Month: December 2016

Weekend Q&A: McDavid’s Wingers, McLellan’s Short Leash, and the Rogers Place Crowd

/ sunilagni / Leave a comment

coppernblue.com.full.54273Below is part of the Weekend Q&A my colleague Minnia and I put together over at The Copper & Blue.

Minnia: With McDavid being paired with so many different wingers, especially on the right side, who have you liked best so far?

Sunil: This guy is so incredible that it doesn’t even matter who his linemates are. Usually we look for which two or three players have chemistry and how they can work off of each other. With McDavid, you look at who needs to play with him to get going or which player’s potential you can maximize. At the start of the year, I would have pushed for Maroon to be McDavid’s left side instead of Lucic, just based on Maroon’s 16 games with Edmonton last season. And I would’ve wanted Eberle on the right side, as he had great chemistry with McDavid and has the speed and finish to produce on the top line with him. Today, I’d rather have Maroon playing on other lines, as he’s been a positive influence on others and has been able to generate goals with McDavid on the bench, something that remains an issue for the Oilers. As of today, I’d keep Lucic with McDavid as they’ve been productive together, and I’d add pretty much any of the depth right wingers to that line. It wouldn’t hurt to have Puljujaarvi there, since the Oilers have no idea what to do with him, or even Slepyshev who has the skillset to fill a complementary role. Whatever the Oilers do, they have to ensure that the other three lines behind McDavid are loaded up so they can start producing goals. This means they may have to have the more depth players on the first line with McDavid.

Minnia: It seems McLellan is quick to “punish” this season, whether it’s reducing ice time drastically for mistakes in-game (Klefbom, Larsson) or sitting players for extended periods of time (Pouliot)– do you like his approach in leaving very little room for mistakes, or would you appreciate some more leeway for players?

Sunil: I think it’s ridiculous how McLellan has been handling/punishing his roster to be honest. A coach is responsible for putting together optimal line combinations and defence pairings, and then properly deploying them to increase their odds of winning games. If a player is struggling within a game, it makes sense to give ice time to the players you’re trusting. But after the game, re-set, work with the player, and base your future decisions not just on one bad play, but a larger sample size. Klefbom is a prime example here. He’s done very well paired with Larsson this year, all areas of the ice, against the other team’s top lines. The duo have a 51.79% share of all of the shot attempts when they’re on the ice this season, which is great considering the competition they face (Source: Corisca Hockey). Their 345 minutes together is the most on the team, so why not stick with that? After a bad play last week, Klefbom still isn’t back with Larsson, who is now getting caved in when it comes to shots against paired with Russell.

Minnia: Taking a look at Bakersfield, who are some of the players on the cusp you’d like to see up on the Oilers in case of injury (#knockonwood)?

Sunil: I would love to see Jujhar Khaira with the Oilers at some point this season. He’s a good skater and showed last season in limited minutes that he can keep up with the more established players. In 2015/16, with Pouliot out, Khaira skated with RNH and Eberle for a total of 53 minutes at 5v5, and played the same level of competition that RNH is accustomed to. The trio had a 50% share of the shot attempts, which was great considering the rest of the team was below 48% for most of the season. Long term, he should be a depth centerman, but it wouldn’t hurt to continue breaking him in on the wing. This season in Bakersfield, he has 5 goals and 5 assists in 12 games and ranks second on the roster when it comes to points per game (0.83). He’s also one of the team leaders in shots per game (3.1).

Minnia: What do you make of the interesting trend that the Oilers seem to lose a lot of games when they win the Corsi battle, yet win a lot of games when they lose the Corsi battle?

Sunil: A lot of that has to do with score effects. Basically, when a team is trailing in a game, which the Oilers have done often recently, they increase their output of shots as they start taking more risks and might shorten their bench and get their top players out there to find offence. On the flip side, when a team is leading, they often slow down their offence, start playing a safer game, off-the-glass-and-out tactics in an attempts to stifle the shots against. At the end of the game where the Oilers lose, it might look like they were outshooting the other team, but that usually happened after they were trailing.

Full article is at The Copper & Blue.

Correlation between shot-share and goal-share

/ sunilagni / 1 Comment

Here’s a graph showing every team’s regular season performance at even-strength (5v5) from the 2007/08 season to the 2015/16 season (Source: Corsica Hockey). Getting a higher proportion of the total goals scored for and against is the key to winning games, so that’s displayed in the vertical axis. On the horizontal axis is the team’s share of all shot attempts (i.e., Corsi). Each dot represents every single team by season since 2007/08, with the lockout-shortened 2012/13 season excluded since each team only played 40 games. Comparing a 40-game season to an 82-game season might give us unreliable results (team’s can get hot or cold over a shortened length) so I left that season out and compared apples to apples.

goal-and-shot-share-2007-2016

The black bars that cut through the graph horizontally and vertically mark the average across all the teams over the eight seasons. Not surprisingly, it’s right at 50.01% for both goal-share and shot-share. I’ve also added a trend line, which has a r-squared value of 0.32. It’s obviously not a perfect correlation between shot-share and goal-share, but you increase your chances of scoring if you’re outshooting your opponents. Teams that posted a higher than average shot-share but didn’t see their goal-share get higher than 50% (bottom right quadrant) often had external factors at play. It could’ve been injuries to key players or a lower than normal shooting or save percentage. On the flip side, team’s might have a lower than average shot share, but still outscore oppoenents (top left quadrant). This would be due to a higher than normal shooting percentage (shooters get hot) or team save percentage (goalie gets hot). Hockey is weird like that.

