Earlier today, I posted a look at how adjustments for score effects can help us make better predictions than narrowing our sample size to look just at tied games. Let’s take a quick look at where Score-Adjusted Fenwick differs from Fenwick Tied for this year’s teams (thanks to George E. Ays for the suggestion).
Full table after the jump.
| Team | Score-Adjusted Fenwick | Fenwick Tied | Fenwick Close |
| DET | 57.0 | 56.5 | 56.8 |
| STL | 56.2 | 56.5 | 56.1 |
| PIT | 55.5 | 54.9 | 55.4 |
| CHI | 53.0 | 53.9 | 52.7 |
| VAN | 53.0 | 50.8 | 51.4 |
| BOS | 52.9 | 52.4 | 52.7 |
| SJ | 52.3 | 51.6 | 51.1 |
| PHI | 51.1 | 48.0 | 50.9 |
| LA | 50.9 | 51.8 | 51.3 |
| WPG | 50.8 | 50.4 | 51.1 |
| COL | 50.3 | 50.2 | 51.3 |
| DAL | 50.0 | 50.0 | 49.0 |
| NJ | 49.7 | 51.3 | 50.3 |
| OTT | 49.5 | 50.4 | 50.1 |
| MTL | 49.2 | 47.8 | 48.6 |
| CBJ | 49.2 | 50.6 | 50.6 |
| PHX | 49.0 | 49.4 | 49.8 |
| NYI | 49.0 | 50.0 | 49.4 |
| FLA | 48.6 | 50.8 | 49.7 |
| TOR | 48.5 | 48.1 | 48.4 |
| EDM | 48.3 | 49.0 | 48.3 |
| WSH | 48.3 | 52.2 | 50.0 |
| CAR | 48.1 | 47.3 | 48.1 |
| TB | 47.8 | 46.7 | 48.1 |
| NYR | 47.8 | 48.5 | 48.5 |
| BUF | 47.6 | 47.1 | 46.8 |
| CGY | 47.6 | 49.0 | 47.8 |
| ANA | 46.3 | 46.0 | 44.9 |
| NSH | 46.0 | 44.0 | 45.2 |
| MIN | 44.2 | 43.7 | 45.0 |
The team that originally started me down this path was the Capitals, who have been very good in tied games but bad in all other situations. I believe Fenwick Tied overrates them considerably.
The other team substantially overrated by Fenwick Tied is Florida (48.6% Score-Adjusted Fenwick, versus 50.8% Fenwick Tied). New Jersey, Columbus, and Calgary also each look a bit overrated by Fenwick Tied.
At the other end of the scale, our very own Flyers appear to be badly underrated by Fenwick Tied (51.1% Score-Adjusted Fenwick, versus 48.0% Fenwick Tied). Vancouver (53.0% vs 50.8%) and Nashville (46.0% vs 44.0%) also appear to be badly underrated.
As should not be a surprise given how the numbers are derived, Fenwick Close often lies about halfway between Fenwick Tied and Score-Adjusted Fenwick, both in absolute value and in predictive accuracy.
