• Last Post on Eberle’s On-Ice S% (and other stuff)

    by  • August 14, 2012 • Hockey • 4 Comments

    I’ve been a bit baffled by some of the comments that I’ve seen about Jordan Eberle’s on-ice shooting percentage since my post on him last week. This appeared in the comments to the post:

    What was his S% playing with RNH for approximately 50 games, versus playing with another centre? Even if RNH were to not improve his play significantly from Year 1 to Year 2, but were to play an extra 25 games as Eberle’s centre, wouldn’t that create an upswing in Eberle’s stats?

    I like to mess around with the data, even if I’m not getting a post out of it. There’s an advantage to doing this, I think – you start to get a sort of intuitive sense as to how these things work. People who don’t do that don’t entirely have it. For their benefit, I thought I’d try and make a few things clear.

    There are 58 forwards who’ve put up a 9% or better ES on-ice shooting percentage over at least 3000 minutes between 2007-12. They probably aren’t all true talent 9%+ guys. So we’re talking about five seasons. What did those seasons look like, in terms of their on-ice shooting percentages? I’ve graphed the frequency of seasons between 7.0% and 7.1% etc. in the table below.

    You can see that the bulk of them are centered around 9.7% or so (the mean is about 9.6%), with tails going in each direction. Eberle’s 12.9% each season is way out at the right end of one of the tails.

    Here’s the same data presented a little differently – it looks at the probability of a season at or below X% for the group of players who we know to be 9%+ ES on-ice shooting percentage between 2007-12.

    As you can see, even for a guy who’s known to be a 9%+ talent over that time period, the odds of a season like Eberle experienced are vanishingly slim – something 2.2% of the 283 seasons played by this group of players over the past five years saw an on-ice shooting percentage that high.

    The list of guys who managed to put up 10%+ over the past five years at ES, again with a 3000 minute threshold, is even shorter: 13 players long. Sidney Crosby, Alex Tanguay, Marian Gaborik, Steve Stamkos, Bobby Ryan, JP Dumont, Evgeni Malkin, Henrik Sedin, Steve Downie, Nathan Horton, Daniel Sedin and Jason Spezza. Some of those guys aren’t true talent 10% types – Steve Downie strikes me as pretty unlikely to be one.

    I put together a graph of the on-ice S% seasons enjoyed by forwards who were between 9%-10% and 10%+ between 2007-12. It’s kind of interesting to me how the 10%+ line is basically the same as the 9-10% line, just shoved over a percentage point.

    What does it mean with Eberle? Well, even if you think he’s a guy who’d shoot 10%+ over a five year period, his on-ice S% last year was extraordinary. On-ice S% is such a powerful thing, with small changes having such a significant impact, that it’s hard for me to imagine how it’s reasonable to expect anything approaching last year for Eberle. The areas in which he could reasonably be expected to improve tend to be difficult areas in which to improve, with past performance strongly predicting the future.

    One more way of looking at this. We know right now that the Oilers have put up a 10.2 S% with Jordan Eberle on the ice over the past two years. We have five years worth of data and it turns out that there have been 71 two year stretches in which a forward has put up 10%+ in at least 1500 total ES minutes. What if we look at what the guy did in the other three seasons? I’ve put together a graph with probabilities of S% of X in the other three seasons.

    The median is about 9.2%. I’d expect Eberle to beat that over the next three years – it’s dragged down to an extent by guys at the end of their careers like Todd White, Jason Arnott, Steve Reinprecht and Cory Stillman. I don’t think I’d expect him to beat it by much though – 68% of guys fell between 8.5% and 10.5%. That percentage would be higher if I cut some of the lower guys from the list. Basically, we can be reasonably certain that the Oilers will be above average finishers with Eberle on the ice, but that’s about it.

    Since my previous post, two things worth mentioning have occurred. First, the Carolina Hurricanes signed Jeff Skinner, who sure looks good as a comparable for Eberle, to a six year deal at a cap hit of $5.725MM. Second, Bob Stauffer dropped a hint on Twitter that the Oilers are likely to sign Eberle and Taylor Hall to new deals in the immediate future, although Ryan Rishaug has said it’s just Taylor Hall that they’re looking at signing now. I’m sort of inclined to believe the guy who’s embedded over the guy who isn’t but it’s odd that there’s a disconnect there.

    As I said in February, there’s no point getting worked up until it happens but signing Eberle now sure looks insane to me. I’m baffled by that Skinner signing because it’s hard to see where the upside is for Carolina, barring a CBA that’s dramatically different in terms of UFA age. It’s interesting to me to see how deals like those given to John Tavares – a contract given in the legitimate expectation that he was about to take a massive stride forward and based in part on his much better pedigree than that of someone like Jeff Skinner – was almost certainly used by Skinner as a yardstick. It strikes me that there’s a bit of an odd thing where contracts given to players who are expected to improve significantly are then used as benchmarks by players who’ve likely had outlier seasons.

