View Single Post
11-19-2012, 06:22 AM
Registered User
Canadiens1958's Avatar
Join Date: Nov 2007
Posts: 14,910
vCash: 500
Facts II

Originally Posted by ContrarianGoaltender View Post
For the record, I pointed out in the original post that the data was incomplete. The exact quote: "The numbers are not fully complete, but a significant majority of the season is covered." I sourced the data, since it was not my own work, I posted the correct numbers as presented by the original compiler, and I offered to share the data with anybody who wanted to see it. Anybody who is part of the Hockey Summary Project group on Yahoo Sports can download their own copy.

When you raised valid concerns about the incompleteness and possible bias to the data, I recalculated the numbers to account for the missing information and posted them. I have also referred to the potential data issues every time I have brought them up since. The suggestion that I was or am trying to hide something does not match the facts of the situation.

There is no mystery hand, only evidence. I am not the only poster in this thread who has brought it up, see BM67's post which does a great job of presenting the case.

The subjectivity of the home scorer can be a source of bias, but it is a consistent source of bias since they usually record games the same way night after night. This means it can be adjusted for, and the best way to deal with under- or overcounting is to look at the evidence, assess the likely magnitude of the effect, and add that back in to the goalie's stats. Another option when there is suspicion of unusual shot counting is to look at road save percentages to try to remove the effect. I have used both methods.

That does not mean that save percentages cannot be trusted, it merely means that there is an additional variable that should be considered in some cases. It's no different than taking into account team effects like shot quality for goalies on strong or weak defensive teams.

I have never claimed there was an effort to name the Vezina winner as the first AST goalie. In the original post on my blog where I pointed out the fact that the two were nearly always awarded to the same goalie, I in fact stated that I think that the voters were sincerely trying to pick the best goalie, and were probably largely unaware of how much weight they were putting on GAA in their evaluations. I still hold that opinion.

Subjective evaluations are influenced by stats, that should not be a controversial statement. Look at the goal totals and plus/minus numbers for Selke Trophy winners, for example. It is very difficult to separate the goalie from the rest of the team. That's why many voters have posted about how difficult this project has been so far.

Again, no conspiracy is required. RTSS stats vary from rink to rink, and there is some subjectivity in recording shots. Nobody has to benefit for this to be the case, it's simple operator error. Two humans will not always perceive the same event in the same way.

My only response to the suggestion that I am using scorer bias as a way to make the numbers fit my preconceived conceptions is to point out that I was originally a skeptic that shot counting was a significant issue. I changed my mind when others presented convincing arguments that it needed to be taken into account. I have almost seven years' worth of posts online that show that I have changed my opinion on a number of topics based on evidence, and when the voting records of this project are released they will show that I have changed my opinion a number of times already based on arguments other voters have presented.

I never claimed that Durnan's shots were definitely undercounted, I pointed out that there may have been potential undercounting because his home shots against rate was lower than his road shots against rate by 5.6 shots per 60 minutes of play, as recorded by S. Klages from the newspaper data.

Furthermore, goalies aren't punished when an adjustment is made for potential undercounting at their home rinks. Such an adjustment adds back the shots that were assumed to have been missed, thereby increasing the goalie's save percentage. It was in fact being charitable to Durnan to suggest it is possible that he may have got credit for fewer saves in Montreal than he otherwise deserved.

If you have specific examples of "nonsense" or data lacking integrity, please bring them up, because I don't appreciate generic and baseless accusations.

And I am not "admitting bias" at all. The whole point of bringing new information to the table is to have other people consider it and weigh in. Sometimes that new information is valuable, sometimes it is limited, that's the point of having a discussion with a group of expert voters. If you want to disregard that data because it doesn't meet your standards, then feel free to disregard it. And if you want to change the mind of the people who don't have Durnan rated as highly as you do (and it's not just me), make a good argument for him. I appreciate your efforts in the last post, although it seemed to completely omit the impact of WWII on the NHL from 1942-43 to 1943-44, and I don't agree that a goalie must have deserved a First All-Star Team because his team was weak offensively.

As for combining GVT numbers going back to 1944, the inventor of the metric Tom Awad did that in an article on the top goalies of all-time for Puck Prospectus, where he rates Durnan fairly highly based on his GAA results. I think it is important to compare across eras, and not just based on All-Star/Vezina voting, but I accept the criticism that it is better to avoid doing so where possible because the metric is based on different statistics for different years.
First two bolded quotes. Being generous you posted data that can be generously viewed as a work in progress. Now you admit that you completed it after you were caught because the data was incomplete and inaccurate. Now you admit that you had access to the additional data that would have made the post in question more complete and had the ability to complete it before posting the original incomplete and inaccurate work. Effectively you chose not to. until you were caught.

Next two bolded quotes, evidence and subjectivity. First do you have evidence that in the 1940s and into the Ron Andrews era of statistics it was actually the job of one person to regularly track shots at the home arena? The subjectivity issue is interesting since it seems to be raised by those who overlook the objective facts. In the 1940s Boston and Chicago had smaller rinks and both teams continued to play on the same smaller rinks into the 1990s. This is an objective and acknowledged fact, yet you never take it into account as a factor in goaltender performance either H/R or over the course of a season or multiple seasons. Furthermore you do not take another objective fact into consideration - scheduling. The Canadiens and Leafs always played the Sunday night game on the road of the string of two weekend games, 3 games in 4 nights, 4 games in 5 nights. Often the Sunday night opponent had the Saturday off or had not played since Wednesday or Thursday. So the SOG differential would be explainable. Yet you never raise these points.

WWII impact - fifth bolded. Not the purpose of the thread and as I've mentioned previously, the depression effects would have to be considered as well. Interesting that you claim to be able to adjust for alleged undercounting but are incapable of supporting your claim of a WWII effect with an appropriate adjustment. If it exists the WWII effect would definitely be objective so an adjustment metric should be available. Yet it is not. Why? You have such metrics - for shot counter bias, an unproven subjective phantom allegation but not for real objective facts - rink size, scheduling, WWII, coaching differentials - Vegas bookies adjust the odds for NFL coaching so it is doable for hockey, to name four.

Tom Awad. So you finally get around to crediting the source of your data. About time. Very uncomfortable with your last bolded. You accept criticism after the damage of your misrepresentation has been done. No effort to delete the damage in fact you further it.
The Broad Street Bully approach. Instead of doing positive advocacy for goalies up for discussion in the project, others get dragged into your trap of having to verify all of your incomplete, uncredited until forced, and inaccurate work. This delays the positive work of others and gives you a net advantage since the positive information gets surpressed or delayed while your incomplete and inaccurate data abounds and dominates.

Canadiens1958 is offline   Reply With Quote