clock menu more-arrow no yes

Filed under:

WOWY Is Worthless: Part 3

WOWY is completely worthless. It's things we know from Corsi plus a big dose of statistical noise. In Part 3, I use Linesmen as a placebo group to look at WOWY.

One of these guys makes Chris Stewart 19% better.
One of these guys makes Chris Stewart 19% better.
Bina Schroer

Once again, let me first review WOWY a little. Looking at the range of observed WDiff we see

group

mean

sd

0%

2.5%

50%

97.5%

100%

n

All

-0.011

0.166

-1.000

-0.445

-0.006

0.432

1.000

24875

100/100

0.000

0.055

-0.237

-0.109

0.000

0.107

0.199

10623

200/200

0.003

0.048

-0.221

-0.091

0.003

0.098

0.196

6947

300/300

0.004

0.045

-0.164

-0.084

0.004

0.094

0.190

4893

600/600

0.010

0.040

-0.117

-0.067

0.011

0.093

0.151

1335

WOWL

So I went through the 2013-14 season and computed WOWY for every skater with each of the NHL Linesmen (WOWL). Linesmen are on the ice when Corsi events happen but have no effect on Corsi production rates. So what does the distribution of WOWL WDiff look like? Once again I looked at all pairs, 100/100, and 200/200. There weren't any skater-lineman pairs with more than 300 events With.

Group

mean

sd

0%

2.5%

50%

97.5%

100%

n

All

-0.001

0.101

-0.581

-0.215

-0.001

0.208

0.635

23651

100/100

0.001

0.056

-0.241

-0.111

0.001

0.110

0.219

8490

200/200

0.002

0.046

-0.138

-0.088

0.002

0.094

0.167

1931

The standard deviation and ranges of the 100/100 and 200/200 groups are essentially identical for WOWY WDiff and WOWL WDiff. The "effect" of Linesmen is the exactly same as the effect of skaters.

Linesmen have no effect on Corsi production but there are a number of skater-linesman WOWL pairs where the linesman appears to have a substantial effect on the player. That's not effect. That is simply randomness.

A Blues Example

Chris Stewart "with" Linesman Bryan Pancich 119 Events 65.5%

Chris Stewart 'without" Bryan Pancich 1068 events 46.5%

WDiff +19.0%