    It reminds me a little bit of the second contracts given to Sergei Samsonov and Joe Thornton, which were identical three year deals worth about $5.15MM. Through the first three years of their careers, Samsonov played 237 games, scoring 66-77-143. Thornton scored 42-66-108 in 217 games. Through the next three years, Thornton went 215-95-145-240. Samsonov went 164-63-93-156. Samsonov was still great value but second contracts were a heck of a lot cheaper then. Since then, Samsonov’s gone 487-106-166-272. Thornton’s gone 645-187-543-730.

    In effect, by focusing on just the time in the NHL and discarding the information gathered prior to the player making the NHL, you’re discarding a lot of information. Eberle went 22nd overall for a reason. Hall went first overall for a reason. If you accept that Eberle’s had a season in which he hit at the far right end of his own personal bell curve – and I think this is inarguable, barring significant improvement in his possession numbers – it doesn’t really make sense to pay him along the same lines as you pay Hall. You’re paying for variance.

    Four other points.

    1) It’s CBA time and there’s a lot of discussion about how the “players” lost the last CBA negotiation, as if they’re a monolithic entity. They didn’t. Some players won and some players lost. Guys like Skinner won. Skinner’s going to make $5.45MM a year for his fourth, fifth and sixth season. It’s a 9.3% annual increase, compounded, over what Samsonov was making on his second contract. Guys like Jason Smith, who played their best days under the old system and then hit UFA age around the same time that the NHL started paying younger guys, lost. The highest paid player in the league was making $10MM in 2000-01. If the highest paid player today had enjoyed 9.3% annual increases, he’d be making $29.1MM in 2012-13.

    2) Robin Brownlee had a line that will go down with his JDD stuff when he wrote: “I happen to believe Eberle will be the best of the bunch in terms of offensive production.”

    3) Young Willis has suggested that the Oilers shouldn’t sign Taylor to a long term extension at this time. Assuming sensible money, this seems crazy to me. Hall took a huge leap forward last year, played tougher competition than Eberle and is clearly a straw that stirs the drink when you look at his underlying numbers. His two injuries seem relatively freaky – the shoulder is supposedly an issue that he’d had for a while which you can consult with your doctors on and the ankle problem can be resolved be telling him not to fight.

    4) We’ve talked about this specific issue a billion times with Oilers’ players. Nilsson (10.6% ES on-ice S% in contract year, 80% ES IPP), Brule (9.6%, 84.2%) and Horcoff (11.36%, 82.5%). Gregor kind of snarked a bit on me for invoking Horcoff but what makes Eberle a potential Horcoff type contract is the term and dollars associated with it. I expect Eberle will score more than Horcoff over the life of his deal but Horcoff does things that Eberle doesn’t and plays tougher competition. In any event, it’d be nice, for once, if the Oilers didn’t bet on this sort of thing.


    4 Responses to Last Post on Eberle’s On-Ice S% (and other stuff)

    1. Triumph
      August 14, 2012 at

      “I’m baffled by that Skinner signing because it’s hard to see where the upside is for Carolina, barring a CBA that’s dramatically different in terms of UFA age.”

      In my opinion, it’s a dangled carrot by a team that hasn’t won a whole lot and despite their solid moves this off-season, still doesn’t look primed to enter the NHL’s elite any time soon. Rather than nickel and dime Skinner when they can, they throw him a big contract when they don’t have to and hope this engenders some goodwill with a very talented player to want to stay in Carolina his entire career. To some degree, they used this philosophy with Eric Staal.

    2. Gally
      August 14, 2012 at

      Eberle went 22nd overall for a reason. Hall went first overall for a reason.

      I have one issue with that argument. Who cares? Alexander Daigle went first overall for a reason. Pavel Datsyuk went 191st overall for a reason. Scouts and GMs fuck up all the time. Look at our management the last 20 years. I’m in no way saying Hall isn’t a bonafide superstar in the making, but the argument that Eberle isn’t, possibly because he wasn’t drafted first overall irks me.

      Keep up the good work though. You’ve introduced me to advanced stats, and changed how I look at the game I’ve loved for so long.

      • Triumph
        August 14, 2012 at

        Oh yeah, that too. Draft position grows increasingly irrelevant the farther from the draft date we go. In a theoretical re-draft of 2008, Eberle probably goes around 7th overall, with Stamkos, Doughty, Pietrangelo, Myers, Karlsson, and Carlson going above him. But that’s a really strong draft year.

    3. Saj
      August 14, 2012 at

      Great post. Made a bet with a friend on Eberle’s goals total next year based on his high shooting% and my expectation of regression.

      He does seem to have really quick and accurate hands, though, so I’m not entirely convinced he doesn’t belong in the Tanguay category

    Leave a Reply

    Your email address will not be published. Required fields are marked *