Sep 032012

A month and a half ago Eric T at had a good post on quantifying the impact on teammate shooting percentage.  I wanted to take a second look at the relative importance the impact on teammate shooting percentage can have because I disagreed somewhat with Eric’s conclusions.

For a very small number of elite playmakers, the ability to drive shooting percentage can be a major component of their value. For the vast majority of the league, driving possession is a more significant and more reproducible path to success.

It is my belief that it is important to consider impact on shooting percentage for more than a “very small number of elite playmakers” and I’ll attempt to show that now.

The method that Eric used to identify a players impact on shooting percentage is to compare that players line mates shooting percentages with him to their overall shooting percentage.  As noted in the comments the one flaw with this is that their overall shooting percentage is impacted by the player we are trying to evaluate which will end up underestimating the impact.  In the comments Eric re-did the analysis using a true “without you” shooting percentage and the impact of driving teammate shooting percentages was greater than initially expected but he concluded the conclusions didn’t  chance significantly.

Overall average for the top ten is a 1.2% boost (up from 0.9% in story) and 5 goals per year (up from 4.5). I don’t think this changes the conclusions appreciably.

In the minutes that a player is on the ice with one of the very best playmakers in the league, his shooting percentage will be about 1% better. For a player who gets ~150-200 shots per year and plays ~40-60% of his ice time with that top-tier playmaker, that’s less than a one-goal boost. It’s just not that big of a factor.

He also suggested that using the “without you” shooting percentage instead of the “overall shooting percentage” would probably result in “more accurate but less precise” analysis.  This is because a guy like Daniel Sedin would get very few shots when playing apart from Henrik Sedin because they rarely play apart and this small “apart” sample size might be subject to significant small sample size errors.

Continue reading »

Quality of Competition in 5v5 Close Situations

 Uncategorized  Comments Off on Quality of Competition in 5v5 Close Situations
Jul 112012

I have been wondering about the benefits of using 5v5 close data instead of 5v5 when we do player analysis and player comparisons.  The rationale for comparing players in 5v5close situations is that we are comparing players under similar situations.  When teams have a comfortable lead they go into a defensive shell resulting in fewer shots for but with a higher shooting percentage and more shots against, but a lower shooting percentage.  The opposite of course is true when a team is trailing.  But what I have been thinking about recently is whether there is a quality of competition impact during close situations.  My hypothesis is that teams that are really good will play more time with the score close against other good teams and less time with the score close against significantly weaker teams.  Conversely, weak teams will play more minutes with the score close against other weak teams than against good teams.

My hypothesis is that players on good teams will have a tougher QoC during 5v5 close situations than during overall 5v5 situations and players on weak teams will have weaker QoC during 5v5 close situations than during overall 5v5 situations.  Let’s put that hypothesis to the test.

The first thing I did was to select one key player from each of the 30 teams to represent that team in the study.  Mostly forwards were chosen but a few defensemen were chosen as well.  From there I looked at the average of their opponents goals for percentage (goals for / [goals for + goals against]) over the past 3 seasons in zone start adjusted 5v5 situations as well as zone start adjusted 5v5 close situations and then compared the difference to the players teams record over the past three seasons.  The table below is what results.

Player Team GF% 5v5 GF% Close Close – 5v5 3yr Pts Avg. Pts
Doan Phoenix 50.3% 50.6% 0.3% 303 101.0
Chara Boston 50.7% 50.9% 0.2% 296 98.7
Toews Chicago 50.4% 50.6% 0.2% 310 103.3
Datsyuk Detroit 50.8% 51.0% 0.2% 308 102.7
Weber Nashville 50.5% 50.7% 0.2% 303 101.0
Backes St. Louis 50.8% 51.0% 0.2% 286 95.3
E. Staal Carolina 50.4% 50.5% 0.1% 253 84.3
Ribeiro Dallas 50.5% 50.6% 0.1% 272 90.7
Gaborik Ny Rangers 50.1% 50.2% 0.1% 289 96.3
Malkin Pittsburgh 50.1% 50.2% 0.1% 315 105.0
Ovechkin Washington 49.9% 50.0% 0.1% 320 106.7
Enstrom Winnipeg 50.1% 50.2% 0.1% 247 82.3
Weiss Florida 50.3% 50.3% 0.0% 243 81.0
Plekanec Montreal 50.4% 50.4% 0.0% 262 87.3
Tavares NY Islanders 50.3% 50.3% 0.0% 231 77.0
Hartnell Philadelphia 50.1% 50.1% 0.0% 297 99.0
J. Thornton San Jose 50.9% 50.9% 0.0% 314 104.7
Kessel Toronto 50.1% 50.1% 0.0% 239 79.7
H. Sedin Vancouver 50.0% 50.0% 0.0% 331 110.3
Nash Columbus 50.9% 50.8% -0.1% 225 75.0
J. Eberle Edmonton 50.6% 50.5% -0.1% 198 66.0
Kopitar Los Angeles 50.6% 50.5% -0.1% 294 98.0
M. Koivu Minnesota 50.7% 50.6% -0.1% 251 83.7
Parise New Jersey 50.8% 50.7% -0.1% 286 95.3
Getzlaf Anaheim 51.0% 50.8% -0.2% 268 89.3
Roy Buffalo 50.3% 50.1% -0.2% 285 95.0
Stastny Colorado 50.3% 50.1% -0.2% 251 83.7
Spezza Ottawa 50.6% 50.4% -0.2% 260 86.7
Stamkos Tampa 50.2% 50.0% -0.2% 267 89.0
Iginla Calgary 50.5% 50.2% -0.3% 274 91.3
50.4% 50.5% >0 97.3
50.3% 50.3% =0 91.3
50.6% 50.4% <0 86.6

The list above is sorted by the difference between the oppositions 5v5 close GF% and the oppositions 5v5 GF%.  The bottom three rows of the last column is what tells the story.  These show the average point totals of the teams for players whose opposition 5v5 close GF% was greater than, equal to and less than the opponents 5v5 GF%.  As you can see, the greater than group had a team average 97.3 points, the equal to group had a team average of 91.3 points and the less than group had a team average of 86.6 points.  This means that good teams have on average tougher 5v5 close opponents than straight 5v5 opponents and weak teams have tougher 5v5 opponents than 5v5 close opponents which is exactly what we predicted.  It is also not unexpected.  Weak teams tend to play close games against similarly weak teams while strong teams play close games against similarly strong teams.

