Tuesday, October 07, 2008

The Outlier

Consider, if you will, the creature called Tanguay. What do we know of him? Not, as it turns out, very much. He is a being from the Western Conference, which we (particularly after just seeing the recent Oilers-Flames preseason game) like to think of as ‘Bizarro Conference.’ For all we know, it’s entirely populated by sentient capybaras and Sheldon Souray. This is only a slight exaggeration.

Of Tanguay, we know the following: He is a consistently reliable even-strength player with moderate-to-high offensive potential with whom the Calgarian notables were displeased, apparently because the offense he produced was more in the range of ‘moderate’ than ‘high’. Essentially, he is an improvement on Ryder- a guy who will put up some reasonable number of points without needing too much handholding (or power play time) to do it and without sacrificing defense. Carbonneau, one assumes, will love him to bits. Moreover, while he will certainly feel the unique pressures of being a local boy playing in Montreal, the Canadiens start this season as an offensively rich team whose major strategy relies on having diversified scoring from many different players, so there will be no particular pressure on him to drive the team’s point totals. All in all, an excellent acquisition, particularly given the paltry exchange and the short remaining contract.

This, however, is not the interesting part. The interesting part is that Alex Tanguay is a statistical outlier of freakish dimensions.

Consider, if you will, the shooting percentage. It is not a beloved statistic. In the great hockey-data war between those who favor advanced statistics (N.B. ‘advanced’ generally means take whatever raw number you’re interested in, break it down according to PP and EV, and then divide by hour of ice time) and those who like their simple counting numbers, the shooting percentage belongs to neither camp. It sits contentedly on the sidelines, referenced often but seldom discussed in any great depth, largely because nobody seems to know what it means. Is it a measure of skill or of luck? Is it driven by the talent of a player or overdetermined by the laws of probability? Can it be controlled or manipulated, harnessed towards a given end? Or is it merely one of those random bits of trivia that just happen to be calculable?

Traditionally, shooting is certainly believed to be a function of free will, or at least of training. There are good shooters and bad shooters, right? On the one hand, there’s the sniper, the Kovalev, who picks his chances carefully and threads shots through improbably tiny holes with laser-precision. On the other hand, there’s the average grinder who just whacks whatever gets to his stick in the general direction of the net and hopes for the best. We know this, don’t we? It’s so logical, it’s so visible, it’s the clear evidence of our own eyes and a good many years of conventional wisdom.

It’s also mostly bullshit. NHL-level shooting percentages don’t tend to vary all that dramatically, and tend to be uniformly lowish numbers. Last season, according to MC79, the overall for forwards was 10.8%, for defensemen around 5.3% (defensemen, obviously, tend to take lower quality shots). The mean is higher for the season’s leading goal-scorers, since a good shooting percentage tends to drive searing success (Mike Ribeiro, for example, shot a ludicrous 25.2%, nearly 10% higher than his career average). But the best shooting percentages don’t necessarily correlate with the most goals- among the top goal scorers, it varied all the way from 10% (or lower than average) to over 20%, while several of the best percentages belonged to guys who didn’t even clear 20 goals. You can have a lame shooting percentage and still get the puck in the net a hell of a lot, similarly you can have a fantastic shooting percentage and still be considered offensively mediocre.

But what’s more curious is that the guys who you’d imagine to be fantastic shooters often aren’t. Make a list, off the top of your head, of twenty or so guys who you’d consider to be ‘snipers’ or ‘natural goal scorers’, and then go look up their shooting percentages. There’s a good chance that a lot of them have no better S% than those you’d consider relatively unskilled grinders. There is, perhaps, a certain logic to this- different styles of game may achieve similar results- but the perplexity is that no particular style seems to correlate closely with high shooting percentage.

It is not surprising, then, that when shooting percentage is discussed it tends to be as a ‘regression to the mean’ statistic to explain why a certain season’s results are unlikely to repeat. Over time, a given player’s shooting percentage tends to rise or recede to his average, and individual player averages in turn tend to rise or recede towards the collective average. So one might hear it said that Jason Blake will improve next season because his 4.8% 2007-2008 shooting percentage is just as freakishly unsustainable as Ribeiro’s 25.2%.

Which brings us back to the curious Tanguay, who has a career shooting percentage of 19.43. To give you an idea of what that means, that is the 8th best shooting percentage in modern NHL history. No other active player is in the top 20, and most of the players on the list scored their goals in the 80s, when goalies were made of recycled tube socks and nets were vast and promiscuous. In a statistical category that has few, if any, sustained outliers, where the law of averages is king, Tanguay beats the odds by a stunning margin and does so consistently. It means, perhaps, nothing, since he evidently takes fewer shots than most of his competition- otherwise he’d be rubbing elbows with Ovechkin on the stats sheets, rather than Mike Green, but it is a suggestive mystery. How? Why? What on earth is so blessed about Alex Tanguay that his shots are almost twice as successful as those of a regular forward, and fifty percent better than most exceptional forwards? Is it training, skill, hockey sense? Is he just congenitally lucky? What, indeed, the fuck is up with that?


Simonus said...

This probably won't illuminate the issue, but one problem with shots on goal is that it doesn't take into consideration all the times a player shoots *at* the net, but misses.

