Advanced Hockey Stats – 101

More Damn Lies, err Stats

As I have mentioned in my previous post, I am a big fan of the advanced hockey stats being done at behindthenet by Gabriel Desjardins.  In this article I am going to explain some of the techniques I use to interpret those stats and how I use them to evaluate hockey in general and the Chicago Blackhawks in particular.

I am going to talk about 4 stats in this article; Quality of Competition, Quality of Teammates, Zone starts and Zone finishes.  We will talk about  ‘+-/60 minutes of ice time on”, and ‘+-/60 minutes of ice time off”  and maybe a couple of others in a future article.





Quality of Competition

The first stat I generally use when looking at a player is Quality of Competition.  Using this stat you can get an idea of who a particular player typically plays against.  The higher the number is positive, the more this player only competes against top end talent.  The larger the number is negative, the more this player only plays against 4th line talent.


This is important when looking at the Chicago Blackhawks because Quenneville is a matchup coach.  Q will match certain players against the top opposing players and others against the bottom of the other teams roster.  Q does this so much it is almost, as if, players like Seabrook and Campbell were playing against different teams.

You can expect your top defensive players when matching up against other teams top lines to have as high as a .08 something QoC.  And players that coaches play exclusively against top level opponents can go into the .120′s.  And you can expect your 4th line players playing against mostly 4th line talent to have somewhere in that -.08 to -.121 range. Though some players go much lower.

Q uses Dave Bolland so much against top end talent that his QoC number is .121 which is the highest rated QoC for a forward in the league.  And only one other player is in that .120 range, Detroit’s Nicklas Lidstrom, who has the highest rating at .128. So you can get an idea of how special Q feels Dave Bolland is defensively.

And at the other spectrum, Ryan Johnson‘s QoC was a -.121 while John Scott’s rating ended up at -.181.  However, for awhile there Scott had the lowest QoC rating for anyone in the league.

Using QoC while looking at centers can give you a good idea of what their responsibilities were:

From the Quality of Competition numbers you will see that Bolland played against the highest level of competition which is something that is pretty well understood.  What is probably less understood is the Toews and Sharp numbers. Typically you would expect number one lines to be matched against a checking line and second lines playing second lines. So you would think Sharp would have the higher level of Competition.  That’s not the case. Some teams would match their top line against the Hawks top line when that team didn’t feel their checking line was up to the task.  So by the end of the year, top lines generally face stiffer competition.

And I know that the local beat writers had Johnson for a stretch as the Hawks third line center but that just was never the case.  Johnson played almost exclusively against 4th line talent.  And you can see it in the QoC numbers.  And if you were wondering why Johnson doesn’t appear to be in the Hawks future plans, well that number might be why.


Quality of Teammates

You can use QoT in a similar way.  Personally, I don’t use QoT as much as QoC.  I feel it is a much smaller sample size so you need to be careful when using it.  So I use it more as a check for other stats.  Is this guys numbers good because of who his line-mates are?

In the table you can see that Bolland was playing with a lower “quality of teammate” then Sharp.  This make sense since Bolland was playing with Bickell and Pisani type players while Sharp was playing with Hossa, Toews and Kane.  That is one of the reasons that Bolland had such a special 2010/11 season.  At least when considering his positive Plus-Minus rating while playing with weaker linemates and against the best competition in the league.

It is also why Sharp’s 5on5 numbers were disappointing, especially at center.  And it is why guys here at BlackhawkUp have been looking this offseason for the Hawks to acquire a better defensive second line center.

Yes, Bowman defended Sharp as a center, that doesn’t mean he wants to play him there.  And it is why I believe Bowman when he says that Kruger is going to get the first crack at that position.  The Hawks drafted Kruger because of his two way hockey skills.

Zone Starts

Zone Starts are another useful tool to determine how a coach views a particular player. If a coach is not manipulating lines, most players ratio of offensive zone draws to defensive zone draws would not change much.  They would all be around the team average which for the Hawks is around 54%.  Yeah, the Hawks are a pretty good team.