Another important observation is how little deviation from 50% there is in each players opposition GF% metrics.  The range for the above players is from 49.9% to 51.0%.  That is an incredible tight range and reconfirms to me the small importance QoC has an a players performance, especially when considering longer periods of time.

I also conducted the same study using fenwick for percentage as the QoC metric instead of goals for percentage but the results were less conclusive.  The >0 group had an average of 93.2 team points int he standings, the =0 group had 93.4 team points in the standings and the <0 group had 83.25 team points in the standings.  Furthermore there was even less variance in opposition FF% than GF% and only 12 teams had any difference between opposition 5v5 and opposition 5v5 close FF%.  For me, this is further evidence that fenwick/corsi are not optimal measures of player value.

Finally, I looked at the difference in player performance during 5v5 situations and found no trends among the different performance levels.  For GF% almost every player had their 5v5 close GF% within 4% of their of their 5v5 GF% (r^2 between the two was 0.7346) and for FF% every player but Parise had their 5v5 close FF% within 1.7% of their 5v5 GF% (r^2 = 0.945).  Furthermore, there was consistency as to which players saw an improvement (or decrease) in their 5v5 close GF% or FF% so it seems it might be luck driven (particularly for GF%) or maybe coaching factors.

So what does this all mean?  It means that in 5v5 close situations good teams have a bias towards tougher QoC than weak teams do.  Does it have a significant factor on player performance?  No, because the QoC metrics vary very little across players or from situation to situation (from my perspective QoC can be ignored the majority of the time).  Does it mean that we should be using 5v5 close in our player analysis?  I am still not sure.  I think the benefits of doing so are still probably quite small if there is any at all as 5v5 close performance metrics mirror 5v5 performance metrics quite well and in the case of goal metrics using the larger sample size of 5v5 data almost certainly supersedes any benefits of using 5v5 close data.


Apr 192012

Prior to the season Gabe Desjardins and I had a conversation over at where I predicted several players would combine for a 5v5 on-ice shooting percentage above 10.0% while league average is just shy of 8.0%.  I documented this in a post prior to the season.  In short, I predicted the following:

  • Crosby, Gaborik, Ryan, St. Louis, H. Sedin, Toews, Heatley, Tanguay, Datsyuk, and Nathan Horton will have a combined on-ice shooting percentage above 10.0%
  • Only two of those 10 players will have an on-ice shooting percentage below 9.5%

So, how did my prediction fair?  The following table tells all.

Player GF SF SH%
SIDNEY CROSBY 31 198 15.66%
MARTIN ST._LOUIS 74 601 12.31%
ALEX TANGUAY 43 371 11.59%
MARIAN GABORIK 57 582 9.79%
JONATHAN TOEWS 51 525 9.71%
NATHAN HORTON 34 359 9.47%
HENRIK SEDIN 62 655 9.47%
BOBBY RYAN 52 552 9.42%
PAVEL DATSYUK 50 573 8.73%
DANY HEATLEY 42 611 6.87%
Totals 496 5027 9.87%

Well, technically neither of my predictions came true.  Only 5 players had on-ice shooting percentages above 9.5% and as a group they did not maintain a shooting percentage above 10.0%.  That said, my prediction wasn’t all that far off.  8 of the 10 players had an on-ice shooting percentage above 9.42% and as a group they had an on-ice shooting percentage of 9.87%.  If Crosby was healthy for most of the season or the Minnesota Wild didn’t suck so bad the group would have reached the 10.0% mark.  So, when all is said and done, while technically my predictions didn’t come perfectly true, the intent of the prediction did.  Shooting percentage is a talent, is maintainable, and can be used as a predictor of future performance.

I now have 5 years of on-ice data on so I thought I would take a look at how sustainable shooting percentage is using that data.  To do this I took all forwards with 350 minutes of 5v5 zone start adjusted ice time in each of the past 5 years and took the first 3 years of the data (2007-08 through 2009-10) to predict the final 2 years of data (2010-11 and 2011-12).  This means we used at least 1050 minutes of data over 3 seasons to predict at least 700 minutes of data over 2 seasons.  The following chart shows the results for on-ice shooting percentage.

Clearly there is some persistence in on-ice shooting percentage.  How does this compare to something like fenwick for rates (using FF20 – Fenwick For per 20 minutes).

Ok, so FF20 seems to be more persistent, but that doesn’t take away from the fact that shooting percentage is persistent and a reasonable predictor of future shooting percentage.  (FYI, the guy out on his own in the upper left is Kyle Wellwood)

The real question is, are either of them any good at predicting future goal scoring rates (GF20 – goals for per 20 minutes) because really, goals are ultimately what matters in hockey.

Ok, so both on-ice shooting percentage and on-ice fenwick for rates are somewhat reasonable predictors of future on-ice goal for rates with a slight advantage to on-ice shooting percentage (sorry, just had to point that out).  This is not inconsistent with what I  found a year ago when I used 4 years of data to calculate 2 year vs 2 year correlations.

Of course, I would never suggest we use shooting percentage as a player evaluation tool, just as I don’t suggest we use fenwick as a player evaluation tool.  Both are sustainable, both can be used as predictors of future success, and both are true player skills, but the best predictor of future goal scoring is past goal scoring, as evidenced by the following chart.

That is pretty clear evidence that goal rates are the best predictor of future goal rates and thus, in my opinion anyway, the best player evaluation tool.  Yes, there are still sample size issues with using goal rates for less than a full seasons worth of data, but for all those players where we have multiple seasons worth of data (or at least one full season with >~750 minutes of ice time) for, using anything other than goals as your player evaluation tool will potentially lead to less reliable and less accurate player evaluations.

