If anybody else wants to be part of the
Goalie
Overall
Analysis
Tinker
Committee
Organization
(i.e. GOATCO)
... just send me a PM and I'll email you the spreadsheet.
The algorithm I've set up so far is really, really close.
Some steps for tinkering:
1. On the far left, there are some red/green cells that align in purple rows.
The purple rows are "goalie battles", where Yahoo "rank" has two goalies with nearly adjacent rank and, thus, very close total scores.
2. The value in the red/green is the "player score" that slides in between those two ranks. As a summary:
Yahoo Rank #307 Danny Taylor (CGY-G)
Yahoo Rank #308 David Moss (PHX). I know he is a skater score = 62.34.
Yahoo Rank #309 Josh Harding (MIN-G)
So... I need to get Danny Taylor's score above 62.34 and get Josh Harding's score below that 62.34. There is also a G-Base score (brown) that I believe is a "base" amount credited to ALL goalies, just to benchmark them among the skaters.
(A goalie with NO stats is pegged as Yahoo Rank #266-#304, which is some value between skater #264 Colin Greening, 63.25pts & skater #306 Clarke MacArthur 62.87pts. I've accurately pegged the skaters, so I know these are close to correct.)
There's all kinds of numbers to tinker with.
Once the Yahoo! formula is pegged, I think it can be analyzed and adjusted for greater accuracy.
For example, Yahoo currently weights SO at approximately 4x more value than W.
However, there was actually 7x more W than SO, so a better scoring evaluator would be to use that same 7x as the multiplier for SO.
(*Note1: For actual scoring - the SO multiplier should be ~7x the W multiplier.)
(*Note2: Though, with SO being a low quantity number and highly variable... the lower multiplier is probably best for projection valuations.)