Dec 152006
 

The NHL keeps track of a large variety of statistics, many of which are subjective in nature. The most subjective might be the hits statistic. What really is a hit? There is the pound the opposing player through the glass hit and then there are the almost incidental bumps which depending on severity may be classified a hit by some and not a hit by others. And looking at the stats, there is no real standard because some stats keepers in some cities seem to give players credit for a ton of hits while other cities are far more stingy in crediting players with a hit. One way to eliminate the bias teams will have by playing in their home arena 41 times a year is to just look at their road hits as a more unbiased indication of how frequently they hit opposing players.

In the table below you will find each teams home and road hits per game, as well as the visiting teams hits per game and the average of all visiting teams road hits per game (exp. hits). The hit bias gives an indication of how many more or fewer hits are credited to players in games in that city which is calculated by (home/road + visitor/exp.visitor)/2.

Team Home Road Visitor Exp.Vis HitBias
Anaheim 13.26 22.07 8.84 15.52 0.59
Atlanta 18.07 18.94 13.13 17.16 0.86
Boston 13.56 17.69 11.88 16.01 0.75
Buffalo 15.43 15.47 14.57 16.95 0.93
Carolina 28.77 14.79 24.38 16.41 1.70
Calgary 16.80 16.00 14.33 15.30 0.99
Chicago 16.71 16.06 12.86 15.82 0.93
Columbus 17.36 16.00 10.43 14.48 0.91
Colorado 9.81 8.87 14.12 15.98 0.96
Dallas 24.44 15.06 26.38 15.47 1.66
Detroit 18.93 12.27 18.53 16.07 1.32
Edmonton 15.59 16.79 15.00 15.12 0.96
Florida 20.69 20.17 14.50 16.94 0.95
Los Angeles 19.30 15.08 17.40 16.09 1.18
Minnesota 11.44 11.07 10.69 14.96 0.85
Montreal 26.18 17.36 18.47 15.81 1.35
Nashville 14.33 13.74 11.50 14.84 0.90
New Jersey 14.29 17.44 13.21 16.54 0.81
NY Islanders 23.75 20.77 16.69 16.15 1.10
NY Rangers 23.40 19.00 22.00 16.44 1.28
Ottawa 27.77 16.60 20.77 15.40 1.52
Philadelphia 14.19 15.53 13.81 16.52 0.87
Phoenix 19.81 15.47 15.38 16.34 1.11
Pittsburgh 18.31 13.79 16.50 16.83 1.14
San Jose 18.13 18.33 14.67 14.71 0.99
St. Louis 14.94 16.93 12.94 15.47 0.86
Tampa Bay 16.18 11.20 18.24 17.74 1.19
Toronto 22.81 16.00 19.56 16.92 1.29
Vancouver 16.76 14.60 22.35 16.12 1.27
Washington 18.31 17.71 18.25 16.89 1.06

As an example, Anaheim gets credited with 13.26 hits per home game and 22.07 hits per road game. Teams visiting Anaheim get credited with 8.84 hits pre game when playing in Anaheim and they average 15.52 hits per game in all of their road games. Those sets of numbers indicate a huge difference between how Anaheim game monitors credit hits than the rest of the league. Overall, hits are credited to players in Anaheim at a rate about 40% less than the average other city. On the opposite end of the spectrum,teams playing in Carolina get credited with 70% more hits than the average city.

So, armed with this information we can adjust team and players hit totals accordingly. Below is a table showing each teams adjusted hit totals which should give a much more realistic representation of a teams level of physical play.

Num Team Hits Adj. Hits
1 Anaheim 583 744
2 Florida 694 697
3 Atlanta 593 592
4 NY Islanders 650 592
5 San Jose 602 573
6 NY Rangers 674 549
7 Ottawa 693 546
8 Montreal 688 545
9 Toronto 621 522
10 Washington 541 521
11 Chicago 491 520
12 Los Angeles 582 514
13 Columbus 499 513
14 Phoenix 549 508
15 St. Louis 476 501
16 New Jersey 479 494
17 Edmonton 500 490
18 Boston 447 486
19 Carolina 655 483
20 Philadelphia 460 483
21 Calgary 476 467
22 Dallas 632 451
23 Buffalo 479 449
24 Vancouver 504 443
25 Pittsburgh 486 442
26 Nashville 433 442
27 Detroit 468 400
28 Tampa Bay 443 387
29 Minnesota 349 371
30 Colorado 290 285

The above table can also be found by clicking on the link in the left menu and that table will be updated manually. You will also find each players adjusted hits in the left menu as well. I will be updating these tables semi-regularly (hopefully at least once a week) so you can all see how your favourite teams and players are doing in the physical play department.

