Ryan’s Hope

No, not that Ryan’s Hope.

At some point in the distant past, Ryan @ Lowetide wondered what would happen if you calculated Corsi by restricting it to playoff teams only.

That seemed more fun and less challenging than watching the Kings curb-stomp the Oilers, so:

Challenge accepted!

Here’s the first cut of Ryan’s stat (I was hoping to have it early in the game but I got delayed because a. dismay at game, and b. had to pause the game and go grocery shopping).

Team EV Corsi % EV Corsi % (good teams only)
ANAHEIM 51.1% 49.2%
ARIZONA 48.8% 47.3%
BOSTON 51.9% 50.5%
BUFFALO 37.9% 36.5%
CALGARY 45.0% 44.0%
CAROLINA 52.1% 51.4%
COLORADO 44.1% 43.3%
COLUMBUS 47.2% 46.0%
CHICAGO 53.8% 52.4%
DALLAS 52.0% 49.7%
DETROIT 53.5% 52.0%
EDMONTON 48.5% 48.0%
FLORIDA 51.0% 50.3%
LOS ANGELES 54.0% 52.4%
MINNESOTA 50.8% 50.9%
MONTREAL 48.2% 46.6%
NY ISLANDERS 52.9% 51.7%
NY RANGERS 49.5% 48.9%
NASHVILLE 52.3% 51.6%
NEW JERSEY 46.7% 47.0%
OTTAWA 50.5% 49.3%
PHILADELPHIA 49.6% 48.4%
PITTSBURGH 53.0% 52.1%
SAN JOSE 51.0% 49.1%
ST. LOUIS 52.1% 51.1%
TAMPA BAY 53.0% 51.7%
TORONTO 46.6% 45.7%
VANCOUVER 49.2% 47.0%
WASHINGTON 51.4% 49.7%
WINNIPEG 52.6% 51.1%

For this calculation I used this list of teams as “good” teams (but can adjust and rerun with any set of teams, so feel free to suggest a different list): ANA, BOS, CBJ, CHI, DAL, DET, LA, MIN, MTL, NYI, NYR, NSH, NJ, OTT, PHI, PIT, SJ, STL, TB, WSH.  They may end up being playoff teams this year, but I could not bring myself to include CGY or VAN!

Next step: Ryan’s analysis.  Challenge accepted?

EDIT: April 3rd.  Ryan’s first stab at assessing the data didn’t reveal much.  Related question was: maybe the criteria above (20 teams) was too broad?  I went back and after some mulling over playoff finalists and PresCup winners the last three years, I came up with this list of eight powerhouse teams: VAN, LAK, CHI, NYR, PIT, BOS, STL, DET (remember, this is backwards looking, hence Vancouver).

With this set, the data is as follows.  Not sure if it’s any better, but at first glance, this set seems more indicative of the ‘true strength’ of the teams vs the more standard metric.  Also helps to explain why BOS, DAL, and VAN are struggling to make the playoffs despite relatively gaudy metrics – they seem to pad their stats beating up on the sisters of the weak.

DET, NYI, PIT, STL, LAK, and WIN might just be dangerous (assuming they make it).

ANAHEIM 51.1% 49.7%
ARIZONA 48.8% 46.3%
BOSTON 51.9% 49.9%
BUFFALO 37.9% 33.8%
CALGARY 45.0% 40.8%
CAROLINA 52.1% 50.5%
COLORADO 44.1% 43.8%
COLUMBUS 47.2% 45.6%
CHICAGO 53.8% 50.8%
DALLAS 52.0% 46.5%
DETROIT 53.5% 51.0%
EDMONTON 48.5% 46.6%
FLORIDA 51.0% 51.0%
LA 54.0% 51.9%
MINNESOTA 50.8% 50.9%
MONTREAL 48.2% 45.9%
NYI 52.9% 51.3%
NYR 49.5% 46.5%
NASHVILLE 52.3% 49.8%
NJD 46.7% 45.7%
OTTAWA 50.5% 50.7%
PHILADELPHIA 49.6% 47.9%
PITTSBURGH 53.0% 52.6%
SJ 51.0% 48.7%
STL 52.1% 51.8%
TAMPA 53.0% 50.2%
TORONTO 46.6% 43.1%
VANCOUVER 49.2% 45.7%
WASHINGTON 51.4% 49.1%
WINNIPEG 52.6% 50.8%

Same chart as above, but with differentials added and data bars to help visualize:

2015-04-04_10-31-21

Advertisements

5 thoughts on “Ryan’s Hope

  1. Hey G, thanks so much for this. My hard drive on macbook died two weeks ago… While I had files backed up, I wasn’t using time machine. I lost my installation of SPSS! Sadly, I am no longer involved with academic research, so I’ll likely never obtain another copy.

    I did play around with your data regardless. So far, it doesn’t seem to help separate teams. It’s possible either that 20 teams is too many since there’s too much overlap or that this idea wasn’t as great as I thought it was. How would the data look if you tried using say just ten teams? The top 10 ev Corsi teams currently holding a playoff position?

    Anaheim and Vancouver seem to flag a little in terms of EV Corsi when bad teams are filtered out. Calgary’s EV Corsi is just so bad either way.

    Like

  2. To reference Mact, I’d say that visually it looks better compared to the prior iteration. I’ll have to play around with these numbers when I get some time today. Great stuff as always.

    I had the idea for this after reading an Eric T article. The links in his article are now dead since they link to extraskater, but here it is: http://www.sbnation.com/nhl/2014/1/9/5278996/los-angeles-kings-stats-shot-differential-really-good

    Anyway, I read this article a day or so after the Kings hammered the Oilers on the scoresheet and Corsi sheet too. His article was sort of trying to find out where the Kings get their Corsi superiority. While he’s talking about zone entries and shots per entry… The Kings had something like a +400 Corsi rating that season when his article was written. I was thinking, 60 of those or so Corsi events just occurred against the Oilers.

    So yeah, I started to wonder if there would be any utility in filtering out the stats padding against the sisters of the poor lol.

    To clarify, your data set is just from this season?

    Like

  3. Would like to do a binomial logistic regression with SPSS comparing the two variables to which teams are currently in a playoff position… At this point, if you sort from highest to lowest Ryan Corsi vs Corsi, you capture 12/16 playoff teams vs 11. The correlation coefficient is just slightly higher for Ryan Corsi and points (0.553 vs 0.549).

    Some teams flag a little when you look at their Ryan Corsi in the context of their PDO. Most notably, both NYR and Montreal look quite poor… Why are they in the playoffs? 🙂 They’re both #1 and #2 in Team PDO.

    http://stats.hockeyanalysis.com/teamstats.php?disp=1&db=201415&sit=5v5&sort=PDO&sortdir=DESC

    They’re also both top five in team SV%.

    Calgary is interesting. Horrendous Corsi and Ryan Corsi. They are 6th in PDO and 18th in team SV%, which screams luck.

    I agree that Vancouver really flags with the Ryan Corsi. They’re bolstered by a 7th ov PDO while not getting great goaltending. When their luck ends and they’re not padding their stats on the sisters of the poor, they’re not a very good hockey team.

    Dallas is also interesting in that they drop from 11th to 21 st in Ryan Corsi. That combined with a 21st ranked PDO is good for an early golf season.

    The other idea behind this stat is to see how useful it is for predicting playoff success. Teams don’t play the Oilers in the second season, so who cares how much they outshot them during the regular season?

    Any chance you could run the data for last season and we could see how well it predicted playoff success?

    Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s