The current 12-game point streak for the Columbus Blue Jackets is cause for some degree of jubilation. This is, after all, a franchise that has been on hard times for years and is navigating its first season after finishing in last place in the NHL. The perceived turnaround of the team makes for a good story, something the local and national hockey fan alike can get behind. A bunch of plucky players are fighting their way uphill against both the negative forces of the previous campaign and the early-season woes that afflicted the squad. What’s not to like?
Unfortunately, it’s a troubling streak for me as a fan of the team. I realize sitting back and enjoying the ride is more rewarding, but troubling signs cloud the future of the club, suggesting that the end of the streak is well overdue… and that the streak is actually more damaging than worthwhile for next year and the year after.
Why do I lack confidence about the Blue Jackets going forward? It’s in the very streak that so many are praising. The way the Blue Jackets are performing is not indicative of a hockey team that can repeatably find victory, at least not without the kind of goaltending they’re currently getting.
Laying the Background. If you follow me on Twitter, you’ve seen me rant, rave, and harp on hockey statistics at varying intervals. A very important method in understanding the current performance and future success of a team is through evaluation of its even strength shot and shot attempt differentials. Fenwick is a measure named after the person who first described the statistic, and it’s merely the sum of goals, shots and shot attempts (that is, missed shots) for a team, minus the sum of goals, shots and shot attempts against a team.
The idea of using shot differential to predict success isn’t overly complicated, but it’s valuable to consider. Michael Parkatti of Boys on the Bus made a quick model of points as a function of shot differential, and there’s a strong correlation (unlike, say, for hits). Whether you choose to believe its utility or not is inconsequential; it is statistically significantly correlated to points.
In fact, Fenwick is also a predictive tool for future success in the NHL. As has been presented by SnarkSD at Fear the Fin, tie-game Fenwick provides a strong method for predicting future success. It’s limited, like any metric, but Fenwick events are more frequent then goals so it’s provides more information over a season. Eric T of Broad Street Hockey and NHL Numbers further explains this element, noting that players are only on ice for around 50 goals for or against per year. Increasing the sample removes (at that sample size, statistically significant) elements of puck luck from our observations. The tie game stipulation is used to remove score effects (that is, if a team is leading, they often act more defensively thus limiting their shot generation ability).
We can improve our sample by including non-tied situations through the use of Score-Adjusted Fenwick. It’s nothing too wild, and this article by Eric T of Broad Street Hockey discusses its development and usage. Travis Yost maintains an update of Score-Adjusted Fenwick, and you can check out the latest one here.
Like most things in hockey, there is a degree of luck involved. Shooting percentage is seen to be random (or luck driven). We can also consider the normal distribution in accounting for Phil Kessel’s early season cold streak. Just because it’s not naturally intuitive doesn’t mean something is outside the realm of probabilistic occurrence. Just watching a puck in play shows situation-based luck, but even a particular percentage is all about luck: 90% goaltending does’t mean always stopping 9 out of 10 shots. It means some games are shutouts, some are blowouts, and some are between. They’re all still events from a .900 goalie, but there is variance.
Fenwick can “fail” when exploring lucky situations: teams with abnormally high shooting percentages or abnormally high save percentages. That doesn’t mean Fenwick-based observations are wrong, just they’re simply being impacted by luck that will regress to a norm over long periods of time.
We can see the impact of variance or luck when looking at PDO. A primer on PDO was written by Cam Charron at Backhand Shelf, and a more rigorous article was done by Patrick D at NHL Numbers. PDO is an example of luck being minimized as a season progresses, a regression toward the mean value of 1000. But on a brief run, luck can run the show. Suddenly hot shooters can overcome the kind of shot differentials that accompany long-term losing, goalies can be greater than Hasek in short bursts.
These kind of luck-driven events shouldn’t be used to infer seasonal success. However, a shortened season means that luck won’t fully dissipate from the final standings. Patrick D explored the implications of this at Fear The Fin. A short streak in a 48 game season might be enough to make or break a team that is otherwise obvious in its 82-game standing.
So What Can We Learn About the Columbus Surge? Unfortunately, Columbus are on the wrong side of shot differential, close-game Fenwick, and score-adjusted Fenwick. Let’s start by using the Boys on the Bus model for points as a function of shot differential (just to give a rough idea of expected performance). For the season, the Jackets are -3.5 shot differential per game. Over an 82 game season, that means they’d have about 84 points a range that’s around 21st to 23rd in the NHL last season. Of the 12 game point streak, they’re averaging -5.2 shots differential per game. That would give them about 80 points, somewhere between 27th and 24th last year.