Now if coaches have a couple of well rested lines, he might decide to change lines depending on whether there is a face-off in the offensive or defensive zone.  A team might use their checking line in defensive zone draws and use their top scoring line in offensive zone draws.  This manipulation can give offensive players an offensive zone to defensive zone ratio in the 60% range.  While defensive players can have offensive zone draw ratios in the 30-40% range.

Typically Centers will have the lowest offensive zone draw ratios because a coach will sometimes send them out as a second center in defensive situations.  This is just in case a center is thrown out of a face-off.  So they might be slightly lower then their normal line-mates.  Ryan Johnson is an example of this.  He would be on the ice for a defensive zone draw and move off if the Hawks obtain the puck.

Looking at these numbers you see that Bickell and Pisani were the primary checkers for the Chicago Blackhawks. At the other side of the spectrum are the players that are primarily offensive players.  It is no surprise that Kane is at the top of that list.

What is a little bit of a surprise is to see Sharp next in line to Kane.  Hossa is over 12 percentage points lower.  Sharp is a player that coach Q used more often in offensive zone draws this year compared to last.  And it is why his reputation as a defensive forward is somewhat suspect.  Coaches will “protect” players who are a defensive liability by putting them in more offensive situations then defensive ones.  As the season progressed, Sharp appeared to be more and more in that category.

Obvious players in that category are John Scott and Victor Stalberg.  These players started around 60% of their non-neutral faceoffs in the offensive zone.  Since they aren’t offensive threats you can assume coach Q was doing this to protect them.

Looking at this year compared to the previous, this year’s 4th line wasn’t as good in either QoC or Zone Starts.  That generally means that Q was protecting it and the reverse means the Hawks couldn’t attack the other teams 4th line as much as they did the year before.  Q had to leave the opponents 4th line for his 4th line just to get his 4th line on the ice.  That also meant getting the 4th line on the ice was harder in general this last year then the cup one.


Zone Finishes

I don’t use zone finishes as much as zone starts.  The stat is based on where play was stopped.   Thing is play doesn’t always stop at the end of a shift so this is an inference of overall play based on the times it does stop.  Basically, it is supposed to even out and I think it “mostly” does.

Now generally what you will see is players that start in the offensive zone will more often finish in the defensive zone.  And this is true in reverse for defensive zone starts finishing in the offensive zone more often.  And that is because hockey tends to move to the middle.

So I mostly use Zone Finishes in comparing players and in looking for anomalies.

When looking at Hossa and Hendry you can see that when Hossa is on the ice the Hawks are much more effective in keeping play in the offensive zone.  Not really a surprise but you can look at players you don’t really know that well and look for comparisons like that.

Finally one of the interesting uses of Zone Finishes is to look for anomalies.  Basically something unexpected.  Certain players seem to be able to move the puck in the right direction.  When looking at Olesz you can see he starts around 50% of his shifts in either zone but his shifts end more often in the offensive one.  That is kind of a big number for a checking line player.  The ability you see there, basically winning puck possession, is possibly why the Hawks are taking a flyer on Olesz and keeping him on the roster.

So stats can by useful.  They can tell you things you might not be seeing during the game.  They might not tell you everything but they can at least tell you where to look.  And it’s the looking that makes them fun.

Tags: Behind The Net Chicago Blackhawks QoC QoT Quality Of Competition Quality Of Teammates Stats Zone Finishes Zone Starts

  • cliffkoroll

    Simply fascinating Neal. This is a great window into the mind of the Stache.

    One thought off the top of my head. I’m guessing these numbers are not simply based on 5-on-5, which would explain a lot of the Sharp/Hossa discepancy in zone starts last year (Sharp on #1 PP, Hossa on #1 PK). Does that sound right?

