Pengwin7's 2010-2011 Goalie Rankings (Draft Version)

[Hey Robbie]

I don't think that save percentage should be penalized if the goalie plays fewer minutes. In a H2H league a .920 save percentage is just that regardless of how many minutes you play.

If the league starts more than 1 goalie then some discount is reasonable. If my #1G plays 240 minutes and faces 120 with a .900 SV% , a #2G who plays 60 minutes and pitches a 30 save shutout (1.000 SV%) gives me a .920 SV% overall, but a #2G playing 240 minutes and giving up 8 goals on 120 shots ( .950 SV%) is actually more valuable, giving a .925 SV% overall.

Still, as I mentioned in a previous post, I think the scaling is too steep as initially proposed.

[Pengwin7]

GAA & SV% has to be scaled to some degree.
You can't just pick a .920 goalie and a .900 goalie and expect a .910 average.

For example, Martin Gerber and Anton Khudobin appeared in about 3 games each this year. They both has a SV% better than .940 and a GAA better than 2.00. If I don't weight their minutes than they both get scores above perfect (10.0+) in each category.

In an "unweighted" analysis, Gerber & Khudobin end up with scores in the low 5.00s.
This makes them more valuable than #17.Hiller, Howard, Roloson, Brodeur, Neuvirth, etc.

Do you really think Gerber & Khudobin were more valuable this year than those guys?

[Rad64]

One way to handle minutes is to tier the goalies based on minutes played. That will give you a true picture of their value.

[Hey Robbie]

The ideal way to deal with this is probably to take a page from sim poolies, and write a program that takes inputs for each goalie such as % chance starts each game, % of games started letting in 0 goals, % games letting in 1, letting in 2, etc. Run H2H matchups many times, and find which goalies have the highest avg # of cats won.

Obvious downside - practicality. I don't think doing the coding would be super hard, but you'd also need to go through last year's stats and come up with the probabilities (or even more of a pain, though more useful, predict probabilities for next year.

I guess we've got a few months before keeper decisions/draft day, and I HAVE been telling myself I'm going to practice my Python, so I may at least fiddle around with the program; if I make any progress maybe I'll lean on DobberNation to fill in a spreadsheet of probabilities for the top 25 goalies.