As for the defensive side of the game, I have not found a single reasonably good predictor of future goals against rates, regardless of whether I look at corsi, fenwick, goals, shooting percentage or anything else.  This isn’t to suggest that players can’t influence defense, because I believe they can, but rather that there are too many other factors that I haven’t figured out how to isolate and remove from the equation.  Most important is the goalie and I feel the most difficult question to answer in hockey statistics is how to separate the goalie from the defenders. Plus, I believe there are far fewer players that truly focus on defense and thus goals against is largely driven by the opposition.

Note:  I won’t make any promises but my intention is to make this my last post on the subject of sustainability of on-ice shooting percentage and the benefit of using a goal based player analysis over a corsi/fenwick based analysis.  For all those who still fail to realize goals matter more than shots or shot attempts there is nothing more I can say.  All the evidence is above or in numerous other posts here at  On-ice shooting percentage is a true player talent that is both sustainable and a viable predictor of future performance at least on par with fenwick rates.  If you choose to ignore reality from this point forward, it is at your own peril.


Mar 022012

A lot has been made about Joffrey Lupul’s “career year” this year and some Leaf fans are even suggesting that now is the time to trade him while his value is at an all-time high.  While it is true that he is on pace for career high in goals and points I would like to suggest that this is not because he is having a ‘career year’ but that he is being given greater opportunity.  He has always been this good and there is no reason to expect that he cannot repeat this years performance next season.

When I analyze a player I like to look at “on-ice” stats because I believe a player can contribute to a teams success without generating individual goals and assists.  But, since on-ice stats are teammate dependent I like to look at how his teammates do with and without the player on the ice with him.  So, let’s look at some of Lupul’s linemates 5v5 close faceoff adjusted goals for per 20 minutes with and without Lupul over the past 5 seasons.

Year Teammate Together TM w/o Lupul % Inc w/ Lupul
2011-12 Kessel 1.418 0.789 79.7%
2011-12 Bozak 1.068 1.268 -15.8%
2010-11 Bozak 0.979 0.718 36.4%
2010-11 Kessel 0.989 0.769 28.6%
2008-09 Hartnell 1.61 0.659 144.3%
2008-09 J. Carter 1.627 0.73 122.9%
2007-08 M. Richards 1.718 0.683 151.5%
2007-08 Umberger 1.915 0.631 203.5%
2007-08 Briere 1.061 0.536 97.9%

The above table includes all players Lupul has played 100 minutes of 5v5 close ice time with over the past 5 seasons including their GF20 together and Lupul’s teammates GF20 when not playing with Lupul.  The final column is how much better the teammates GF20 is playing with Lupul compared to without Lupul.  As you can see, in every single season Lupul has made his linemates significantly better offensively.  This is a good thing.

So, why are Lupul’s individual offensive numbers so much better this year?  A lot of it has to do with greater opportunity and the most important factor in opportunity is ice time.   Let’s take a look at Lupul’s even strength goal production over the past 5 seasons and compare it to his even strength ice time.

Year ES TOI ES G Min. bt goals
2011-12 984:59 17 57.9
2010-11 688:23 10 68.8
2009-10 299:05 10 29.9
2008-09 1039:42 19 54.7
2007-08 744:47 13 57.3

The “Min. bt goals” column is the average number of minutes that he spent on the ice at even strength between his even strength goals.  As you can see, this season is pretty much on par with what he has done in the past.

Another interesting thing to look at is his on-ice shooting percentage in 5v5 close zone start adjusted situations.  Over the past 5 seasons, starting with 2007-08, they are 14.04%, 12.05%, 9.09%, 11.64%, and 13.73%.  These are exceptional numbers, and among the best in the league.  I know not everyone believes in shooting percentages but I believe they are an integral component of producing offense.  As a result, a corsi-based analysis of Lupul will fail to show his true offensive value.

So, in conclusion, Lupul’s offensive production this season is not an anomaly, it is his ice time that is the anomaly.  He has almost as much even strength ice time this year than he has ever had and he has capitalized on it at more or less the same rate as he has in the past.  He is on pace for 32 goals this season and there is no reason to believe that he can’t be a 30 goal scorer next year as well.  The Leafs shouldn’t be considering trading Lupul this summer but rather they should be re-signing him to a long-term deal before his value really sky rockets in 2013 after putting up back to back 30+ goal, 70+ point seasons.


What is Rick Nash?

 Uncategorized  Comments Off on What is Rick Nash?
Feb 142012

So word has come out over the last day that Rick Nash is, at least on some level, available in a trade from the Blue Jackets.  So, the question is, who is Rick Nash and would you want him on your team?

Nash has been a Blue Jacket from the day he was drafted first overall in 2002.  He has played 648 regular season games and has scored 277 goals and 527 points.  Since the lockout he is 10th in goals (only Ovechkin, Kovalchuk, Heatley, Iginla, Staal, Lecavalier, Marleau, Vanek and Hossa) and 25 in points.  He has a pair of 40+ goals seasons and has been a 30+ goal scorer six times.  He has just 4 NHL playoff games under his belt when he scored 1 goal and a pair of assists.  He was a member of the 2010 Canadian Olympic team scoring a pair of goals and 3 assists in 7 games on route to the gold medal.  That is the raw facts that we all know about Nash.  But what about advanced statistics.

Here are my HockeyAnalysis ratings for Rick Nash over the past 4 seasons plus this season as well as his 2007-11 four year average.

2007-08 2008-09 2009-10 2010-11 2011-12 2007-11 (4yr)
HARO+ 0.991 1.070 1.257 1.502 1.079 1.200
HARO+ rank 142/235 118/241 59/245 8/260 116/229 60/217
HARD+ 0.827 0.992 0.802 0.882 0.732 0.895
HARD+ rank 164/235 96/241 196/245 162/260 197/229 162/217
HART+ 0.909 1.031 1.030 1.192 0.905 1.047
HART+ rank 172/235 115/241 123/245 36/260 169/229 95/217

HARO+ is an offensive rating, HARD+ is a defensive rating and HART+ is his total/overall rating which is simply an average of his HARO+ and HARD+ ratings.  These ratings are for 5v5 close zone adjusted situations and the rank includes any players who played 400 ore more minutes in single seasons, 300 minutes for 2011-12 partial season (through this past Saturday’s games) and 1500 minutes for the 4 year average.  These ratings take into account quality of teammates and quality of competition.


