21st Century Hockey Analysis Begins

by Dirk Hoag on Oct 9, 2006 5:07 PM EDT

It's late in the game, you need a goal, and there's a key faceoff coming up in the offensive zone. You need to get a shot on net right away, but who are you going to send out to take the draw? Common wisdom would suggest running down the player statistics by Face Off Win Percentage, and sending your best player out there, assuming he's not winded from a recent shift.

Chances are, however, that the common wisdom is wrong.

I've recently gathered detailed game information from the entire 2005-2006 regular season, and analysed the results of over 45,000 non-neutral zone faceoffs that took place across more than 1,200 games. Based on when a faceoff in either end of the ice resulted in the offensive team getting a shot within 5 seconds of the drop of the puck, I've ranked the players as to how often offensive zone faceoffs result in immediate shots (given a minimum of 100 such draws). On average, a shot (defined as either a goal, save, blocked shot, or missed shot) took place within 5 seconds of the puck drop 9.09% of the time last season. So who fares better or worse than expected when looking at this particularly important slice of game performance? The answers might surprise you.

Take Yanic Perrault, for example, often lauded as one of the best faceoff men in the game. In 358 offensive-zone draws, only 19 shots resulted, for a paltry 5.38%. If you look at the league leaders in FOWin% from last season, Anaheim's Andy McDonald and Atlanta's Bobby Holik ranked 6th and 8th respectively. Pretty strong faceoff guys, right? When it comes to creating shots they rank near the bottom, however, producing at only 6.5%, well below the league average. Anaheim would do better using Todd Marchant (13.33%), and Atlanta using Steve Rucchin (10.45%).

On the plus side, look at Chris Kelly, from Ottawa. He ranks 80th out of 85 players in the NHL's official faceoff numbers, but he produced shots in the offensive end 12.81% of the time, 20th on this list and second-best on his team behind Antoine Vermette. The leader in this category was Montreal's Tomas Plekanec, who produced shots at a 17.78% clip. Montreal would be wise to invest in some heavy-shooting defensemen, as Steve Begin came in 3rd at 16.00%.

Peruse the list below, and add some helpful commentary the next time you're yelling at the hometown coach during your next NHL game. And stay tuned for more of this kind of detailed analysis coming in the weeks ahead. I'm excited about the data I've assembled and am looking for promising areas to research, so if you have a suggestion, feel free to leave a comment...

The Top Shot Producers (with last year's team):
Tomas PLEKANEC, MTL 17.78%
Antoine VERMETTE, OTT 17.42%
Steve BEGIN, MTL 16.00%
Marty REASONER, EDM 15.88%
Rod BRIND'AMOUR, CAR 15.44%
Jamaal MAYERS, STL 15.38%
Michael PECA, EDM 15.11%
Brett MCLEAN, COL 14.61%
Jarret STOLL, EDM 14.17%
Eric BELANGER, LA 14.16%
Travis GREEN, BOS 13.90%
Jan HRDINA, CBJ 13.87%
Clarke WILM, TOR 13.64%
Alexei ZHAMNOV, BOS 13.59%
Stephane YELLE, CGY 13.59%
Alyn MCCAULEY, SJ 13.50%
Todd MARCHANT, CBJ/ANA 13.33%
Petr CAJANEK, STL 12.91%
Jerred SMITHSON, NSH 12.87%
Chris KELLY, OTT 12.81%

The Rally Killers:
Mark BELL, CHI 3.30%
Marcel GOC, SJ 3.83%
Dave ANDREYCHUK, TB 3.96%
Keith TKACHUK, STL 4.20%
Serge PAYER, FLA 4.42%
Teemu SELANNE, ANA 4.64%
Rob NIEDERMAYER, ANA 4.91%
Vaclav PROSPAL, TB 5.07%
">

Tweet Comment 12 comments | 0 recs |

Do you like this story?