Highlighted in the graph are eight orange dots with a black ring around it. These are the Oilers eight seasons, and not surprisingly they’re all in the bottom left quadrant where you don’t want to be. They’ve been terrible at outshooting opponents, and as expected, they haven’t been able to get a higher share of the total goals except for that one time in 2008/09. They had one of the worst shot-shares in the league that season (46.35%), but thanks in large part to a slightly higher than normal shooting percentage, and some great goaltending from Roloson (a 92.45 save percentage), the Oilers managed to get 50.9% of the total goals.

This season the Oilers would be in the top right quadrant, as they’re getting 52.00% of the total shot attempts, which has them 9th in the league, and are also getting 52.88% of the total goals (i.e., goal-share). Generating shots is a repeatable skill and is dependent on the coaching staff’s on-ice tactics and the skill of the players. A team’s ability to generate shots very often remains consistent through a season, and season-to-season, as long as the coaching staff remains and players stay healthy. The Oilers have a good thing going with their ability to out-shoot opponents, but this in no way guaranetees goals. A lot has to happen for a shot to become a goal, and goalies can always go on hot and cold stretches. But as long as they keep their share of shot attempts up, they’re putting themselves in a position to succeed.

Thoughts on the Oilers: Current state, Eyeballs first and Defence core

/ sunilagni / Leave a comment

cutAfter 28 games, the Oilers are sitting 3rd in a weak Western conference, 1st in the Pacific, with a record of 14-10-4.  They’ve outscored their opponents 55-49 at even-strength, a 52.88% goal-share that ranks them 8th the league. Their 55 goals ranks them 4th in the league, while their 49 goals against is the 7th highest. The Oilers currently have a 52.34% score and venue adjusted Corsi, which ranks them 9th in the league behind Boston, LA, Carolina, Washington, Florida, San Jose, St. Louis and Nashville (Natural Stat Trick). A team’s share of shots tends to remain consistent through a season as it’s influenced by the coaching staff’s tactics and player deployment, as well as what the player’s actually do on the ice. Really the only way the Oilers might see a drop in their shot-share is if the coaches change or if the team loses significant players.

There’s no indication that the Oilers are relying on luck-driven factors including shooting percentage and save percentage. The Oilers have a 7.74% shooting percentage and a 92.63% save percentage, which ranks them right around league average.  Quite often, team’s go on hot streaks either converting more of their shots into goals or preventing more shots from becoming goals. If either of these factors is significantly higher or lower than the league average, it’s worth monitoring as team’s tend to regress to the mean over time. I do think this team can compete for a playoff spot as long as McDavid and Talbot stay healthy. They can generate scoring chances at 5v5, ranking 4th in the league when it comes to high danger shot attempts (11.65 per hour), but they also allow a lot of chances against, allowing the 3rd highest (11.34 per hour).

Please note that I look at 5v5 play to assess a team and individual player performance as it’s when both teams are trying to score and trying to defend. Special teams play is important, but they involve specific players doing a specific task, which would skew the overall numbers for the team and players if included.

Couple thoughts:

The comment from Todd McLellan a few weeks ago that a lot of anti-stats people loved, but that I found very amusing, was the whole “eyeballs first….” thing. This implied that people who do statistical analysis or analytics aren’t watching the games enough and that their work is not well informed. Here’s the thing: watching the game and doing the analysis should always be dependent on one another. What you see/hear/read informs your analysis, but your analysis informs your viewing of the game or topic as well. That’s how it works in any industry that relies on stats and analytics, with hockey, a game, a goal-scoring contest, being no different. If you’re really only doing stats after, you’re at risk of overlooking key information or letting confirmation bias set it. It’s a recipe for failure. Can you imagine a real-world CEO having an “eyeballs first” perspective on his or her business? They’d be ridiculed.

With the upcoming expansion draft and the fact that a number of key players including McDavid, Draisaitl and Nurse will be needing contracts, it’s a bad time to start over-valuing depth defenceman like Kris Russell. I really don’t mind Russell as a depth player, but he’s been way over played by this coaching staff, and the underlying goal and shot metrics don’t paint a pretty picture (OilersNation). The team would be better off if  he was playing at his established NHL level, but the coaching staff has other plans (which includes reducing Klefbom’s minutes, awesome). The Oilers have to protect three defencemen this upcomg off-season, and it should be Sekera, Larsson and Klefbom. Signing Russell in January means someone is moving out, either one of Sekera, Larsson or Klefbom, and the future contract for one of the young players is being impacted. I don’t think it’s worth getting into it with Russell, and that they could move Russell at the deadline, especially if players get healthy and he’s been bumped down the depth chart by then. My ideal pairings: Klefbom-Larsson, Sekera-Benning/Fayne (7th defenceman) and a combo of Nurse/Davidson/Gryba (8th) for the third pairing.

Just want to throw this out there, but if anyone has any questions about analytics, feel free to email me (sunilagni23 at gmail dot com). There’s a lot of bad information floating around, so if I can be of any help, let me know and I’ll try my best to answer it. I’ve been following the discussion around stats and analytics for a long time now (learned a lot from Vic Ferrari, JLikens, Dellow, Zona, Reynolds, Battle of Alberta, Parkatti, plenty more), so I’ve seen the same cycle of topics come over and over again.