Overall in 5v5 close situations Rick Nash looks to be a solid offensive player, but not elite overall and defensively he is relatively weak.

To put Nash’s 4 year numbers in perspective, the most closely ranked players in terms of HARO+ are Cammalleri, Weiss, Hemsky, Jussi Jokinen, Vanek, Boyes, Bertuzzi, Grabovski, Alfredsson and Parise.

How about Nash’s 5v4 power play numbers.

5v4 HARO+
2007-08 1.010
2008-09 0.853
2009-10 1.203
2010-11 0.902
2011-12 0.951
2007-11 (4yr) 0.967
2007-11 rank (500 min.) 154/184
2007-11 rank (750 min.) 92/99

Generally speaking, his PP numbers are quite poor relative to other top PP forwards.

An interesting comparable is Joffrey Lupul.  It is an interesting comparable because it is quite likely that the Leafs will have an interest in Rick Nash and also because Lupul is an interesting case because he has really had a break through season this year.  Or so it seems anyway.

Nash Lupul
2007-11 5v5close HARO+ 1.200 1.385
2007-11 5v5 HARO+ 1.080 1.118
2007-11 5v4 HARO+ 0.967 1.246

It’s interesting that Joffrey Lupul ranked better than Nash in each of the three categories.  Due to injury Lupul didn’t put up 1500 minutes of 5v5 close ice time (he had 1374:44), but of all 251 players to play 1350 minutes of 5v5 close ice time Lupul ranked 10th.  When looking at these numbers it is actually not a surprise to see Lupul tied for 5th in points and 17th in goals.  He is finally being given an opportunity to play big time first line minutes with offensive zone starts and #1 PP unit ice time and as a result, he is producing.

So, getting back to Nash, let’s take a look at how he has done with his various linemates over the previous four seasons.  Here are the scoring rates (goals for per 20 minutes) for all the forwards who have played at least 250 minites of 5v5 close zone adjusted minutes during the 2007-11 four year time period.

Linemate TOI Together Nash /wo Linemate Linemate /wo Nash
Huselius 969:45 0.969 0.938 0.907
Vermette 607:35 0.757 1.016 0.782
Umberger 448:34 0.803 0.985 0.845
Brassard 441:22 1.359 0.860 0.930
Voracek 426:33 1.313 0.873 1.020
Malhotra 425:06 0.894 0.963 0.790

Nash played best when he was paired up with Voracek and Brassard and only Voracek, Brassard and Huselius made Nash a better offensive player when playing with him.  Vermette, Umberger and Malhotra were drags on his offensive numbers.  When playing apart, Voracek’s numbers are better than Nash’s.  Same for Brassard’s (who is doing it again this year, 0.782 GF20 vs Nash’s 0.613 when apart).  As an aside, the numbers suggest that Voracek is a very good offensive player  and it was probably a big mistake to trade him.  It also suggest that the Flyers aren’t getting full value from him by playing him primarily with Maxime Talbot.  If someone acquired Voracek and put him in the right situations, he could be the next Joffrey Lupul.

So, to summarize, yes Nash is a good offensive player who may put up better numbers playing with better offensive players but he is probably not an elite offensive forward.  Also, he isn’t a great defensive forward so offense really is what you get him for.  If I were Columbus I would be willing to trade him if I can get a quality NHL ready player capable of playing in their top 6 forwards, a top tier prospect and a first round pick.  If I were other teams, I would be very wary of over paying because he is not an elite player but he is paid like one ($7.8M cap hit for 6 more seasons).



Feb 092012

It has been shown on numerous occasions that players can influence their own teams on-ice shooting percentage be that through their talents or their style of play.  An example is the PDO vs Luck article I posted the other day.  In that article there is a table that clearly shows that shooting percentage varies across players and that players who are given more ice time (presumably because they are better players) have higher shooting percentages.  The same was not true for on-ice save percentage though.  On-ice save percentages were not ‘stratified’ according to ice time. That study looked at forwards and I have since looked at defensemen and have also attempted to see if organizing players according to defensive zone starts percentages would allow for ‘stratification’ of on-ice shooting percentages but to no avail.  But I am stubborn and didn’t give up.

The next thing I chose to do is compare a players on-ice save percentage with the weighted average of the save percentages of all the goalies the player played with.  The weighted average is based on the number of shots against the goalie and the player were on the ice together for.  So, lets say for example Player A was on the ice for 100 shots against, 30 of those shots were when he was on the ice with Goalie A and 70 were when he was on the ice with Goalie B.  When Goalie A is not playing with Player A his save percentage is 91%.  When Goalie B is not playing with Player A his save percentage is 92%.  The weighted average of the two goalies is (91% * 30 + 92% * 70) /100 or 91.7%.  I then compare that goalie save percentage 91.7% to the players on-ice save percentage by dividing the players save percentage by the goalies save percentage.  So, for example, if Player A’s on-ice save percentage is 92% then I calculate 92% divided by 91.7% to get 100.33.  Any numbers above 100 indicate the player improved his goalies save percentage and any numbers below 100 indicate the player hurt the goalies save percentage.

In order to get an indication of whether the player could produce that much of an improvement due solely to luck I employed a binomial distribution estimation of the likelihood that the player would have an on-ice save percentage greater than the one he posted considering the goalies he played in front of.  The results of all of this are below.  Forwards first followed by defensemen and top 25 and bottom 25 for both.  The data I used was 4 year 2007-11 5v5 zone start adjusted data and only using players with 1250 shots against.

