Sunday, January 2, 2011

Vs Florida 3-2 win 12/31/10

A game of hockey is usually won by the team that wins the most one-on-one battles on the ice. If we break down the game into specific situational battles the resulting numbers should tell us who won the game. These numbers also allow us to quantify players' performance by giving players credit for battles won, while keeping track of battles lost.

I've broken these one on one battles into 13 different categories.
-One on one wall battles won
-One on one wall battles lost
-Defensive zone face-offs won
-Defensive zone face-offs lost
-Offensive zone faceoffs won
-Offensive zone face-offs lost
-Giveaways
-Takeaways
-Shoot ins leading to possession
-Shoot ins leading to loss of possession
-Carry ins leading to possession
-Carry ins leading to loss of possession
-Point shots through to the net


The spreadsheet has them broken down into 26 specific situations in order to break down players' successes or failures by zones (offense, defensive, or neutral), and by manpower (power play, shorthanded or even strength).

Data gathered during the Montreal Canadiens 3-2 overtime win over the Florida Panthers on December 31st, 2010.

Legend below:



6 11 13 14 15 20 21 32 44 46 52 57 58 67 68 75 76 81 94
OZWW 1 3 7 1 2 1 1 1 1 1
1
4 3
2 8 2 OZWW
OZWL
2

1
2

2



1

1
OZWL
SKI +
5 3 1 1 1 2 1
2 1 3
2


1 2 SKI +
SKI-
3
2

1
1 2 1 1
1




SKI-
DI+ 1 2 1 1 3
1
1 1 1 1
3 2 1 2 1 1 DI+
DI-
1










1

1

DI-
DZWW 2 1 1
3 5 1 1 2

1
1 2 5 8 2 4 DZWW
DZWL 1 3 3

2 1
1 1 1 1
1
1 2
1 DZWL
NZWW

1
1 1












NZWW
NZWL
2
















NZWL
DZTA







1









DZTA
DZGA 1
1

1


1



3 1


DZGA
NZTA







1
1



1


NZTA
NZGA
2
1

1
1

2
3




NZGA
DZFOW


2














DZFOW
DZFOL


1














DZFOL
DZPKFOW


3 1













DZPKFOW
DZPKFOL



1












1 DZPKFOL
OZPPFOW
2
2














OZPPFOW
OZPPFOL
1
1














OZPPFOL
OZFOW
2
1












2 1 OZFOW
OZFOL
1
4












2 1 OZFOL
PS THRU

3

6

1




5
1

PS THRU
PS BL
1 1















PS BL





















WINS 4 15 16 11 11 14 5 3 7 4 3 6 0 10 12 7 13 14 10 WINS
BATTLES 6 31 21 20 13 17 10 3 10 10 5 10 0 16 16 9 16 17 13 BATTLES
GRADE 67 48 76 55 85 82 50 100 70 40 60 60 # 63 75 78 81 82 77 GRADE
PLAYER 6 11 13 14 15 20 21 32 44 46 52 57 58 67 68 75 76 81 94

LEGEND













OZWW OFFENSIVE ZONE WALL BATTLE WON







OZWL OFFENSIVE ZONE WALL BATTLE LOST







SKI + SUCCESSFUL SKATING OF PUCK INTO OFFENSIVE ZONE




SKI- UNSUCCESSFUL SKATING OF PUCK INTO OFFENSIVE ZONE




DI+ SUCCESSFUL DUMP IN WITH PUCK POSSESSION





DI- UNSUCCESSFUL DUMP IN (NO PUCK POSSESSION)





DZWW DEFENSIVE ZONE WALL BATTLE WON






DZWL DEFENSIVE ZONE WALL BATTLE LOST







NZWW NEUTRAL ZONE WALL BATTLE WON







NZWL NEUTRAL ZONE WALL BATTLE LOST







DZTA DEFENSIVE ZONE TAKE AWAY








DZGA DEFENSIVE ZONE GIVE AWAY








NZTA NEUTRAL ZONE TAKE AWAY








NZGA NEUTRAL ZONE GIVE AWAY








DZFOW DEFENSIVE ZONE FACE OFF WON








DZFOL DEFENSIVE ZONE FACE OFF LOST







DZPKFOW DEFENSIVE ZONE PENALTY KILL FACEOFF WON





DZPKFOL DEFENSIVE ZONE PENALTY KILL FACE OFF LOST





OZPPFOW OFFENSIVE ZONE POWERPLAY FACEOFF WON





OZPPFOL OFFENSIVE ZONE POWERPLAY FACEOFF LOST





OZFOW OFFENSIVE ZONE EVEN STRENGTH FACEOFF WON





OZFOL OFFENSIVE ZONE EVEN STRENGTH FACEOFF LOST





PS THRU POINT SHOT THROUGH PAST SHOT BLOCKING FORWARD




PS BL POINT SHOT BLOCKED BY SHOT BLOCKING FORWARD OR DEFENSEMAN



The players are given a score of one for each individual achievement or mistake. Succesful battles (dzone wall win) or a decision leading to offensive possession (successful dump in) go into the positive category, while lost battles(dzone wall loss) or a decision leading to a negative result (neutral zone giveaway) is placed in the negative category.

