View Single Post
Old
02-23-2013, 02:53 PM
  #171
Replacement
Now with 9% more zen
 
Replacement's Avatar
 
Join Date: Apr 2005
Location: Hockey Hell
Country: Canada
Posts: 38,072
vCash: 500
Quote:
Originally Posted by Red Deer Rebel View Post
I'm sure he uses a crawler to mine data off the NHL site, and doesn't put these numbers in manually, but you are right - that site is pretty much an amateurish mess.

I've noticed errors before, and this one comes to mind: Look at Rob Schremp's +/- on the nhl.com website for the 2008/09 season => +2.

On behindthenet.ca, it is 0.

I went through the official Game Reports at the time (for the 4 games in question), and behindthenet was wrong.

I mention this because I noticed this discrepancy back when BBO was getting in Schremp debates, and his detractors kept quoting the inaccurate stats from behindthenet. The comments section on Lowetide's blog is a major offender in citing this garbage as well.

Glad you brought this up, because the "stats experts" both here and on the Oilogosphere have never noticed this stuff, even though they are constantly quoting it - which raises concerns about their own competence. This topic is worthy of its own thread, actually.

As the old adage goes: "Garbage in, garbage out" - and this pretty much summarizes behindthenet.

ps. For ease of reference, here are the links to the 4 official Game Reports showing Schremp with a net +2 at even-strength for the 2008-09 season:

http://www.nhl.com/scores/htmlreport...9/GS020418.HTM - #88 not on ice for any GF or GA => +0
http://www.nhl.com/scores/htmlreport...9/GS020390.HTM - #88 on ice for 1 5x4 GF => +0
http://www.nhl.com/scores/htmlreport...9/GS020379.HTM - #88 on ice for 2 5x5 GF, 1 GA => +1
http://www.nhl.com/scores/htmlreport...9/GS020365.HTM - #88 on ice for 2 5x5 GF, 1 GA => +1
We definitely agree on this. I've noted inconsistencies in their data for years. One of the easiest ways to detect it is if the GP for the player don't matchup with what they are on the NHL site. For instance due to missing the LA game the GP for the players is one less. Sometimes they find these errors much later.

Also, Behind the net used to have easy to find disclaimers that go into detail on how they tabulate and where there might be some difference. For instance at one point they didn't count (afairc) GA, GF that were obtained in pulled goalie situations, then they did, then they didn't due to it not being representative of typical 5 on 5 play. Trouble is if you can't make up your mind how to tabulate and what to include and you change it up from time to time then year to year comparisons tend to be null and void.

This not even getting into so many of the "advanced stats" being suspect on their own criteria.

Replacement is offline   Reply With Quote