Forward Sv% Infl. Chance > Forward Sv% Infl. Chance >
TAYLOR PYATT 101.94% 0.54% MATT STAJAN 98.87% 90.41%
MANNY MALHOTRA 101.95% 1.00% DEREK ROY 98.98% 90.50%
ZACH PARISE 101.86% 1.08% DAVID BACKES 98.90% 90.86%
JEFF CARTER 101.61% 1.32% SAM GAGNER 98.83% 91.74%
LEE STEMPNIAK 101.70% 1.34% HENRIK ZETTERBERG 98.77% 92.60%
JORDAN STAAL 101.50% 2.45% SIDNEY CROSBY 98.83% 92.92%
TEEMU SELANNE 101.51% 2.95% SHANE DOAN 98.98% 93.52%
TRAVIS MOEN 101.30% 3.59% PATRICK KANE 98.76% 93.64%
CORY STILLMAN 101.34% 3.62% DAINIUS ZUBRUS 98.67% 93.73%
RADIM VRBATA 101.34% 4.54% RICK NASH 98.77% 94.30%
TRAVIS ZAJAC 101.22% 5.22% MARTIN HAVLAT 98.64% 94.72%
BRIAN GIONTA 101.11% 6.15% MARTIN ERAT 98.75% 95.04%
SAMUEL PAHLSSON 101.22% 6.30% DAVID BOOTH 98.61% 95.77%
RADEK DVORAK 101.08% 6.99% PAUL STASTNY 98.44% 96.62%
VALTTERI FILPPULA 101.28% 7.14% ANDREW LADD 98.42% 96.99%
JASON POMINVILLE 101.01% 7.72% MARK RECCHI 98.54% 97.07%
WOJTEK WOLSKI 101.07% 8.24% EVGENI MALKIN 98.48% 97.67%
MIKE KNUBLE 101.03% 8.40% ALEXANDER FROLOV 98.16% 97.93%
MARC SAVARD 101.05% 9.02% RYAN KESLER 98.29% 98.12%
CHRIS THORBURN 101.07% 10.39% THOMAS VANEK 98.41% 98.39%
CHRIS DRURY 100.98% 11.55% TODD WHITE 98.05% 98.45%
MICHAEL RYDER 100.88% 11.62% CHRIS KELLY 98.02% 98.63%
RENE BOURQUE 100.98% 11.81% KRISTIAN HUSELIUS 97.85% 99.39%
NICKLAS BACKSTROM 100.87% 12.22% BRANDON DUBINSKY 97.51% 99.89%
MIKKO KOIVU 100.84% 12.65% ILYA KOVALCHUK 97.65% 99.96%


Defenseman Sv% Infl. Chance > Defenseman Sv% Infl. Chance >
KENT HUSKINS 102.22% 0.26% AARON WARD 99.21% 81.98%
NICKLAS LIDSTROM 102.09% 0.31% JORDAN LEOPOLD 99.23% 83.79%
ROB SCUDERI 101.78% 0.52% KEVIN BIEKSA 99.13% 84.68%
SEAN O’DONNELL 101.55% 1.26% JAROSLAV SPACEK 99.25% 84.75%
BRYCE SALVADOR 101.87% 1.28% NICK BOYNTON 99.14% 85.31%
SHANE O’BRIEN 101.63% 1.52% DAN BOYLE 99.19% 85.70%
MIKE WEAVER 101.61% 2.15% STEPHANE ROBIDAS 99.13% 87.86%
ROSTISLAV KLESLA 101.60% 3.15% SHEA WEBER 99.22% 87.88%
TREVOR DALEY 101.23% 3.16% JOHN-MICHAEL LILES 98.98% 89.07%
BRYAN MCCABE 101.25% 3.30% LUBOMIR VISNOVSKY 99.04% 90.41%
TIM GLEASON 101.20% 3.55% DENNIS WIDEMAN 99.11% 91.36%
ROB BLAKE 101.48% 3.86% MARK STREIT 98.79% 91.57%
MARC-EDOUARD VLASIC 101.22% 3.95% BRENT SEABROOK 98.94% 92.21%
PAUL MARTIN 101.37% 4.29% SHAONE MORRISONN 98.80% 92.52%
MIKE LUNDIN 101.51% 4.97% SCOTT NIEDERMAYER 98.82% 93.29%
ANDREJ MESZAROS 101.09% 5.88% ANDREJ SEKERA 98.71% 94.38%
NICK SCHULTZ 101.00% 5.96% FILIP KUBA 98.63% 94.44%
KEITH YANDLE 101.00% 6.79% MARTIN SKOULA 98.61% 95.27%
ANDREI MARKOV 101.07% 7.22% DUNCAN KEITH 98.80% 95.91%
MATT GREENE 101.14% 7.30% BARRET JACKMAN 98.73% 95.96%
ROMAN HAMRLIK 100.81% 9.44% DAN GIRARDI 98.69% 97.15%
TONI LYDMAN 100.83% 10.05% ZBYNEK MICHALEK 98.74% 97.37%
DUSTIN BYFUGLIEN 100.98% 10.12% FEDOR TYUTIN 98.53% 97.74%
JAN HEJDA 100.89% 10.14% DAN HAMHUIS 98.63% 97.87%
CHRIS PRONGER 100.89% 10.72% JACK JOHNSON 97.80% 99.95%

There were a total of 172 forwards and 141 defensemen in the study.  What is interesting is that there were 15 defensemen (10.6% of them) that had their binomial chance of posting their on-ice save percentage at 5% or lower when we would expect 7 by chance.  That means there were more than twice as many really really good on-ice save percentages for defensemen than we would expect by chance alone.

For forwards, there were just 10 who had their binomial chance at 5% or lower which equates to 5.81% so not far off of what we would expect.  We had 10 we expected 8.6.  There were 19 forwards with binomial chance <10% when we should expect 17 by chance.  Not a huge difference.  Conversely, there were 14 forwards with binomial chance >95% or 8.1% compared to the expected 8.6 players and there were 25 forwards above 90% when we should expect 17.

It seems the really good defenders are defensemen and the players most apt to hurt their goalies save percentage are forwards.

That was a pure numbers analysis, what if we looked at the players themselves.  Looking at the list of forwards with better than expected on-ice save percentages we see a lot of third line players that primarily play defensive roles (Pyatt, Malhotra, Moen, Pahlsson, Drury, Staal, etc.).  The bottom 25 forwards contain a lot of more offensive oriented players (Kovalchuk, Huselius, Vanek, Frolov, Malkin, Recchi, Stastny, Booth, Havlat, Nash, Kane, Crosby, Roy, etc.).  There is actually only a 0.04% chance (one in 2500) that Kovakchuk’s on-ice save percentage was due to luck alone.