These results are then calculated to create a grade for each particular player for that game. This is done by dividing the positive results total by the total results (positive + negative). This result is used as the players grade for each game.

For example:

#6 Jaroslav Spacek

Positive results:
1 offensive zone battle along the wall won
1 dump-in leading to offensive zone possession
2 defensive zone battles along the wall won

Negative results:

1 defensive zone giveaway
1 defensive zone battles along the wall lost

4 positive / (4 positive + 2 negative)
=4/6

= .67
*100 (to create the grade average)
= a grade of 67% for that game

I fully understand that it is impossible to create completely objective results using this system. It is based on my ability to spot these situations and tabulate then during a game. That said, it allows us the opportunity to compare our impression of a players play during a game with results obtained from specific battles and decisions made by those same players.
I look forward to posting my calculations of coming games, as well as an ongoing total value for each Montreal Canadien skater.


6 comments:

  1. I really like what you're trying to do here. It is a more objective way of evaluating a player than is their points, plus minus, or most other statistics. The one question I have is this: how do you define whether a battle is "won" or "lost"? Is it a matter of possession? In other words, Spacek goes into a battle with Little, and after that battle whichever team gets the puck determines whether Spacek won the battle or not?

    I think this is a great first step toward an interesting rating system. A further step would be to somehow weight a "battle won" based on who that battle was against. In other words, a battle won against a superstar player might count for more than a battle won against a 3rd or 4th liner.

    All in all, great work.

    ReplyDelete
  2. I'm looking at one on one board battles for the most part. If a puck is loose and two players fight for the puck, whoever comes away with the puck won the battle. Situations are not often black or white so I have to use some hockey logic here too.

    ReplyDelete
  3. That makes sense. It would be good if you could come up with some sort of objective deciding factor for a battle won. I think your stats would be more robust if you could.

    Consider a battle in the corner of the defensive zone. If the defensive defenseman is able to knock the puck up the wall to his team mate (a winger) on the wall, wouldn't that be considered a "won battle" for the defenseman? In many ways, that d-man knocking the puck up the wall is better than him gaining possession himself. Not universally, but often.

    Maybe you could come up with the potential scenarios that can result from a one on one battle, and then define each of those scenarios as a win or a loss. For example, you could have the following results of a battle:

    A) Player comes out of battle with the puck
    B) Opposing player comes out of battle with the puck
    C) Player moves puck to a teammate
    D) Opposing player moves puck to a teammate
    E) Player moves puck to a more advantageous area (away from his net in the d-zone, toward opposing net in o-zone)
    F) Opposing player moves puck to a less advantageous area

    Then you could classify each of those outcomes (and any others you can think of) as wins and losses. This would make your statistic more transparent, and more repeatable by others.

    Again, I'm not trying to rain on your parade. I think you're doing great work here. Just hoping to add in a few thoughts to help you improve your metric.

    ReplyDelete
  4. I feel no rain here. I'm on board with you 100%. With most results not being black and white I use hockey logic to decide how to score the play. If a d-man chips the puck up the wall to his winger on the half-wall I'll score that as a puck battle won. A dump in deep resulting in a good change is scored as a good dump. A bad pinch resulting in an odd-man rush is scored as an offensive zone wall battle lost. Thanks for the interest. As I go along I'll work on possibilities for each metric and establish a system for scoring each outcome.

    ReplyDelete
  5. It might be a good idea to separate things like dumps and pinches from one-on-one battles. Not because they're unimportant, but because they're different. I think it takes a different set of skills to properly dump a puck, and to pinch properly, than it does to win a one on one battle. You might be able to come up with some more specific (and easier to track) stats if you defined a one on one battle as any time two opposing players engage in a battle for a puck.

    Then you would have some very clear defined outcomes.

    This isn't to say you couldn't track things like good dumps and good pinches, but you would just separate them out into other statistics.

    I'm very interested in the work you're doing here. I will likely continue to follow your blog closely, and chip in with the odd bit of commentary when I have the chance. As long as you don't mind!

    ReplyDelete
  6. Thanks. I enjoy the imput. I'm working on separating the metrics. My biggest concern is ensuring that the grade is representative. If I throw in too many metrics it becomes harder to track and it may skew the numbers.

    ReplyDelete