Jets’ TOI Quality of Competition



Eric T. has been performing yeoman work this summer over at NHL Numbers, and he added another excellent article to his collection last week with a look at Quality of Competition. What makes this look at the subject a bit different is that the data used in measuring that quality was time on ice of opposing forwards and defencemen, and not out-shooting.

Over the last few years, the hockey mathosphere has largely adopted the quality of competition metric developed by Gabe Desjardins at Behind the Net as a means of quickly assessing which players were facing the most difficult opposition. For those of you with longer memories, you’ll recall Gabe added Corsi as a component, after a suggestion by Vic Ferrari, in an attempt to improve on his original formula. Corsi Relative Quality of Competition has become the current standard for those trying to best judge the degree of difficulty faced by players.

That ranking has always come with the standard caveat that the QoC stats should be used as an ordinal for players on the same team, and the raw numbers shouldn’t be used to cross between teams. In other words, although Olli Jokinen’s 2.056 last season was a higher raw number than Jim Slater’s .900 and likely suggests he had a tougher go of things than his new teammate, the one solid judgement worth making from those numbers was that both players routinely faced top lines, since Jokinen’s figure was second behind Curtis Glencross on Calgary while Slater’s figure topped all Jets’ forwards.

By the Corsi Relative QoC measure, the GST line faced the toughest competition of all Jet forwards, followed by Ladd-Little-Wheeler, followed by the others all the way down to Tim Stapleton. That matched the eye test for anyone that watched the games as well, and Claude Noel openly admitted that he was using GST as his hard minutes group, particularly at the MTS Centre where he controlled last change. 

What Eric’s work hints at, however, is that the current measure might be missing a component, because it seemed to heavily weight to which forward groups a player faced with somewht less consideration to the quality of defenceman on the other side of the ice. This question often arises when teams like Winnipeg go about things in the manner they did last year. The GST guys might have routinely faced the best forwards, but common sense suggests it would have been unlikely that opposition coaches would have targeted them with their primary shutdown defence pair, since GST posed virtually no offensive threat.

So, with that in mind, what do we see when competition based on opponent ice time is measured? Here’s a lovely data plot from that post, and note that opposing forward and defence TOI measures the average amount of time per game those players were on ice for the entire year, and not just when they played against the Jets:



That pretty much squares with my statement above. The X axis shows us that the GST line faced the opposition forwards that averaged the most minutes over the course of the season, with Ladd’s line close behind, followed by a clear gap back to the next 5 guys, then an ocean before you reach Stapleton. That rating also squares almost perfectly with the Corsi Relative QoC figures for the team last season. 

The data on the Y axis, though, almost certainly speaks as to which Jet forwards were perceived as threats by the other teams. Ladd’s line faced the other team’s top defenders based on TOI, followed by the other decent skill forwards, with the GST dudes, Stapleton and Miettinen in the bottom tier. Again, not wholly surprising, because matching your D based on the other team’s forward skill is a routine most coaches have followed since roughly the dawn of time.

As a result, if one were blend the data, that would suggest that Ladd’s line faced the toughest overall opposition, by dint of meeting top-six forwards on the regular along with the other team’s top pairings on D, and that the likes of Kane and Burmistrov, while lagging in terms of forwards faced, get a boost in degree of difficulty from facing most teams’ second D pairings, rather than the parade of 3rd pairs that GST, Stapleton and Miettinen worked against. 

I’m not entirely sure that I buy that formulation in total, though. I don’t for a second diminsh what a really good top defence pairing can accomplish, but I’ve long held the belief that forwards drive the bus in terms of out-shooting, and that with few exceptions, what makes a team difficult to play against is their quality up front. 

With that in mind, I thought I’d run some numbers for my own amusement. I selected, based on EV ice time, the top 120 D (30 x 4), and top 270 forwards (30 x 9) in the NHL last season. That sort of sweep seems likely to cover all of the worthwhile contributors at even strength.

Next, since very few top 4 defenders get the sort of super-cushy Zone Starts that guys like the Sedins or the top Chicago forwards receive, I controlled the Zone Start percentage for forwards to be no higher than Eric Johnson’s 59.2. That reduced my population to 246 forwards.

What I looked for next was Relative Corsi. If defensemen are as valuable as forwards in equal measure, you’d expect the number of realy high end out-shoooters for both positions to line up at approximately the same rate. Since I now have populations of 120 D and 246 F, you’d think that high end out-shooters should favour the forwards by about 2 to 1, if everything is equal in terms of ability to drive play. 

My line in the sand for Relative Corsi was +10, which means a player would, over the course of 60 minutes of EV icetime, have been on the ice for 10 extra Corsi events in his team’s favour, relative to his team’s overall performance at out-shooting. That was a somewhat arbitrary number I picked for a standard, but +10 seemed like a pretty good demarcation for top out-shooters. Your mileage, as always, may vary.