When a player that has a low number of shots and a high shooting percentage, it might indicate that he is actually shooting the puck often, but is missing the net. If a guy is shooting consistently at the corners of the net, he is likely to miss the net a lot (and avoid lowering his shooting percentage), but when he actually gets the puck on the right side of the post, it is more likely to go in.

Check these articles out:


CheGordito said...

Good points E and simonus. Tanguay is an anomaly - so my hopes were even higher when he joined the Canadiens!

One point I'd like to bring up is examining what constitutes a shot. Maybe some players are in the 'right' place near the net to scramble for the puck. If they touch it, they have a chance to 'shoot' or guide the puck towards a net, resulting in either a goal or block, or miss the net or miss the puck altogether.

Maybe Tanguay makes such an effort to shoot the puck well that when he does reach it in those situations, it goes in most of the time when he reaches it. And maybe he misses it altogether (and is subsequently not counted) more often than other players.

olibou said...

Two things:

a) Simonus rises a very interesting point: good shooters shoot at the sides, bad shooters shoot at the net and hope for the best. I'd add that good scorers know they are good, and their team know they are good at shooting. For a very good shooter, that translates in an absolute ton of SOG. What differentiate Ryder from Ovechkin is very apparent when you look at the total amount of SOG at the end of the season. With passeur extraordinaire Koivu and plenty of Power play time, Ryder ended with an average of 226 Shots on goal between 2003 and 2007. Ovechkin? 426! A good scorer isn't simply an efficient shooter. A good scrorer is a prolific shooter. Most probably, a good scorer is, first and foremost, a prolific scorer. (I'm looking at you, Higgins)

2) A good player is a good player is a good player. I suspect there is some kind of sweet spot for those fine passers with strong possession ability. They don't generate a lot of shots because puck possession is mostly synonymous with absence of shooting angle. But they *are* talented. So when they shoot, it's usually in a situation that is far more likely to end up in a goal. If Tanguay is on the left side with some weird angle, he'll keep the puck, cycle a bit and feed a pass to somebody venturing in the slot (I'm looking at you, Higgins!). If Andrei K is in the same situation, he's more likely to shoot (and suceed). Tanguay probably has a good S% because he doesn't shoot that much (around 115-120 SOG a year), only pulling the trigger in high-reward opportunities, opportunities that he is very good at generating.

I think we won't find it hard to like him.

Matt said...

Flames fan here! Subjectively, Tanguay has a high Shoot% because (A) he has an accurate shot, and (B) he doesn't shoot if he doesn't think it has a chance of going in; either the goalie has to be un-squared, or he has to be close in enough so that an accurate shot goes in and the goalie can't possibly react fast enough to stop it.

He's an interesting player. Your typical great passer relies on deception a lot -- Tanguay really doesn't (I think this explains his mediocrity on the PP), what he does extremely well is make very fast decisions and execute them immediately (e.g. exploiting a Dman who's slightly out of position before the guy even realizes it).

Also E: I think you'll find some consensus among the advanced stats types that they like shooting percentage, because on average, the Shoot% of a given player remains fairly constant over time, especially at EV. A season where a player is below his career rate is a Buy opportunity (e.g. Jason Blake right now); where he's above, a Sell opportunity (e.g. Jason Blake a year ago).

Anyway, between Tanguay and Lang (a terrific player), the Habs should be miles better at EV this season.

Jaybird said...

Great to see you back! Your insight luminates the C+ graded blog writers union. Only thing I have to say is "TANGUAY...SHOOT!"

Hopefully I won't be screaming that at the tele too much.

Jaybird said...

I meant to say most blogs are C+....not yours though ;)

E said...

simonus- i get the point, but if you look at the first article you site, it's clearly not the case with tanguay, since he also misses a smaller than average portion of his shots. his total shooting percentage is just as ludicrously high as his standard shooting percentage.

che- to me, the curious thing is that tanguay would be so different in his style of play (regarding deciding when to shoot) than basically every other player of his generation. you could certainly cluster guys by careful shooters and lazy shooters, and you might find a couple of percentage points difference favoring the former, but it's still bizarre that tanguay is so far ahead of the pack.

olibou- oh, certainly. one of the things i wanted to do in this post is differentiate shooting percentage from scoring value. olli jokinen got 34 goals at a 10% s.p., andrew cogliano got 18 while shooting 18.4%. you're best scorers aren't necessarily your most efficient shooters.

matt- [deferential curtsy] i'm looking forward to watching him, just to see if there is something in the way he plays which is so different from the average better-than-average player as to explain this anomaly. in a way, it's sort of hard to believe that it might be something (gasp!) actually perceptible- we're so conditioned to think of talent as mysterious.

and yeah, i do see that stats folk like shooting percentage in certain conditions, i just think the tendency is for them to interpret it as one where the power rests with the average, not in the will of the player. which may be true, but as i said, it makes tanguay either a) the exception that proves the rule, or b) a massive freak.

jay- oh, i think you'll be screaming at the television a lot. sounds like part of the reason for the high s% is selective shooting, which means... yeah, he's a methodical guy. higgins, on the other hand, might do well to try the jokinen method.