Much the same can be said for the defensemen.  The defensemen that are  best at improving on-ice save percentage are often defensemen we consider to be defensive defensemen (Huskins, Scuderi, O’Donnell, Salvadore, Weaver, Vlasic, Martin, etc.) or elite 2-way defensemen (Lidstrom, Blake, Yandle, Pronger, etc.) and the ones at the bottom of the list are more offensive specialists (J. Johnson, Keith, Kuba, Sekera, Wideman, Liles, Visnovsky, Boyle, Streit, etc.).  Yes, this is more evidence that Jack Johnson is a horrific defensive defenseman.

All things considered, there does seem to be some order in the list and order is the enemy of luck and the binomial analysis indicates that there may be more going on than one would expect purely from luck.  It seems that players can, to some degree, influence on-ice shooting percentage.  We can’t credit, or blame, the goalies all the time.

Continue reading »

Feb 052012

One of my beefs in the analysis and evaluation of hockey players is the notion that PDO (on-ice shooting percentage plus on-ice save percentage) can be used as a proxy for luck.  A perfect example of how PDO is used as a proxy for luck is this article by Neil Greenberg about the Washington Capitals.

For example, when Alex Ovechkin has been on the ice during even strength this season, the team has a shooting percentage of 8.2 percent and has saved shots at a rate of .917. So that makes his PDO value 999 (.082+.917=.999), which is almost exactly the league average. In other words, Ovechkin has seen neither very good nor very bad “puck luck” this season.

What’s useful about this metric is that it’s “unstable,” and over a large-enough sample will regress to 1000. Why 1000? Because every shot that is a goal is a shot not saved, and vice versa.

My beef with such an analysis is the notion that for all players PDO regresses to 1000 and any players with PDO above 1000 are lucky  and any players with a PDO below 1000 are unlucky.  While I do believe luck can influence PDO over small sample sizes, not all players have a natural PDO level of 1000 and there are two reasons why.

1.  Not all players play in front of perfectly average goalies which will have a major impact on the save percentage portion of PDO.

2. Players can drive shooting percentages.

To show you what I mean on point 2, I took 4 years (2007-08 to 2010-11) of 5v5 zone start adjusted data and grouped forwards based on their ice time over those 4 years and then calculated the on-ice shooting and save percentages and PDO for each group.  Here is what I found.

TOI (minutes) SH% SV% PDO
<500 7.5% 90.9% 983.5
500-999 7.9% 91.2% 991.2
1000-1499 8.0% 91.2% 992.2
1500-1999 8.2% 91.2% 993.4
2000-2499 8.6% 91.1% 997.0
2500-2999 9.0% 91.2% 1001.9
3000-3499 9.3% 91.2% 1004.4
3500-4000 9.8% 90.8% 1006.1
4000+ 10.4% 90.8% 1012.4

PDO varies from 983.5 up to 1012.4 depending on the group’s ice time.  This is largely driven by shooting percentage which varies from 7.5% to 10.4% with the players with the lowest amount of ice time having the lowest on-ice shooting percentage and the players with the most ice time having the highest shooting percentage.  Order is the enemy of luck so seeing shooting percentages ordered this nicely tells me something other than luck is happening.  Driving on-ice shooting percentage is a skill.  This means more talented players can have a natural PDO (the PDO that they should regress to) above 1000 and less talented players can have a nautral PDO below 1000.  Factor in the goaltending and a player could have a natural PDO well above or well below 1000.

Now, this is not to say that luck isn’t a factor in a players PDO, especially over small sample sizes, it’s just we can’t estimate that luck by assuming every players natural “regress to” PDO is 1000.  Daniel Sedin has a PDO of 1043 this season (through Thursday February 2nd).  Is it fair to suggest he has been luck and should see his PDO regress to 1000?  When you consider his4-year PDO is 1035 (and his 3 year PDO is 1054) probably not.  His natural, “regress to” PDO is probably not that far off his current 1043 PDO.  Now if you are talking about Todd Bertuzzi this season it’s a different story.  Through Thursday he had a a PDO of 1056 while his 4-year PDO is 994 and he hasn’t had a PDO above 1000 in any of the previous 3 seasons.  It is probably fair to presume that Bertuzzi’s natural regress to PDO is much closer to 1000, maybe even below 1000 in which case it is fair to conclude that Bertuzzi has probably been quite lucky so far this season and is unlikely to continue at this pace the remainder of the season.

When used properly PDO can be an indication of luck but to do so we need to consider the context of a players PDO, not just assume all players PDO’s will necessarily regress to 1000.


Zone Start Effects on Stats

 Uncategorized  Comments Off on Zone Start Effects on Stats
Feb 012012

Over the past week or so I have talked about a simple and straight forward method for taking into account variations in zone starts.  The method is to simply ignore the 10 seconds following an offensive or defensive face off.  By adjusting for zone starts in this manner we can see a fairly significant impact on stats and today I’ll take a look at what gets impacted and how.

To do this I took a look at 3 year data using the 2008-09, 2009-10 and 2010-11 seasons.  Using 5v5 data for players with at least 1000 minutes of ice time I identified the 25 players who had the highest percentage of their face offs in the offensive zone and the 25 players who had the highest percentage of their face offs in the defensive zone.   I then compared their 5v5 zone start adjusted stats to their non-adjusted 5v5 stats.  The statististics I looked at are on-ice goals for percentage, on-ice fenwick for percentage, shooting percentage, opposition shooting percentage, goals for per 20 minutes, goals against per 20 minutes, fenwick for per 20 minutes and fenwick against per 20 minutes.  The changes are as follows:

Top 25 OZPct Top 25 DZPct
GF% -1.17% 2.58%
FF% -0.99% 2.32%
SH% 15.00% 12.40%
OppSh% 15.31% 11.86%
GF20 2.40% 7.00%
GA20 4.69% 2.12%
FF20 -8.28% -2.89%
FA20 -6.36% -6.93%

What is interesting is that there are relatively small differences in GF% and FF% but differences in shooting percentages are very large (note that 15% change is from, for example, 10% to 11.5%, not the actual difference in shooting percentages).  Goal and fenwick event rates are somewhere in the middle but while goal rates rise when we ignore the 10 seconds after an offensive/defensive zone  faceoff, fenwick rates drop.  This means that while a lot of shots are taken in the 10 seconds after the faceoff, very few of those shots end up as goals.  As I mentioned yesterday, the league-wide shooting 5v5 percentage in the 10 seconds after the faceoff is around 3% while it is almost 9% the rest of the time.

