Friday, April 23, 2010

timeonice.com Script Notes. Part II.

Part I is here. You will need to read that first.
I have added some features to the even strength statistic scripts at timeonice.com:

  1. Appending &hv=1 will only report data from road games, &hv=-1 returns home game data.
  2. Appending &shawn=22 will only report events that occurred when the player wearing #22 was on the ice. By way of example http://timeonice.com/playershots0910.php?team=EDM&first=20001&last=21230&shawn=18 will output the results of every Oiler player during the time they were on the ice with Ethan Moreau at even strength during the 09/10 season.
  3. Appending &skip=20336 20636 20874 21173 will cause these games to be omitted from the results. In this case all of the EDM vs PHX games would be skipped.
If you really want to get your geek on, you can use item 3 to filter out the games that were not recorded by the scoring chance trackers for any team. And if you do that for the Oilers, accessing oilogosphere commenter quain's summary table, then you can easily look for trends, such as the one below (click to enlarge).


If you can't spot the pattern there, well you're just not trying.

13 Comments:

Blogger Olivier said...

Hah, easy!

The guys in the upper-right corner are about to get traded for energy players and draft picks.

Do I win something?

4/24/2010 1:11 am  
Blogger Sunny Mehta said...

thanks a ton, vic.

fyi appending the "&vh=1" doesn't seem to work for me. (i.e. it returns all games instead of filtering for road games)

4/24/2010 9:00 am  
Blogger Vic Ferrari said...

D'oh! That should read "&hv=1". I'll change the original post.

4/24/2010 9:22 am  
Blogger Scott Reynolds said...

This is awesome Vic! I thank you very much! And to Olivier, the sad part is that it's quite possible.

4/24/2010 9:53 pm  
Blogger speeds said...

Vic:

(1) What's the correlation coefficient? Assuming it to be high (my stats memory is fuzzy), should one take this to mean we don't really "need" scoring chance numbers, as don't seem to tell us much more than Fenwick?

(2) Is there some reason, other than chance, that we'd expect some players to have a higher scoring chance%:fenwick%? Nilsson seems to be higher by that measure than Moreau, does this tell us something about an ability for Nilsson, on average, to create higher quality chances?

(3) Interesting to see the difference between Nilsson and O'Sullivan.

4/24/2010 10:04 pm  
Blogger speeds said...

Vic:

(1) What's the correlation coefficient? Assuming it to be high (my stats memory is fuzzy), should one take this to mean we don't really "need" scoring chance numbers, as don't seem to tell us much more than Fenwick?

(2) Is there some reason, other than chance, that we'd expect some players to have a higher scoring chance%:fenwick%? Nilsson seems to be higher by that measure than Moreau, does this tell us something about an ability for Nilsson, on average, to create higher quality chances?

(3) Interesting to see the difference between Nilsson and O'Sullivan.

4/24/2010 10:04 pm  
Blogger Vic Ferrari said...

Speeds,

1. I think the correlation coefficient for this sample set was .92 for Fenwick, .89 for shots and .90 for Corsi. So much of a muchness, you wouldn't be able to tell the difference by eye. The Oiler's shot recorder is pretty good, assuming he hasn't been fired.

2. I don't know. You've picked the right guys as outliers though. If you remove those two then there was no repeatability from 08/09 to 09/10 in terms of "scoring chance % outperforming fenwick %" ... same applies to shots and corsi.

Overall the season to season "outperforming fenwick" correlation is r=.24. That's not much, and the sample is really small, so sensible people shouldn't be drawing any conclusions. Still, if we were to list the Oiler most guilty of trying to pass it into the net, Nilsson would win the ballot. If we were to vote for the Oiler most likely to shoot from an angle outside the dots ... I suspect that Moreau would win in a landslide. So I suspect that there is something to it, just that the effect is very small in the population, and very difficult to see through the noise.

3. Yeah. I've never been a Nilsson fan, but in the very first posts I ran on corsi since he became an Oiler ... he's a guy I expected to be brutal, but he manages to do okay. Scoring chances even moreso. Might be his style of game, or just a personal bias, but the chipstack is never as bad as I expect it will be for Robert.

I'm sort of the opposite with O'Sullivan. I'm always surprised by just how terrible the underlying numbers are for the guy.

And this stuff really repeats from year to year, especially if you make any sort of effort to account for the context of the icetime. In fairness to Patio, he is useful on special teams.

4/25/2010 9:06 am  
Blogger Jonathan Willis said...

Just fantastic stuff, Vic.

Thank you so much.

4/25/2010 10:30 am  
Blogger Coach pb9617 said...

Awesome stuff, Vic.

You're right, Nilsson is not nearly as bad as he's made out to be in fan circles. Much like Penner last season, Nilsson sometimes looks terrible and that sticks. Confirmation bias, I guess.

4/25/2010 5:33 pm  
Blogger rananda said...

The Oiler's shot recorder is pretty good, assuming he hasn't been fired.

solid gold. team culture aint gonna change itself, though.

4/26/2010 3:07 pm  
Blogger 332wil0佳穎4burc_custis said...

辛苦了!祝你愈來愈好!........................................

4/26/2010 8:39 pm  
Blogger 茂恒 said...

Well done!......................................................................

5/02/2010 5:25 am  
Blogger Jim Philips said...

Thanks for the two part. I read some of the second part without the first part and you are right. It doesn't make so much sense.

price per head free demo

5/30/2013 10:09 pm  

Post a Comment

<< Home