  • DaleHalas

    @cliffkoroll Sorry, Cliff. Should have specified that these are all 5on5 numbers. I mostly use 5on5 numbers when determining play. So yea the discrepancy is worth pointing out. Sharp jumped 5-7% in offensive zone starts – this year compared to last. He became a much more one way player this year.

    Brouwer’s numbers are also useful in looking at the difference between the fans view of the player and the coaches.

    And I totally agree with your point on the coach. I try to use stats to see what the coach is doing. Then you can try to figure out why…

  • cliffkoroll

    Re: Kruger. I thoght he showed good awareness/positioning in his brief stint up last year. While the Hawks aren’t hanging their hat on Kruger as 2C by any means, I’m intrigued to see how this works. Seems like a lower-pressure situation than it sounds, considering the QoC and QoT stats of the Hawks 2nd line (look ma, advanced stats!)

  • DaleHalas

    @cliffkoroll Congratulations and see it’s easy. And it is those kinds of observations that make it interesting.

    I don’t think Q is going to ask Kruger to be a two way player against other teams top lines. I think they are looking for a defensively responsible center in mostly offensive situations. And I agree that is a little bit easier position for him to fill. At least possible for him to fill…

  • ChicagoNativeSon

    @cliffkoroll I think I see what you’re getting at cliff/Brian. How can Sharp’s Zone Starts be almost identical to Kane’s and so distant from Hossa’s if he split a lot of 5v5 time between the 1st and 2nd lines? Also, how can Hossa’s be so similar to the percentages of the checking line players?

    So, cronologically, here’s what I think happened…

    Sharp and Hossa start the season together and have very similar Zone Starts.

    Hossa is out for a couple of weeks and then for a month. During these periods Sharp’s line is a defensive liability, and therefore he gets an extremely high % of offensive zone starts. So a big gap is initially created.

    As Hossa and Sharp spend time on the same line again, the delta starts to slowly reduce due to basic math.

    Later in the season, Sharp starts skating almost exclusively with the 1st line, so the chasm slowly widens again.

    Next, Bolland goes down for the final month of the season and the 3rd line isn’t what it used to be. Q starts using a 2nd line including Hossa and Frolik in more defensive situations, and therefore Hossa’s zone start % falls further away from Sharpy’s.

    Make sense?

    Of course, I could be completely wrong here.

  • ChicagoNativeSon

    Great article Neal. I wish I could permanently pin this to the front page of not only this site, but to countless others.

    I completely agree about QoT and Zone Finishes. I have very little use for them except for when I’m looking for anomalies.

    As you explained, it’s easy to see how Zone Finishes can be deceiving. It’s much harder to unravel QoT.

  • DaleHalas

    @John Schultz @cliffkoroll Hey John, when Bolland went down, Q used Toews against the toughest opponent. Then ran the checking line against the number two offensive line, Against the toughest top lines Hossa played with Toews. Otherwise Hossa and Sharp were playing offensively.

    Also Hossa played for awhile with Bickell and Bolland in that hybrid “Havlat” style second line that plays the other teams top lines.

    And at the very end of the year Hossa was double shifting to get more time and some of that went with the checking line.

    The thing about Sharp is if he was as good defensively as people think, Q would have him getting the second toughest opponents (Bolland getting the toughest). Think about it, Toews is stuck with Kane and Sharp gets Hossa but Toews gets the defensive assignments?

    To me Sharp should have gotten numbers similar to the year before. His splits should look like Toews, not Kane. The fact they don’t tells me Q thought of him as a defensive liability this year and treated him accordingly.

  • DaleHalas

    @John Schultz Thanks John

    QoT, to me, just has too small a sample size. The only time Bickell wasn’t on Bolland’s hip was when Bolland was hurt. Then the team isn’t as deep so Bickell’s numbers aren’t as good.

    QoC has all teams playing all their opponents so the sample size is bigger and therefore more accurate. It is just a much more useful number.