Let’s look at some specific examples.  Henrik Sedin gets a lot of offensive zone faceoffs and as a result 19.6% of his fenwick against events come within the 10 seconds after an offensive/defensive zone faceoff but only 8.0% of his on-ice goals do.  In real numbers, Henrik Sedin was on the ice for 2634 fenwick for events and 523 occurred within 10 seconds of an offensive/defensive zone faceoff.  He was also on the ice for 212 goals for while only 17 occurred within 10 seconds of an offensive/defensive zone faceoff.

Manny Malhotra is the opposite of Henrik Sedin and gets a lot of defensive zone faceoffs.  As a result, 17.3% of all his fenwick events against occur within the 10 seconds after an offensive/defensive zone faceoff, but only 4% of his on-ice goals against do.  In real numbers, Malhotra was on the ice for 1710 fenwick events against at 5v5 over the past 3 seasons, but 296 came within 10 seconds of an offensive/defensive zone face off.  He was also on the ice for 75 goals against, but only 3 came within 10 seconds of an offensive/defensive zone faceoff.

What does this all mean?  It means that if you are doing a corsi/fenwick/shot/shooting percentage based analysis accounting for zone starts is really important because it can have significant impacts on these stats (less so for ratios though).  The impact on goals is much less significant but probably not something we would want to ignore depending on the analysis.  May as well use the 10 second zone start adjusted data for all player analysis.


Jan 262012

With the re-signing of John-Michael Liles the Leafs now have an abundance of defensemen signed under control for a number of years, many with big dollar contracts too.  We all have our varying opinions on the relative values of each of these defensemen but I thought it would be an appropriate time to take a closer look at them statistically.


2011-12 HARO+ 2010-11 HARO+ 2010-12 HARO+ 2011-12 FenHARO+ 2010-11 FenHARO+ 2010-12 FenHARO+
JOHN-MICHAEL LILES 1.23 1.03 1.11 0.96 0.99 1.00
CODY FRANSON 1.20 1.06 1.10 1.05 1.05 1.03
LUKE SCHENN 1.10 1.08 1.08 0.85 1.02 0.99
DION PHANEUF 1.01 1.08 1.05 1.00 0.99 1.03
CARL GUNNARSSON 1.05 1.00 1.02 1.04 0.92 0.94
MIKE KOMISAREK 1.10 0.96 1.00 1.02 0.90 0.86
KEITH AULIE 0.90 1.02 0.99 0.78 0.86 0.89

The above list are my own offensive ratings (goal based and fenwick based) for 5v5 zone start adjusted (10 seconds) situations sorted by their year and a half (2010-12) HARO+ ratings.

The list generally fits with what we might expect though the one surprise is probably Luke Schenn being rated so highly offensively.  I had a debate with a few people last week where I suggested that Schenn is as good offensively as Phaneuf and got ridiculed for making that statement but the numbers do in fact support that.  The above are based on ‘on-ice’ numbers but individual stats make Schenn look good too.  This season Phaneuf has 2 even strength goals and 13 even strength points while Schenn has 1 even strength goal and 11 even strength points but Phaneuf has played more than 30% more even strength minutes than Schenn.  Last season Phaneuf had 5 goals and 17 points at even strength versus Schenn’s 5 goals and 21 points in 20% more minutes.  Combined Schenn has 6 goals and 32 points in 2237 ES minutes while Phaneuf has 7 goals and 30 points in 2207 ES minutes.  That’s awfully close offensive production if you ask me.  The difference in their overall totals is solely due to Phaneuf’s PP minutes and Schenn’s lack of them.

Getting back to the rest of the team, it is no surprise to see Liles and Franson at the top of the list.  They are known to be more offensive specialists and the stats bear that out.  The reverse is true for Komisarek and Aulie who are viewed as more defensive defensemen and that is the role they are assigned.  They simply do not produce much offense.  We only have half a season of Jake Gardiner, but so far so good.  While his fenwick offensive numbers aren’t crazy good, his HARO+ rating is very very good.  I think Gardiner is someone we can be cautiously optimistic will develop into a very good (maybe Liles-like) offensive defenseman.

For interest sake, here are the players raw offensive numbers for the last 2 seasons combined sorted by GF20.

2010-12 GF20 2010-12 FF20
LUKE SCHENN 0.88 12.62
CODY FRANSON 0.88 13.22
DION PHANEUF 0.85 13.01
KEITH AULIE 0.81 11.41

Gardiner’s GF20 is 0.96 and FF20 is 12.62 so far this season.


This is the defensive equivalent of the above offensive rating chart.

2011-12 HARD+ 2010-11 HARD+ 2010-12 HARD+ 2011-12 FenHARD+ 2010-11 FenHARD+ 2010-12 FenHARD+
CODY FRANSON 0.77 1.39 1.15 1.02 0.98 1.00
KEITH AULIE 0.71 1.22 1.05 0.89 0.87 0.85
DION PHANEUF 0.87 1.07 1.00 1.04 0.94 0.97
CARL GUNNARSSON 1.04 0.86 0.95 1.00 0.94 0.99
LUKE SCHENN 0.83 0.88 0.88 0.89 0.93 0.90
JOHN-MICHAEL LILES 0.85 0.86 0.87 1.00 1.00 0.99
MIKE KOMISAREK 0.78 0.74 0.76 0.90 0.93 0.95

There are definitely some surprises in the above list and there are probably some small sample size issues going on.  Franson looked awesome defensively last season but terrible this season when considering their goal based HARO+ numbers.  The same is true for Aulie, and to some extent Phaneuf while the reverse is true for Gunnarsson.  For each of them their Fenwick numbers are a little more consistent.

