Shot quality

Toronto Maple Leafs v Boston Bruins - Game Two

If you’re an assistant coach of a National Hockey League team and you say something that immediately sets off the bullshit detector of hundreds of hockey die-hards and analysts, it may be time to seriously question the direction of your franchise.

Meet Greg Cronin, the man behind such lines as “I think Orr has proven he’s more than just an enforcer”, who gave an interview to Alec Brownscombe of Maple Leafs Hot Stove that made the rounds this week. Everything in it just seemed off, and while Cronin seemed to be aware that there are, er, certain metrics that showed his Toronto Maple Leafs were not as good as their wins and losses record indicated.

Much of his argument focused on what we have begun to call “shot quality”.

Look at some of the comments Cronin made:

“We prided ourselves on quality possessions with the puck”

“We want to encourage our guys to have quality puck time in the offensive zone. If there’s a chance to take the puck to the net with a quality shot, then take the shot.”

“Our even strength goals produced must’ve been pretty high, which is a pretty good reflection that our quality of shots or quality of possession strategy was fairly successful.”

There’s also a bizarre comparison with the Leafs to a mythical team that manages to take three shots from poor areas on the ice, and how the Leafs’ methods were much better. What more, Cronin talks about how Randy Carlyle’s system is superior to the one that an NHL head coach at a panel was presenting that dealt with putting lots of pucks on the net to create rebounds. He also “unequivocally [does] not believe” that the Maple Leafs were out-possessed last season, despite the statistic that correlates so strongly with zone time saying otherwise.

There’s just a lot of wrong things. I happen to know somebody who sat in on the same panel at the draft as Cronin, who told me who the NHL head coach was. It’s not like the coach that Cronin is referring to coaches a squad that can’t put the puck in the net—it was a top-10 offensive team over the last two seasons, and one far from the bottom of the leaderboard in team shooting percentage.

I guess for those of you new to the debate, some analysts like to use a metric called “Corsi” as a proxy for offensive zone time, and that happens to correlate with winning and predict future wins better than a team’s record does. The basic argument against the metric, which tallies up all shot attempts, including blocks and misses, is that Corsi doesn’t take into account the quality of a shot.

The Maple Leafs had the lowest “Corsi %” in the NHL, controlling just 44.1% of all shot attempts, but they did convert on over 10 per cent of their shots, which was the highest rate in the league in 2013. From a predictive perspective, we’d expect the team’s shooting percentage to regress to a league-wide mean over a long stretch of games and the Leafs goals and against rate would more closely reflect their puck-possession rate.

I’ve written that above paragraph so many different ways over two seasons. Obviously, a player like Sidney Crosby or Henrik Sedin will put more pucks in the net per opportunity than a player like Shawn Thornton or George Parros. Michael Parkatti used six seasons of data to prove that. The problem is that there are only so many Crosbys and Sedins going around, and you can’t construct a top six with the very best players in the world in a time where there’s both an entry draft and a salary cap. The talent gets divided among 30 teams.

Every year, after 30 or 40 games or so, there will be one team that is high up in the standings and they do not deserve to be. The 2010 Colorado Avalanche, the 2011 Dallas Stars, and the 2012 Minnesota Wild are all the cousins of the 2013 Toronto Maple Leafs, teams that simply didn’t belong to be in the playoff race. We had those debates, the Minnesota one was legendary. They all boiled down to the same theme. “Why should it matter if we’re getting out-shot every night? We’re winning games, that’s what counts, no?”

Well… winning counts, and to win you need to score goals, and to score goals you need to get scoring chances. It sounds easy enough to suggest that you want to eschew bad shots and take good shots, but that isn’t the case. Sidney Crosby and Henrik Sedin’s teams don’t have high shooting percentages when they’re on the ice because they wait around for the perfect set up, they just plain set up scoring chances more than other players possibly can. With other guys, it’s different and for a guy with a buttload of goals in his career and the reputation of having a deadly wrist shot, Alexander Ovechkin is simply a player that takes a tonne of shots, and sometimes they go in.

Obviously, you’d be an idiot to not prefer a 30-foot shot from the slot to a 50-foot point shot. The problem is that you can’t just wait for those shots to come. Hockey moves quickly and teams need to be pro-active. Defences are too well-structured to beat the same way each time into the zone. They need to be worn down with aggressive play, with speed, with passing, with shots, with… well, a bunch of different things.

Basically, Cronin is saying that he thinks the Leafs can shoot 10.56% again next season. This is very unlikely. The Washington Capitals of 2010 are the only other team since 2007-2008 to shoot above 10%, and the next season, they fell off a cliff, moved towards the trap and fired their coach (UPDATE – as pointed out in the comments, that was the year *after* their shooting fell to 7.5%). They wound up shooting below league average. Coincidentally, the Los Angeles Kings of 2012 have the lowest recorded shooting percentage recorded over an 82-game season. Not only did the 2012 Kings win the Stanley Cup, but their shooting percentage in 2013 was actually higher than the Capitals of 2010-2011.

When you talk shot quality, the inference you’re making is that you set up shots that lead to a higher percentage of pucks going in. But let’s test that out. Hockey Analysis keeps single-season raw data for goals and shots. By copy and pasting a few pages into Excel, I have five consecutive seasons of shooting percentages. All we do is sort the 150 seasons of team shooting rates into five “buckets” and see how those teams performed the next year:

Shot % in Year 1 Shot % in Year 2
Bucket 1 9.0% 8.0%
Bucket 2 8.3% 7.9%
Bucket 3 7.9% 7.9%
Bucket 4 7.3% 7.9%
Bucket 5 6.8% 7.8%

Incidentally… any difference between the top teams and bottom teams is minimal. Teams that shoot poorly one season combine to shoot so close to league average the next season. If shot quality was a repeatable thing that teams could gameplan around, would the higher-rated teams in Bucket 1 not be able to sustain the same rates rather than fall to one-tenth of a percentage above league average rates? Not unless they were deliberately sabotaging their own teams.

You can expect stars with better shots to have higher rates than grinders, but if there was an indication that teams could influence shooting rates, it would show up somewhere in the data. It doesn’t. There’s a lot of randomness when you’re looking at one season of shooting percentage data, but when you combine 30 teams you’re looking at something closer to the truth.

So… none of this is to say that high quality shots don’t help teams win. The question is whether teams that get high quality shots on Wednesday will also be able to get them on Thursday.

Next up: shot quantity.