Comments

Wow my friend,very interesting.Do the numbers change when looking at say 7 seconds?I'm very impressed with the amount of work you've done.Thanks.Will post a link to your story on my site.

www.LightningHockeyCentral.com

TLW

by TLW on Oct 10, 2006 5:55 AM EDT reply actions

Brilliant work

by Martin on Oct 10, 2006 8:12 AM EDT reply actions

Sorry to use a comment for this request, but I can't locate your email address. I'd like to take a look at the compete data you've compiled to see how things compare team by team, if possible. I'd appreciate it if you could contact me at martinthornell@gmail.com
Thanks

by Martin on Oct 10, 2006 8:18 AM EDT reply actions

Thanks, TLW - I've added a link to your site in my sidebar. Nice site there!

by The Forechecker on Oct 10, 2006 12:00 PM EDT reply actions

It would be interesting to see the raw numbers that those percentages were derived from and do some statistical test on the significance.

by Steve C on Oct 10, 2006 3:22 PM EDT reply actions

I am glad people are starting to do this kind of research, but this particular study seems flawed - if indeed it is based on all offensive draws. Clearly the objective for most offensive zone faceoffs is not simply to generate a shot as quickly as possible, but rather to gain puck possession in order to generate the best possible offensive opportunity.

Thus it is probably not surprising that there are so many defensive forwards on your list - in even strength, non-"close-and-late" situations, simply drawing it back and firing it on net is a relatively conservative, low percentage play. Most higher percentage faceoff players and better offensive players would likely try to generate higher-quality scoring opportunities when they have the time to do so.

It might be interesting, as tlw suggests, to look at a longer timeframe - say, 10 to 15 seconds - not as a means of judging individual faceoff-takers, but as a means of figuring out which forward lines are capable as a whole of generating quick scoring chances (though even these results might be of limited use, considering how many close-and-late situations are 6 on 5 or other unfamiliar odd-man situations).

by Anonymous on Oct 10, 2006 4:54 PM EDT reply actions

As with any statistical analysis, this has to be set in context - what exactly do the numbers tell us, and in this case we're talking about a particularly narrow slice of the game.

I could certainly rerun the numbers on a team-by-team basis using a 10-second window, and will post the results in another piece here. Any more than 10 seconds, and you're looking at enough time to clear the zone and come back in, which pretty much removes the immediate effect of the faceoff.

Keep the feedback coming - these are early days for working with this type of information, and refinement will be important going ahead.

Thanks!

by The Forechecker on Oct 10, 2006 5:18 PM EDT reply actions

Anaheim would do better using Todd Marchant (13.33%)

Sorry I'm commenting late, but even though these statistics are interesting, is there anything to really suggest that the ability to shoot right after the faceoff really leads to goals?

I don't see a lot of the leaders on this list as 'assist' leaders, so other than contributing in a trivial (guess who's best?) way, what real benefit would there be to have Todd Marchant replace Andy McDonald in the offensive faceoff circle?

More saves for the goaltender?

by Earl Sleek on Oct 11, 2006 5:01 PM EDT reply actions

You won't see this as a determiner of top assist guys, because, as you say, goals immediately off a faceoff are a small percentage of the overall total. That said, the intent here is to isolate how often shots get created - whether a goal results is more a function of the shooter and the goalie.

You're right, it's a measure of a particular slice of the game, not the whole package. But in general, this type of information is sorely lacking in the game of hockey. Look at baseball, and it's easy to find who's best at driving in runners from scoring position with 2 outs, or in football, a quarterback's performance on 3rd & long. What I'm trying to get started here is a dialogue as to how we can develop such information around hockey - there's real potential here to map out new territory.

by The Forechecker on Oct 11, 2006 5:46 PM EDT reply actions

I'm not against data exploration by any means (in fact, you just earned a spot on our sidebar), but there's a difference between driving in a run and generating a shot on goal. One affects the score and one might or might not.

Shots within 7 (or 10) seconds of a won faceoff--with what percentage to they go in the net? Is this different than what we might observe for all shots from similar distances?

Does getting a shot off a faceoff indicate much more than what has been measured?

by Earl Sleek on Oct 11, 2006 6:10 PM EDT reply actions

I recently did a quick analysis on face-offs and found that shooting % for these shots is much lower (about half), so it's bad to shoot after a face-off...

At the end of the game I'd want the player with the best shooting %....

by JavaGeek on Nov 26, 2006 4:23 PM EST reply actions

You should probably also check that the players you are examining have taken a minimum number of offensive zone faceoffs before including their percentages. Higher percentages resulting from smaller samples would skew the data quite significantly.

by Stephen on Oct 22, 2007 10:41 AM EDT reply actions