by PuckStopsHere on 11/24/09 at 02:11 PM ET
It is a relatively widely held belief among proponents of hockey sabermetrics that goals scored and allowed better predict future success than the win/loss record of a given team at a point in time. This can be explained to mean that luck tends to even out over the longterm and some teams might be lucky enough to win a few close games, but in the future they will lose some close games as well (and vice versa). The teams that are managing to significantly outscore opponents in their wins are likely better teams than those who barely beat opponents. This may not be true in all cases. For example, a team may be very good at shutting down their opponent when they get a lead (and hence win close games) but be completely inept at coming from behind and often allow goals when trying to force thing offensively (and hence lose blowouts).
A generally believed extension of that is that, in general, shots are a good predictor of goals and a good team can be predicted from its shot differential. The problem here is that goaltending is removed from the equation. Teams with outstanding goaltending will win more often than their shot differential predicts and teams with poor goaltending will underperform their shot differential.
There is a significant test to this theory in the NHL so far this season. The Toronto Maple Leafs are tied for last place with the Carolina Hurricanes (both have 15 points so far). Both have very bad goal differentials. Carolina allows 1.31 more goals per game than they score and Toronto allows 1.10 more goals per game than they score. It is very rare for a team to be this far in the negative regions. For example, last year’s worst was the New York Islanders with 0.92 goals allowed per game more than they scored. The good news to this is that unless you believe that Carolina or Toronto are significantly worse than last year’s Islanders (and I don’t), things should get better in both cities. They are both doing unsustainably poorly.
When we look at shots per game, we get a very different story in Toronto. After last night’s 60 shot outing against the New York islanders, the Leafs have moved into the league lead in shots per game with 34.7. The last place team takes the most shots per game. That is a nearly impossible feat. Toronto has a positive shot differential. They take 4.2 shots more than opposition each game. Detroit, Pittsburgh and Chicago are the only teams that rank ahead of this in the league (and Buffalo is tied with them). The Toronto Maple Leafs shot differential suggests the Leafs should be a pretty good team. How can that be?
The biggest problem when looking at shot differentials is that it removes goaltending from the picture. Toronto has had poor goaltending this season. Jonas Gustavsson leads the team with a .900 saves percentage and a 3.18 GAA. Those are not the numbers of a top starter. They are more acceptable for a seldom used backup. The other goalies in the system have been worse. Vesa Toskala has a .865 saves percentage and a 3.91 GAA (those numbers are improving from a horrid start to the season). Joey MacDonald has a .875 saves percentage and 3.91 GAA. Neither Toskala nor MacDonald have put up numbers that are good enough to keep them in an NHL job, so far this season. Thus Toronto will be a worse team than their shot differential shows when taking into account goaltending.
Shot differentials also assume that all shots are equal (i.e. over the longterm all teams will have roughly the same shooting percentage). If this is true, Toronto has been very unlucky in terms of their shooting. The average NHL team has a shooting percentage around 8.5-9%. Toronto has a 7.3% shooting percentage so far this season. It has been shown that teams who play with the lead generally have higher shooting percentages because they tend not to take the lower quality shots that a team pressing to catch up in a game might take. This works against the Leafs, as they are usually trailing in games this season.
Despite these caveats, the Toronto Maple Leaf shot differential suggests they are better than a last place team. It certainly isn’t a fair conclusion that they are fourth best in the league (despite a fourth best shot differential) because they suffer from poor goaltending and tend to be behind in their games and both will tend to weaken the correlation between shot differential and winning, but their shot differential is too good for a last place team. The only logical conclusion is that Toronto is not as bad as they have appeared so far this year. They are unlikely to finish last. The playoffs are too much to expect in the positive direction. I expect, given their poor start, the Leafs may wind up a lottery team, but not last place. I also expect that Carolina is better than their record suggests and will improve when Eric Staal and Cam Ward return from injury. I think that a season’s end, there will likely be another team at the bottom of the NHL standings. Minnesota, Florida and the New York Islanders look like possible candidates (despite being ahead of Toronto and Carolina at this point).
Toronto is an interesting study this year. It is very unlikely that the team that takes the most shots in the league could also be the last place team. Toronto’s shots on goal give them a good shot differential. This is one that ranks them among the best teams in the NHL. While it is true that this removes effects of goaltending and of shot selection from study, the Leafs shots differential suggests the team has been unlucky so far this season to be doing as poorly as they are in the standings. Hence if Toronto were to fire Ron Wilson and see an improvement under their next coach that would not prove the coach is a good coach. That improvement is predictable regardless of coaching.
Be the first to comment.
Add a Comment
Please limit embedded image or media size to 575 pixels wide.
Most Recent Blog Posts
About The Puck Stops Here
Who am I? A diehard hockey fan.
Why am I blogging? I want to.
Why are you reading it? ???