clock menu more-arrow no yes

Filed under:

Fenwick Close and Predicting Success in the NHL

New, comments

A look at the Washington Capitals' relationship with success-predicting statistics as compared to the rest of the League.

Geoff Burke-USA TODAY Sports

No doubt if you're a reader of this blog, or of similar websites, you've heard the phrase "Fenwick Close is a predictor of success" or something like it. But exactly how good of a predictor is Fenwick Close%? Elite NHL teams will almost invariably outpossess their opponents, but is "teams that perform well at even-strength in close games will be successful" really a revelation?

That probably depends on just how strong that connection between close-game possession and success (standing points) is. Below, that relationship is visualized for each of the past 4 NHL seasons (2010-2011 thru 2013-2014).

The  "Ranking Difference" refers to the difference between a team's end of season ranking in the NHL standings by point totals, and that team's league ranking in Fenwick Close %.

Ranking Difference

# of NHL Teams

2011

Between 0 & 3

10

Between 4 & 9

13

Between 10 & 14

4

15 or Greater

3

So what we saw in 2011 was that 1/3 of all NHL teams saw their place in the league standings end up three or fewer spots away from where they were in Fenwick-Close rankings. Let's call this a very strong correlation. 13 NHL teams saw the relationship between four and nine ranking spots, which we'll call a moderate correlation, and seven NHL teams had a difference of 10 or greater between their standings rankings and their FFClose% ranking - we'll consider that to not be a positive correlation at all.

Here's how those numbers held up in the three subsequent seasons:

Ranking Difference

# of NHL Teams

2012

Between 0 & 3

13

Between 4 & 9

13

Between 10 & 14

1

15 or Greater

3

Ranking Difference

# of NHL Teams

2013

Between 0 & 3

10

Between 4 & 9

13

Between 10 & 14

4

15 or Greater

3

Ranking Difference

# of NHL Teams

2014

Between 0 & 3

10

Between 4 & 9

13

Between 10 & 14

4

15 or Greater

3

Alright, let's take a peek at how all of this data looks when aggregated:

Ranking Difference

# of NHL Teams

Total 2011-2014

Between 0 & 3

43

Between 4 & 9

52

Between 10 & 14

13

15 or Greater

12

So over the span of four seasons - 4410 hockey games, accounting for the lockout shortened season (and more than a sufficient sample size) - more than a third of completed team-seasons finished with a very strong correlation between Fenwick Close % and their place in the standings.

A whopping 79% of team-seasons have ended with the FFClose ranking within nine spots of the final standings ranking.

Approximately 1/5 of team-seasons ended with FFClose% having no evident relationship with success, failure, or mediocrity.

It shouldn't come as a surprise to anyone who has followed this team as advanced statistics have grown more prevalent and relevant over the years, but in the past, the Capitals have under-performed with regards to many of the widely acknowledged advanced indicators of a good hockey team. However, that hasn't stopped them from winning games, and at the end of the day isn't that what sports are about?

Indeed, if you look at each team's relationship between FF% and Points over the last four years, the Capitals have the third weakest relationship.

Teams all the way to the left are those with the closest correlation of FFClose% to standings points over the last 4 seasons. Only Anaheim and New Jersey have had a weaker relationship between Fenwick Close and regular-season success than the Washington Capitals.

But the above visualization only tells part of the story. The simple lack of a relationship does not explain whether the team has been more or less successful than their Fenwick Close would imply. By allowing negative values instead of absolutes, we can illuminate this variation.

See here that only Anaheim has performed better in the standings versus their Fenwick ranking than Washington. The only other team whose standing points defied their Fenwick performance to a greater degree than the Caps - New Jersey - saw less favorable results, which is to say they were rewarded with standing points at a much lower rate, despite better Fenwick play.

At a glance, there isn't much to takeaway from the above. At least there aren't any obvious trends. The majority of teams underperform in the standings, as compared to their Fenwick, which of course does not take into account special teams (more on this in a moment). Teams that have been pretty great the last four seasons - the Chicagos, Los Angeleses, San Joses, and Bostons of the world - are found all over that graph. Those who have struggled mightily over that same period - Buffalo, Florida, Edmonton, Calgary - are also loathe to reveal anything telltale.

But teams change. Turnover on players, coaches, and management all contribute to shifts in performance year over year, and perhaps these variances are great enough to muddle the data, hiding insightful trends.

What changes less is the standard of performance needed to reach the Stanley Cup playoffs each season, which raises the question of whether or not there tends to be a stronger relationship between FFClose% and standing points for playoff teams versus non-playoff teams.

The below graph shows this relationship for each of the 30 positions in the standings over the past 4 years.

You can see above that the relationship between FFClose% and success seems to be a bit more temperamental amongst teams in playoff slots, particularly in rankings one through ten, before leveling out for the most part the rest of the way.

(Note that these rankings are not an exact representation of playoff vs non-playoff teams, because rankings 1-16 do not account for division winners, or for conference overlap.)

FFClose vs Standings Ranking Diff

Standing Position

Average Difference

1 thru 15

6.33

16 thru 30

6.1

So there's nothing to indicate that playoff-caliber teams have a different relationship with FFClose % than weaker teams. Both calibers of team tend to have a discrepancy of about 6 spots in rankings between the NHL Standings and the Fenwick Close standings. The intuitive explanation for the lack of a relationship between Fenwick and standings points would be special teams play, since Fenwick doesn't account for them.

The other, altogether more nebulous culprit, would be luck.

This below graph shows each team's ST% (PP% + PK%) as an average over the past four years. The teams whose data is shown in red, are those 10 teams with the weakest relationship between Fenwick and Standings Points over that time.

You'll notice that three of the six teams with the best special teams over the past four years had a weaker Fenwick/Standing Points relationship, while the two teams with the worst special teams in the league also had weaker relationships.

So what does this mean for our Washington Capitals?

Well, they're presently 4th in the league in FF% in close situations. By percentages, they've got the 2nd best powerplay in the NHL, and a middle of the pack PK unit.

If the Capitals sustain their level of play on both special teams units, and consistency in possession, and maybe factor in a bit of luck going their way (or at least the laws of probability behaving accordingly) the Capitals could be good this year. Very good.