You can find current team Fenwick percentage values at Behind the Net. Right now the Blue Jackets are third-to-last on the season with a 44.76%. That is, the Jackets only generate 44.76% of shots attempts in their games; their opponents get the majority of Fenwick events. The latest publication from Travis Yost sees Columbus also third-to-last in score-adjusted Fenwick with 45.19% of events going to the Jackets.
Over the streak, this hasn’t changed much. Prior to their point streak, the Jackets outshot or equaled their opponents in only 7 of 19 games (4 times with a positive Fenwick). In the 12 game streak, this has happened 4 times (all four positive). It’s a slight uptick in number of positive games, but continues to see their shot separation below zero. It’s not as bad now, but it’s still bad. When a team is so low in close Fenwick and score-adjusted Fenwick, they should be expected to finish near the bottom of the league.
So Why Worry About This Anyway? If the Columbus Blue Jackets are being so routinely outshot, out-shot-attempted, they’re not structured to be successful in the NHL. Shot differential is statistically linked to standing points, and for future success the Jackets need to find a way to overcome this weakness, to drive play in their games. This is particularly concerning in the middle of a lengthy point streak, as there is little being done by the skaters to mitigate the statistical disparity.
Shot differential on a player-by-player basis isn’t inspiring for much of the team, goal totals during the streak aren’t keeping the club afloat (only 2.5 goals per game over the last 12, which would be good for somewhere between 20th and 23rd in this season’s NHL). Without skater defense or scoring to keep this alive, a fall seems inevitable and support, necessary.
For a team so likely to finish near the bottom of the NHL, a strong way to improve is through high draft picks (you may have heard some hype about a defenseman from Portland or two forwards from Halifax?). By “playing” their way out of the basement, Columbus remains a very poor group of skaters but is now unlikely to have a chance at drafting one of the three top players. It’s hard for anybody to say that winning is bad… but the kind of outmatched victories the Jackets are seeing this season are not good news for the future of the team.
Is There Anything Positive to Take Away? The answer is “yes” but like the streak itself, it’s more complicated than a knee-jerk response. The positive is that Sergei Bobrovsky is both responsible for this streak, and proving himself to be a starting-caliber goaltender. We have already soundly eliminated team offense and defense as reasons for Blue Jacket points, so the remaining answer lies in net.
And the numbers stick out immediately. Bob’s overall save percentage is now a stunning .932 (second only to the injured Craig Anderson), and his even strength mark is an insane .936. During the last 12 games, Bobrovsky has been the starter for 11. Those games saw Bob put up an unbelievable .949 overall. Despite the narrative of team hard work or improved coaching style, the blemishes are being covered by the netminder.
In fact, this is a performance that is among the greatest of all time. Only 17 goalies in NHL history have played 20 games and had a save percentage above .930 (Bob is number 18 if he continues his pace). This suggests that Bob won’t keep up his inspiring pace, but his work over the long-term (now taken over three seasons) suggests that Bob is the kind of starting goalie any team would want. His success is good to see, but it’s a shame that could come at the expense of a draft pick.
The Future is More Confusing. Without this point streak, the remainder of the season would have been easier to predict: shed trade parts, allow the new draft-oriented GM to prepare for the June 30th event. Now the Blue Jackets sit in an unenviable position, perched precariously between the abyss and Sergei Bobrovsky’s play, all the while managing to move further away from a top prospect boost.
There are two things that might mitigate the problems brought by this streak. The first is the previously mentioned netminder: Bob’s play alone might be able to elevate Columbus’ profile, making the team more attractive to free agents. If they want to play for a winner, they’ll have it… even this season’s winning was only because of one man.
The second saving grace is the new front office members of the Blue Jackets. If Jarmo Kekalainen and John Davidson are successful at talent evaluation and salary cap management, the rebuild will continue unabated. While a top pick would make their jobs easier, a good hockey ops staff won’t be derailed by this kind of setback. It is in their hands that the success of the team currently sits.
So today we will wait to see when the streak ends and remain solemnly confident that the current Blue Jackets skater core is not built to win, no matter how hard they work. Our hope for victory right now is only in goal. That’s not enough unless you’re counting on unwavering historic save percentages every game. But our hope for the future is in the GM’s office and in the NHL draft. It remains to be seen if the latter will be game-changing.