All-in all though, Franson looks like he could be a more than respectable defenseman defensively.  His fenwick ratings are pretty solid and his 2-year goal ratings are very good.  On the other side of the spectrum, Komisarek looks awful, regardless of whether you consider goal ratings or fenwick ratings.  This is not good for a guy who doesn’t produce offense either.  Luke Schenn’s defensive numbers are a little better than Komisarek’s but still not great, but at least he is producing offensively.

Again, for interest sake, here are each defenseman’s 2-year raw defensive numbers.

2010-12 GA20 2010-12 FA20
CODY FRANSON 0.67 13.06
KEITH AULIE 0.73 15.36
DION PHANEUF 0.78 13.48
LUKE SCHENN 0.88 14.51

Gardiner’s GA20 is 0.80 and FA20 is 13.83 so far this season.

Contract Status and Moving Forward

Phaneuf and Komisrek are signed for 2 more seasons at $6.5M and $4.5M cap hits respectively.  Liles and Schenn are signed for 4 more seasons each at $3.875M and $3.6M cap hits respectively.  Carl Gunnarsson is signed for another season at $1.325M when he becomes an RFA and will be due a substantial raise.  Cody Franson is set to become an RFA this summer and will deserve a sizeable raise from his current $800K salary.  Jake Gardiner has 2 years left on his entry level deal with a $1.1M cap hit and Keith Aulie is an RFA this summer.  The Leafs also have Korbinian Holzer, Jesse Blacker and others in the farm system ready to make a push for a roster spot on the Leafs in the next year or two.

The Leafs salary cap hit for their defensemen next season will be $21M plus whatever Cody Franson gets on a new contract which quite likely will be around the $1.5-2.5M range.  That would bring their expenditures on defensemen to $23M which actually isn’t all that ridiculous if the salary cap is $65+M.  That said, if they are looking to free up salary to spend on forwards and/or are looking to open up a roster spot for their young defensemen there are a few options.

The first option is to trade (if possible) Mike Komisarek.  He provides no real value to this team but then he will probably provide no value to any team so trading him might be difficult.  He also has a limited no trade clause limiting the number of potential trade partners as well.  He would be a perfect candidate to have his contract buried in the AHL (in actual dollars he’ll earn $3.5M in each of the next 2 seasons and coincidentally Jeff Finger’s buried $3.5M contract expires this summer) but he has a no movement clause which means he cannot be demoted.  The only option to get his contract off the books is via trade.

Another option is to trade Luke Schenn.  He provides some value to the Leafs with his offensive ability but that is not an area where the Leafs are lacking (most of their defensemen have offensive capabilities).  His poor defensive numbers make him expendable in my opinion and being young and on a reasonably priced long term contract he should have a lot of value on the trade market.  He could feasibly be used in a package to land the Leafs the big two-way forward they desperately need.

The other options are trading either Franson or Gunnarsson.  Neither would save the team as much cap space as either Komisarek or Schenn but both would have good value on the trade market.  That said, I would not be a proponent of this as I think they both provide good value to the Leafs, and are likely to provide good value for many years.  Gunnarsson has developed into a solid all-purpose defenseman and I think Franson has that ability too.


Importance of Quality of Competition/Teammates

 Uncategorized  Comments Off on Importance of Quality of Competition/Teammates
Jan 252012

Whenever I get into a statistical debate over which player might be better than another the inevitable argument that comes up is “yeah, but player A plays against tougher competition and gets tougher assignments” which is a valid argument to make.  But how valid?  The other day I looked at a simple, straight forward method for accounting for zone start differences (which can be significant) and today I thought I’d take a look at quality of teammates and quality of competition.

Whenever I browse through my site or in my own database I have always been curious about the general lack of variation in the quality of competition and to a lesser extent quality of teammate stats (especially over multiple seasons of data) and I thought it would be worth while taking a look at it more closely.

My stats site has a number of metrics that we can look at but let me define a few.

  • GF20 – Goals For per 20 minutes of ice time.
  • GA20 – Goals Against per 20 minutes of ice time.
  • TMGF20 – Weighted average (by ice time played with) of teammates GF20
  • TMGA20 – Weighted average (by ice time played with) of teammates GA20
  • OppGF20 – Weighted average (by ice time played against) of opponents GF20
  • OppGA20 – Weighted average (by ice time played against) of opponents GA20

I also have the same stats for fenwick as well identified with an F instead of a G in the above abbreviations.

So, let’s take a look at a players offensive capabilities.  Things that would affect a players GF20 are the players own offensive talents, the offensive talents of his teammates and the defensive talents of his opponents.  We know that not all players have the same talent level, but what about the talent levels of his teammates and his opposition?  What is the variation among them?

The above table shows the mean goal production (GF20) in blue along with lines representing + and – one standard deviation.  Also included is TMGF20 in green and OPPGA20 in red and their + and – standard deviation lines.  I have included data for one, two, three and 4 seasons of data and skaters with a minimum of 400 minutes of 5v5 ice time average per season.

As you can see, there is very very little variation in quality of opposition, almost to the point we can almost  ignore it.  The variation in quality of teammate is significant and cannot be ignored and while it seems to get reduced over time, it’s impact cannot be ignored even when using 4 years of data.

Here is the same chart except using fenwick stats instead of goal stats.


We see pretty much the same thing when we look at fenwick data as we do goal data.  There is very little variation in quality of opposition, but significant variation in quality of teammate.  What about on the defensive side of things?


Once again, the quality of opposition has very little variation across a group of players almost to the point that it can be ignored.

All of this tells us that when comparing/evaluating players, the quality of competition a  player faces varies very little from player to player and we should be really careful when we use arguments such as “Player A faces tougher quality of competition” because in the grand scheme of things, the quality of competition probably only has a very minor influence on Player A’s on-ice stats.  And if you think about it, this probably makes sense.  If you have a great offensive player, the theory is your opponents will want to match up their great defensive players against him.  But, at the same time you are trying to match up your great offensive player against their weakest defensive players.  When at home, you get the line matching advantage, while on the road your opponent does.  When all is said and done everything more or less evens out.