Projections for Rick Nash on the Rangers
Eric T.
July 30 2012 07:23AM
By 5of7 (Rick Nash), via Wikimedia Commons
Corey has already laid out some Rick Nash facts. Now that we know where he'll be playing next year, I'm going to take a stab at projecting how his production might change if he is on a line with Brad Richards.
To do that, I'll make estimates at how much the following factors might adjust his performance up or down from last year:
- How much more (or less) ice time will he get? (Affects goals and assists)
- How many more (or fewer) shots will his new line generate? (Affects goals and assists)
- How did last year's shooting percentage compare to his established career performance? (Affects goals)
- How should we expect his new linemates to affect his shooting percentage? (Affects goals)
- How much better (or worse) are his new linemates' shooting percentages? (Affects assists)
- How will his power play production change? (Affects goals and assists)
Obviously this will require some guesswork, and I'm not arguing that the season will play out exactly according to my arithmetic, but working through the results in this manner gives us a good baseline expectation for what is reasonable and for how important each of these factors is. Let's see where it takes us.
The factors
Ice time: Last year, Nash got 14:20 of 5v5 ice time per game. Richards got 15:13 last year, so if Nash gets boosted to that, we can expect a ~6% increase in his 5v5 production from ice time.
Shot rate: Last year, Nash's team got 46.9% of the shots when he was on the ice. The Rangers got 49.1% with Richards on the ice, so we can expect a ~5% increase over last year's 5v5 production from being on a line that generates more shots. (Note: there's no particular reason to assume Nash's presence will raise this number, since as Corey has shown, he hasn't appreciably improved his linemates' shot differential in the past.)
Shooting percentage regression: Last year, Nash scored on 7.3% of his unblocked shots at 5v5, continuing a steady decline in recent years from 10.5% to 9.3% to 8.4% to 7.3%. Some of this might be due to aging, but a significant fraction is likely just bad luck; let's assume his true talent at this point in his career is something like 8.5%.
Linemates impact on his shooting percentage: Richards is one of the elite playmakers in the league, the rare player who appears to boost his linemates' shooting by 1%. That assessment didn't include missed shots; factoring those in, the impact is about 0.7%. If we assume Nash will be with Richards most of the year, we might put him at about 9% -- good for almost a 25% boost on his 5v5 goals over last year.
Linemates shooting percentage: Nash's teammates shot 8.4% last year when he was on the ice, and the Rangers shot 8.7% when Richards was on the ice. That difference might be good for a ~3% increase on Nash's assist total.
Overall at 5v5, last year Nash had 21 goals and 16 assists. With all of these factors working in his favor, we might project a total for the coming year of something like 29 goals and 18 assists. Most of these gains come from a projected improvement in the notoriously variable shooting percentage, but a small boost comes from just playing on a better team.
The power play is even more fickle, but if his 3:20 of 5v4 ice time per game goes up to the 3:53 that Richards got last year, that's a ~16% increase. His shooting percentage at 5v4 the last two years has been terrible (4/53 and 5/54, and we're excluding missed shots now), as has his team's shooting percentage with him on the ice (8.59% and 8.53%). But if we assume that is just variance, we might guess that he'll rebound to his career ~13% 5v4 shooting and his teammates will also shoot better. Between the extra power play time and the improved shooting, we might see his 4 goals and 11 assists last year become 8 and 15 this year.
Finally, we need to factor in his health. Nash played 82 games last year, but it was just the first time in his career. He's averaged about 78 games per season in recent years, so let's shave 5% off for missed time.
The results
We're left with projections of 35 goals and 31 assists in 78 games played. That might be a bit optimistic, since it assumes his ice time increases significantly and some troublesome trends in shooting percentage at both 5v5 and 5v4 reverse themselves, but it's a believable figure.
Since every step of this process involved some assumptions and guesses, let's look at how the answer would change if any given one of those assumptions turns out to be wrong:
- Same ice time per game as last year instead of what Richards got: 33 G, 28 A
- Rangers get 51% of shots with Nash on ice instead of 49%: 36 G, 32 A
- Nash shoots 8% at 5v5 (including missed shots) instead of 9% (still up from last year's 7.3%): 32 G, 31 A
- Nash stays healthy for 82 games instead of 78: 37 G, 33 A
- Nash continues to underperform on the power play: 32 G, 28 A
Of course, all of this assumes there's no significant talent decline this year. Given his age, there will probably be a small decline, but for the most part, it seems reasonable to put him at around 65 points.
I've seen people argue that playing with a good team for the first time will make him an 80-point player, but I don't see how to get to those numbers without some extremely optimistic assumptions -- he'd have to match the best shooting percentage of his career and turn a line that got 49% of the shots last year into one that gets 55% this year. It's possible, but it's certainly not the midpoint of the possible range, which is what a projection is meant to capture.