  10 Responses to “Adjusted hit totals”

  1.  

    Be suspicious of any sabremetric result that is entirely a function of normalization. Anaheim leading the league in hits is clearly an example of this.

  2.  

    First off, why? Why is normalization to factor out a problem that is clearly evident is such a bad thing? Why is Anaheim being at the top of the list automatically make the process wrong? Anaheim has 2 things that are key to dishing out hits: size and speed.

    And second, these are just hits which don’t really mean a whole lot so lets not take them too seriously by calling them a sabremetric.

  3.  

    I am not claiming your result is wrong. I am just very suspicious of it.

  4.  

    Why? Just because Anaheim is at the top of the list? Well, if you look, Anaheim is also at the top of the list of most road hits with 22.07 per game so they are getting credited with a lot of hits by a lot of different people in a lot of different games.

  5.  

    Why? Because you have a mid-level team in terms of their raw hits that is number one after normalization. You better be pretty certain that normalization is correct and doesn’t have any unintended consequences or miss anything significant because it is the whole story. When your analysis makes a mid-level team into the big leader, you better be 100% confident in your analysis if you want to believe the results have any meaning.

  6.  

    Understandable, but if you look at the numbers, it makes total sense. Anaheim has the most road hits and visiting teams that play in Anaheim have the fewest hits (but a relatively large margin). Either whoever is keeping track of hits in Anaheim are pretty stingyor every team that plays in Anaheim (including the Ducks themselves) play like softies when they don’t otherwise. I doubt the latter is true.

  7.  

    Clearly you didn’t make a mathematical error. Anaheim home games don’t have many hits. Its the assumptions that are behind your math i am uncertain about.

    Lets imagine that
    1) the guy who gives out hits in Anaheim is stingy
    2) the guy who gives out hits in Anaheim is strongly biased toward the Ducks and is thus really really stingy when it comes to giving visiting teams hits and only moderately stingy with the Ducks
    3) the other Pacific division hits guys are quite a bit more generous

    disclaimer: I havent ever watched a game with an eye on the guy giving out hits so I have no idea if any of these assumptions are true but they are somewhat consistent with the data.

    With these assumptions, we would have Anaheim getting far more hits than opponents at home (which is obsevred). We would have Anaheim getting more hits on the road than at home (specifically in inter-division games – this bias would serve to make Anaheim look good in your analysis and the rest of their division look bad – and looking at your statistics this may also be observed). This could very well produce a normalization result with Anaheim on top when the reality is it is a fluke based on the way hits are given out leaguewide.

    I am not claiming my scenario is reality – but I do not have enough information to rule it out. One would have to watch several games while simultaneously watching the guy who gives out hits to attempt to test my scenario.

    The problem is that I think hits (the way they are determined by the NHL) is a garbage stat. The definition varies throughout the league. And garbage in garbage out.

  8.  

    You have to also add assumption 4 that hit guys aren’t biased towards their home team in other cities. (BTW, as far as I know the stat recorders are independent of the teams and are employed by STATS Inc. so there shouldn’t be any bias)

    But we also need to factor in that Anaheim has had 15 road games in 13 different arenas (twice in LA and twice in Van) and only 3 road games against pacific division teams so that pretty much nullifies your assumption #3 as being valid.

    As for hit counter bias towards Ducks at Ducks home games, well that doesn’t seem like a valid assumption either. The Ducks have a home hits per game stat that is 60% of their road stat. Ducks opponents when playing in Anaheim have hits per game stats that are 57% of their road games hit per game averages. That is pretty consistant stinginess between Ducks and Ducks opponents.

    I am not sure hits is a garbage stat if you factor out the inconsistencies between those monitoring the games. But, I don’t think it is necessarily an significantly important stat in terms of winning games (Buffalo, Dallas, Nashville and Detroit are all near the bottom of the adjusted hits list).

  9.  

    I am not sure hits is a garbage stat if you factor out the inconsistencies between those monitoring the games

    We agree on this statement. But how certain can you be that the inconsistencies you have attempted to factor out are the only inconsistencies that exist? I think there could be more different inconsistencies than those for which you are normalizing.

  10.  

    Well, yeah, and the inconsistencies that I have factored out I probably haven’t factored out 100% accurately. But, I suspect I have factored out the most significant inconsistancy to a fairly significant level and I am fairly confident that my adjusted hit numbers are much closer reflection of a teams physical play than the actual NHL numbers.

Sorry, the comment form is closed at this time.