On fancystats and accurate estimates

Stats post, wading into deep water without the little swimmy things…

Expansion and sources on Wednesday’s story:

Start Here, @joshweissbock’s post Sunday night with estimated Fenwick Close and PDO numbers for the AHL. “Bridgeport is bad,” he writes.

“WATCH THE GAME, NERD” I hear a cry. Well, I do, usually. Try to. Attempt. When not chasing early deadlines and what have you. Along the way I try to tick off Corsi and note blocked shots to turn it into Fenwick. I’m sure I miss one or two, but over time I’m hoping they even out. Then it’s just counting. (Which, honestly, math is easier.)

The first definition of “Fenwick Close” that I came across was this one. I’ve since seen others. You’re killing me, nascent community codifying your ideas. But I’ll start with the first way: Shots attempted or on goal, not blocked, while the game was tied in the third period or within a goal in the first two.

And Bridgeport comes out better in that count than the @joshweissbock estimate. Hang on: it’s not an ordinal better.

Basically, the counting is not complimentary to Bridgeport in the first quarter. Click here for a big jpeg of numbers.

(FC F/A: Fenwick Close for and against in all situations in regulation time. FCES: At even strength in regulation time. FC5: At five-on-five. FCOT4: In the four-on-score part of overtime. FCOT3: In three-on-three. OT notes for special-teams time in overtime. The totals for FC and FCES first sum only regulation time, then the next line adds in overtime.)

So more or less, estimates are awesome. (Estimates are cool when you’re part of a team.)

Two weeks ago, my overall number was within a few hundredths of a percent of Josh’s estimate, which was pretty cool. The trend had been slightly upward, so I was curious which way things would go. I wanted to double check some things and get in a few more games — which turned out not to be pretty games — before I went writing. So there it is.

Michael Fornabaio