  • ChicagoNativeSon


    Well, I didn’t mention line matchups, which would affect QoC, not necessarily Zone Starts as much. I recall Q going 1vs1′s, 2vs2,s, etc a lot while Bolly was out. Here’s an example of those line matchups vs FLA on 3/23/11:

    If I remeber correctly Q was putting Hossa, Frolik and typically Kopecky out for more D-zone draws when he could control the matchups, which he could do b/c the other team was throwing their 2nd line out there for offensive zone draws and their 3rd line out for defensive zone draws. So our 2nd line was starting in the D-zone and our 3rd line in the O-zone more often than usual, especially compared to when Bolland was in the lineup.

    On SCH, members were often bitching that Kopecky was taking D-zone draws late in some games. This was the reason.

  • DaleHalas

    @John Schultz @cliffkoroll Hey John, actually you would be surprised how much the matchups affect Zone Starts. The opposing coach is manipulating his lines too. So he is throwing his top line out there more often in offensive situations and his checking line in defensive ones.

    So Sharp playing a primarily offensive game gets the checking line and more offensive zone starts. Playing a two way game it is more even. Or even for the Hawks at 54%.

    And when Bolly was out, the Hawks tended to match 1vs1′s but then matched 3vs2′s, 2vs3′s and 4vs4′s. Though I admit when Kruger and Frolik first showed up, Q was trying out everything.

    BTW, I’ve have a detailed breakdown of the last two games vs the Wings that show the 2vs3, 3vs2 matchups if you are interested? With everything on the line Jake Dowell was checking the Wings second line.

    Where we totally agree is with Frolik centering Hossa’s line. Q played that line the way you would expect a two-way center and a MARIAN HOSSA line to be played. They were in the defensive zone for draws. Looking at how Q played lines with Frolik at center compared to lines with Sharp at center is exactly my point. Frolik was indeed a heavy lifter two-way player in Florida and Q used him that way here.

    One of the reasons everyone likes Bickell/Bolland/Frolik as a third line is that Frolik is a two-way player, Sharp has that reputation but Q didn’t use him that way last year.

  • ChicagoNativeSon

    @DaleHalas That’s why i said “as much” and “when [Q] could control the matchups.”

    We’re pretty much on the same page. I understand matchups and zone starts go hand in hand. What I’m specifically talking about is when Q used Frolik-Hossa- as a two-way line, therefore getting more defensive zone draws then a normal 2nd line. That’s partly why Hossa’s offensive zone starts were so low as compared to Sharp. Sharp obviously wasn’t on the line at that time.

  • DaleHalas

    That’s one of the things I wish we could do; be able to pull stats from periods of the season. It would be so much easier to show the “pre-leddy” team, then when Frolik and Campoli showed up, all the way to the end when Kruger showed up. Totally different team by the end of the year. If they started the way they finished, well, that wouldn’t have been an 8th seeded team.

    And if we could do that, we could look at the impact to players like Sharp and Hossa when other players were added to the mix. Maybe it will be enough to look at line play game by game to “see” the trends…

  • ChicagoNativeSon

    @DaleHalas Ah ha! You’re the perfect setup man, Neal.

    I have been planning a program that does exactly that.

    Unfortunately, to be able to add all the bell’s and whistles I want, and have it auto-update, I need programming help.

    I’m gonna try contacting Verstig from SCH, but first I’m going to see if anyone has already started said project.

    A site like BtN isn’t really to difficult to duplicate if you know where to get the data. What we really need is a site that takes that raw data and makes it sortable based on date ranges, line partners, etc. Obviously the data is input shift-by-shift and game-by-game, so the data to do so already exists too.

  • K_Dog

    Man, this was great. Thank you! Should be required reading for advanced stats n00bs like me; I understand so much better now.

    The article from arcticicehockey you linked to help a lot with furthering my understand the guts of how QoC is derived.

  • DaleHalas

    @K_Dog Hey dog, thanks and glad I could help. To me, Hockey is interesting in the different levels of competition intertwined inside the exact same game. Campbell and Seabrook did play completely different classes of opponents and the advanced stats really help to point that out.