8%. A small improvement by Talbot, but as I mentioned, it’s still too early to evaluate. But even a 1% improvement is meaningful. Extrapolate that to 6 high-danger shots/game over 60 games, that’s 360 shots and saving 1.2% more of these is 4 fewer goals. For the record, Talbot’s high-danger save% last season was 86.17%, which ranked him 8th overall.
One thing I failed to mention is that St. Louis and Nashville were two of the best teams last season from the toughest division, the Central. They finished 2nd and 3rd, respectively, in the Western conference. In fact, St. Louis tied Anaheim with 109 points, but was relegated to 2nd place because of they had one fewer win, which didn’t go to shootout. In other words, these are elite teams. If the Oilers can show improvement against the best teams in the NHL, regardless of the losses, I’m left feeling hopeful.
Thanks for reading. Please leave any comments or questions below.
Walter