So what do we learn from this exercise? Nash had 67, 66, and 59 points the last three years in Columbus, but we project that playing with an elite playmaker on a top team will only get him to 65 points in the coming year. In the end, team effects turn out to be small -- his line might push play forwards 5% more, but that's negligible when his shooting percentage is down 44% from its peak.
In other words, players make the team good, not the other way around.
Recently by Eric T.
I'm glad to see the impact of linemates (or lack of) is finally getting some depth from the advanced stats community.
I've always been a believer that linemates had a negligible effect on a players production (say 5% - 10% on the extreme), but a lot of people seem to think ridding shotgun with an elite player will add 30 points to your season.
Great point about players making the team rather than the other way around. The problem, it seems, is that Nash just isn't as good as people generally believe.
Did you have any particular reason for putting Nash mostly with Richards at evens? It seems possible to me that Tortorella will give him an OZ push like he did with Gaborik and Stepan (perhaps even alongside them). I don't know that it would help his numbers much (if at all), but Corey did show that he had an excellent scoring rate at evens when he was put in that situation by the Jackets in 2010-11.
Great work Eric. More proof trading Pavelski (who's averaged 67 points per 82 games over the past two seasons and provides so much more two-way value than Nash) would have been an incredibly bad move.
As for the power play estimates, doesn't it also make sense to take into account the relative strengths of Columbus and New York's power play units? The Jackets were fourth in the NHL in 5v4 SF/60 last season, the Rangers were fourth-worst by that measure. I think it's reasonable to predict Nash will not be able to match his PP shot total from a season ago.
@Scott Reynolds
Not really; I was just thinking that putting him with the best linemates possible should show the largest effect possible and also most directly address the claim that playing with a good team would impact his results.
It's definitely possible that he'll see more OZ starts than the 54.4% Richards had last year, which is why I included a 51% Fenwick option as one of the alternative scenarios near the end. But unless the zone starts get extreme enough to boost Fenwick to the ~54-55% range, it shouldn't even be a 10% impact on his point total.
I think this is true; he's definitely a top-tier talent, but not even in the conversation for best scorer in the game. Nash started his career with a 41-goal season on 15.2% shooting and a 31-goals-in-54-games season on 18.2% shooting, and a lot of opinions were formed. Those opinions are hard to change, even as over the next six seasons he shot 11.9% and only cracked the top 10 in goals once.
@Derek T.
I considered looking at power play shot rates, but I got nervous about that -- scorer effects (MSG, augh) could affect the SF/60 numbers, and I can imagine that a player like Nash drives shot rates on the power play significantly even if he has less effect at even strength.
"The problem, it seems, is that Nash just isn't as good as people generally believe."
He's big, Canadian and he can dangle.... the perfect storm for overating a player.
I'd take the over on Nash at around 65 points in a heartbeat. Who do you think is likely to set Nash up for not just more goals but easier goals? David Vyborny, Kristian Huselius or Brad Richards? Nash somehow managed 40 goal seasons with Vyborny and Huselius as the set up men on his team and Richards is light years ahead of both of them as a passer. Nash has been playing on a team where he is the only guy other teams have to focus on shutting down. Now that he will be surrounded by better players and other scoring threats, he's not going to face the same level of checking that he had to deal with in Columbus. You can crunch your situational numbers all you want but they don't tell the true story in this case.
@Captain Obvious
So to summarize:
Yes, we know that Brad Richards' linemates over the last four years shot 9.5% without him and 10.6% with him, but you expect his impact on Rick Nash's shooting percentage to be much larger than +1% for some reason.
In fact, it sounds like you expect him to do better than the measly 40 goals he scored with Vyborny and Huselius. So if he has another ~300 shot season, I guess you're expecting him to shoot something like 14%.
This is where I suppose I should point out that the last two years he shot 10.5% and 9.8%, so you're looking for a 4% boost from playing with Richards and regression. You can refuse to crunch your situational numbers all you want, but your opinions about the importance of Richards' passing skills don't tell the true story in this case.
In 2010-11, Ryan Callahan shot 12.8% and scored 0.38 goals per game. Then he got paired with Richards last year and shot 12.3% and scored 0.38 goals per game.
In 2010-11, Loui Eriksson shot 15.1% and scored 0.34 goals per game. Then Richards left and he shot 13.9% and scored 0.32 goals per game.
Can Nash hit 40 goals? Sure, he definitely has the talent to make that a possibility -- shooting percentage is highly variable and he's had a couple of 15% years. But it's definitely not the midpoint of the range, and if he does, it's not going to be because Richards magically made him into a 14% shooter.
Let's mark this down and see what you have to say at the end of the year.
It's like people never learn... every year, some random angry fan pops up on one of these prediction posts and just goes off 'blah blah not the whole story blah blah watch the games blah blah numbers don't tell every thing blah blah...' The Wild last year, Stars the year before, Avalanche before that... I mean at some point people should start realizing that these advanced stats predictions are fairly accurate. Even if they don't understand the reasoning (as Captain Obvious obviously doesn't), why argue with something you should know by default is most likely gonna be right??