8 defencemen reached that criteria, with 2 of them, Byfuglien and Enstrom, playing on the same pair. Brian Campbell just missed out at 9.9, and he played most of last year with Jason Garrison, who was one of the 8. The others were Weber, Gologoski, Karlsson, Streit, and Chara. That’s a hell of a group, but it’s pretty small in number, and I don’t think anyone would consider that collection the last word in defencemen by any means. 

Up front, 44 forwards, or 5 1/2 times as many, made the cut. Even accounting for quality of competition differences that some top nine forwards might face when they play on teams that hang certain forward groups out to dry for the benefit of others, that’s a significant differential. I’m not making the case that this is the last word or anything of that nature, but my sense from the general data collected over the last few years is that there really aren’t that many defenders that can drive play. Most of them, even at the highest levels, just try to survive their ice time against top end forwards without getting torched, and because of that, I’m loathe to assign equal weight to defence pairing quality. It’s important, but not at the level of forward quality, imo. 

The other item of interest in this entire discussion of competition is something Eric wasn’t trying to measure, but still seems worth mentioning when we discuss degree of difficulty. Where players start their shifts has a meaningful effect on how hard their circumstances end up being. In Winnipeg’s case, The GST line, beyond facing top forwards, did so while starting in their own end a fair bit, which is almost certainly a tougher task than average. In fact, I’ve always operated under the premise that it’s because I’m an avid, if thoroughly mediocre, golfer, that the idea of positioning makes intuitive sense to me.




Anyone familiar with pasture pool knows that courses have multiple ratings, which are based on which set of tees a player uses. As an example, if you hit the following link and scroll down to the Scorecard header, you’ll find the various ratings for the Stadium Course at TPC Scottsdale. For a male scratch golfer, the expected difference between a round played from the whites and one played from the tips is 5.9 strokes, ranging from 68.7 and 74.6, or from 2 under par to nearly 4 over, just by changing where you begin play. Same golf course, but it’s a totally different experience, since longer distances on holes will leave players with longer, less accurate clubs being used for approach shots, and less accurate approach shots will leave longer putts or greens being missed outright. A higher score will be, on average, inevitable.

That sort of handicapping is at the heart of why I tend to regard what the Sedins face in terms of competition to be less stressful than what a player like Anze Kopitar faces. If you go back to the source article, those three gents are in the top right of their team plots, facing the toughest combo of forwards and defencemen, and they all had superb out-shooting numbers. The Sedins, though, did that while having a OZone start percentage in the 79% range, compared to about 51% for Kopitar. The Sedins are great, but playing from the forward tees makes any golf course easier, and there’s significant evidence that a heavy ratio in favour of OZone starts is less of a strain on the ice as well.

In the case of the Jets, although the Ladd line certainly didn’t have things easy, I’d still base virtually any analysis about last year’s team on the premise that the GST line faced the toughest ice time, since they mixed top forward comp with very difficult Zone Start percentages, followed by Ladd’s line, with the Burmistrov-Kane-Wellwood line getting the easiest pickings of the main forwards. That would be right in line with the Corsi Relative QoC numbers, but if there was a complete scale encompassing this data and the BTN stuff, I suspect that the spread between the top nine forwards would likely be very tight.

All that noted, neither of my two areas of questioning render Eric’s work any less valuable, since they largely fall outside the bounds of his current exercise, and for teams with with a more balanced forward utilization than the Jets exhibited last year, the quality of defencemen faced might well be the tipping point in determining who had the tougher lot in life. I don’t doubt that we’ll see home-road splits for this data at some juncture, as well as a similar set of data collected and plotted fordefenceman, and that’s all to the good.

From my POV, what’s really valuable about this work is that it’s one more bit of context to add to the pile, and that it approaches quality of competition from an angle that forces us to consider if the current consensus is wholly merited. I might not weight defence QComp as quite as highly as forward QComp, but I’m not oblivious to the fact that the contributions that a genuinely top defenceman makes can have a significant effect on QComp for his opposition. When we consider Zone Starts and Corsi Relative QoC alongside what Eric’s collected, it seems to me that we’re another step closer to fully answering the question about which players really drive play.

  • Tim in Kelowna

    Really interesting stuff.

    I agree with you re: difficulty of some players. Good players become ‘great’ players in part because of the quality of the rest of their team. It’s a trickle down effect where someone like the Sedins have Malhotra and Kesler to give them the high ground, while Iginla has… Stajan.

    The defenseman issue is an interesting one too. I’d like to see a graph of how the defensemen line up against opposition fwd TOI. It’s pretty obvious to me (and the graphs seem to confirm) that coaches see shut-down dmen as just as,if not more, important in preventing chances/goals/corsi events. All the teams stars’ see the big minute dmen as often as possible.