Following up on our look at the extent to which personal judgement might affect the compilation of the NHL's Extended Statistics, I bring you today a look at Giveaways and Takeaways. In a nutshell, the numbers seem to show once again that the level of Giveaways and Takeaways per game is primarily dependent on the site of the game, not the teams involved. Compared to the Missed Shot, Gives and Takes don't have much of a direct role relative to other hockey stats, in the sense that whether or not an official scorer calls something a Giveaway or a Takeaway doesn't impact the value of key stats such as Shots, Goals, Assists, or Saves. Nevertheless, if the NHL is going to publish statistics, it has an interest in ensuring valid data is being put out there, and what I'm seeing is a hodge-podge of data that defies logical explanation.
Within the following tables, the Road column shows a teams average events per game while on the road, Home shows the number for home games, and Visitors shows the value for other teams playing in their building. Let's start with Giveaways first...
So what is it about those Oilers fans that cause Edmonton to give the puck away three times as often on home ice? And gosh, is the ice so bad in Dallas that the puck hops away from players who can't control it? Jumping down to the bottom of this list, it seems like teams hardly give it up at all in Columbus or Chicago. When you look at how teams perform on the road (when they play in front of a myriad of official scorers), the numbers fall into a pretty tight range, with all but a few between 7 and 9 Giveaways per game. When you isolate the action down to a single official scorer, however (in the case of the Home and Visitors columns), the numbers go all over the place. The Visitors column helps isolate the official scorer's tendency even more, because we're basically seeing how they perform relative to a wide variety of teams (albeit facing a common opponent). The bottom line is that there is clearly some major differences in opinion as to what consitutues a Giveaway between scorers for the teams at the top and bottom of this table.
And now, the Takeaways...
Again, we have a similar story. At home, the Canadiens and Avalanche (and their visiting opponents) are borderline kleptomaniacs, but on the road, they fall right into line with the average value of 6.2 Takes per Game. Similarly, Takeaways are almost unheard of in Buffalo, Chicago and Pittsburgh. Buffalo and Pittsburgh are especially obvious problems when you see that their Road totals are two of the three best in the league, so the consensus opinion would seem to indicate they swipe the puck more than most.
It's truly unfortunate that we see such wide disparity in these numbers, based on the location of the game (and hence the official scorer involved). Some level of individual judgement is naturally expected, but the levels seen here too troubling to be ignored, and the NHL would be wise to address this, either by drafting and rolling out some standards (much like they do with refereeing) or by abolishing these stats altogether. In the past, I've done some pieces that I thought were pretty interesting, and while I knew the data to somewhat open to interpretation, the insight gleaned from relative values could still be useful. This look at things makes me doubt whether much use, if any, could be made out of using Giveaways and Takeaways at all, however. Unless anyone has some good suggestions, I think I'll hold off from using these stats in any future analysis...
Within the following tables, the Road column shows a teams average events per game while on the road, Home shows the number for home games, and Visitors shows the value for other teams playing in their building. Let's start with Giveaways first...
So what is it about those Oilers fans that cause Edmonton to give the puck away three times as often on home ice? And gosh, is the ice so bad in Dallas that the puck hops away from players who can't control it? Jumping down to the bottom of this list, it seems like teams hardly give it up at all in Columbus or Chicago. When you look at how teams perform on the road (when they play in front of a myriad of official scorers), the numbers fall into a pretty tight range, with all but a few between 7 and 9 Giveaways per game. When you isolate the action down to a single official scorer, however (in the case of the Home and Visitors columns), the numbers go all over the place. The Visitors column helps isolate the official scorer's tendency even more, because we're basically seeing how they perform relative to a wide variety of teams (albeit facing a common opponent). The bottom line is that there is clearly some major differences in opinion as to what consitutues a Giveaway between scorers for the teams at the top and bottom of this table.
And now, the Takeaways...
Again, we have a similar story. At home, the Canadiens and Avalanche (and their visiting opponents) are borderline kleptomaniacs, but on the road, they fall right into line with the average value of 6.2 Takes per Game. Similarly, Takeaways are almost unheard of in Buffalo, Chicago and Pittsburgh. Buffalo and Pittsburgh are especially obvious problems when you see that their Road totals are two of the three best in the league, so the consensus opinion would seem to indicate they swipe the puck more than most.
It's truly unfortunate that we see such wide disparity in these numbers, based on the location of the game (and hence the official scorer involved). Some level of individual judgement is naturally expected, but the levels seen here too troubling to be ignored, and the NHL would be wise to address this, either by drafting and rolling out some standards (much like they do with refereeing) or by abolishing these stats altogether. In the past, I've done some pieces that I thought were pretty interesting, and while I knew the data to somewhat open to interpretation, the insight gleaned from relative values could still be useful. This look at things makes me doubt whether much use, if any, could be made out of using Giveaways and Takeaways at all, however. Unless anyone has some good suggestions, I think I'll hold off from using these stats in any future analysis...