@OB1 Team Yakopov - F.S.T.N.F
I think the WOWY work that's been done is in-depth.
You're all obviously missing my point. Who did the opposition have to focus on when they played Columbus? Rick Nash and............ That's it. Just Rick Nash. Doesn't it make sense that maybe his shooting percentage would drop when he rarely gets a chance to get all alone in a good shooting position because he's the focus of the best checkers that other teams have to offer? "Last year, Nash scored on 7.3% of his unblocked shots at 5v5" Where were those shots taken? Where they all in an actual scoring area? Were 20% of those shots low percentage slappers from the point as he went off for a line change? Anyone can lay down stats to support an argument but that doesn't mean they're accurate. You talk about his percentages declining possibly due to aging. He just turned 28. He's still well in his prime. What kind of prediction did you make for Kovalchuk last year? He came off an uninspired 60 point season and he's even older than Nash. Did the "numbers" support his 23 point increase? @James.P - If these advanced stats predictions are fairly accurate you should be able to provide a few links to where we could see this accuracy from past seasons. You're right about people popping up and commenting on these "accurate" prediction stories because people write them constantly. What you don't see is comments on the articles about how accurate last years predictions were because they don't exist. In 6 months no one will remember this article because these carefully selected numbers that back up the predictions are about as accurate as Nostradamus. The guy made barrels of predictions and people twist a couple of them to fit where they want it and all of a sudden he's an accurate prognosticator. I could write an article predicting players point totals based on a video game simulation and it would be as accurate as this one but without the very important stat breakdowns but just like these numbers, a video game wouldn't tell the whole truth. Nash is a legitimate superstar and not picking the over on a 65 point prediction is like betting that someone will be breaking the 150 point mark next year. Bad cash management.
@Captain Obvious
Wow, that's a big wall of text. Allow me to parse it out.
And you're missing the counterpoint that we've accounted for that already. When people play with Brad Richards, their shooting percentage is 1% better than when they don't. That's the effect of moving from playing on a second or third line in Dallas to playing on a top line with Richards.
If you're going to argue that Richards' effect on Nash will be much more than a 1% boost, then you have to argue that guys like Huselius and Voracek are much worse than the second or third line guys who people played with in Dallas when they weren't with Richards. That doesn't seem reasonable to me.
I guess the argument you're trying to make here is that because he played with bad linemates, he was forced to take a bunch of low-percentage shots? If that's what you're trying to say, then I refer you to the argument I just made above, that those linemates weren't much worse than the guys people played with on Dallas when they weren't playing with Richards.
And for what it's worth, please remember that I do think his percentage last year was anomalously low -- you'll notice that in my projection I assumed his real talent is 8.5%.
And yet you've laid down no stats, accurate or otherwise, or video, or historical examples, or anything. Anyone can make up claims out of thin air to support an argument, but that doesn't mean they're accurate.
25 is a scorer's peak. See: http://www.arcticicehockey.com/2010/1/21/1261318/nhl-points-per-game-peak-age
By 28, shooting percentage is clearly on the decline. See: http://www.arcticicehockey.com/2010/4/15/1418423/shooting-percentage-vs-age
Obviously these are population statistics, and some individuals age better than others. But Nash going from 10.5 to 9.3 to 8.4 to 7.3 over the last four years doesn't suggest he's one of the people on the good end of the aging curve.
I didn't make a prediction for him, but if I had, I certainly would have predicted a sharp increase in points. After all, it was widely observed that the Devils were having a historically unlucky shooting percentage, see http://www.arcticicehockey.com/2010/12/28/1896389/poll-of-the-day-new-jersey-vs-dallas for example. The team shot 7.4% with Kovalchuk on the ice that year, after shooting 12.0% the previous year (and 12.1% the year before in Atlanta). That's why his assists were so low, and it was perfectly reasonable to predict that it would come back up using exactly the same method I've used here.
Also, let's note that Kovalchuk's shooting percentage has gone 18.4% - 15.6% - 14.1% - 12.7% - 11.9%, making him not the best example you could choose to support an argument that Nash will have a sharp rebound in shooting percentage this year.
Sure. I already gave you a link about the Devils outperforming the Stars last year. Here's one about the Wild collapsing: http://www.arcticicehockey.com/2011/11/12/2557194/the-columbus-blue-jackets-are-not-the-worst-team-in-the-league
Here's one about the Kings being much better than people thought: http://www.broadstreethockey.com/2012/3/27/2903536/nhl-playoff-preview-score-adjusted-fenwick or http://www.broadstreethockey.com/2012/4/9/2936018/shot-differential-score-adjusted-fenwick-nhl-playoffs
You could do the prediction that way, yes. But it wouldn't be as accurate. If you're equating Nash being under 65 with someone scoring 150, then I think you're hopelessly ungrounded in reality.