## Re: [hockeydisk] Digest Number 16

Expand Messages
• On 30 May 2000 10:55:38 -0000 hockeydisk@egroups.com wrote: Bits and snips taken from today... My thoughs. ... True... And to make sure we got rid of
Message 1 of 1 , May 30, 2000
On 30 May 2000 10:55:38 -0000
hockeydisk@egroups.com wrote:

Bits and snips taken from today... My thoughs.

> deleted by me without further reading. If you expect some level of
> realism in your league, you MUST limit GP.

True... And to make sure we got rid of statistical deficienties, we even
limited the number of games a player must have to play. Any player that
doesn't have 20 games played doesn't play in our league. That gets rid
of goaltenders who played 4-6 games against bad teams and end up with
above .900 average. It's also important to note that we have a twelve
team league, and so it means that every team can have to 2 normally
regular goalie on hand. At the other hand of the spectrum, if a guy
played 50gms (40gms for a goalie), then he is a regular and will play as
many games as the computer will let him (we play with injuries).

So maybe it's one way to tweek the goaltenders for people who really
wanted to.

Obviously, we are still all out to find these players that can make a
difference for the 20+ games they played. 2 years ago when Fedorov was
on strike, I was really happy when he finished the season with 21gms. it
meant I could play him for 21gms and it meant I was in the playoff right
up until the last month, even though everybody didn't see me doing them

If you really want to tweek the stats, then how do you go about it?

- make an average for "groups" of team, and then pro-rate sav%
accordingly?
- take an average for goalies that have played against all team for each
"groups" and then re-calculate the sav% accordingly?
- lower these goalies a couple of sav% points to make them less
desirable (from .920 to .900...)

It also brings the question: if we lower the guy that only played
against the worst teams, why don't we up the goalie who didn't play
against these same teams?

> But take a look at his breakdowns:
>
> 1) While there is definitely a huge split in Billington's
> numbers, it doesn't appear to mean anything. He played a
> lot more time against quality opponents than I would have
> thought. The actual breakdown was 240 minutes (.893) vs.
> quality opponents, 110 minutes (.756) vs. middle opponents
> and 264 minutes (.966) vs. the bottom of the food chain.
> That's a fairly balanced distribution of playing time.

Yeah, but can a game against the Islanders be counteracted by a game
against Dallas? Now, that's something to think about...

> 2) On the other hand, Billington's stats are hardly gaudy,
> just a little above league average. He's not really in the
> data group we are targeting for adjustments. In fact, there
> are really only 4 guys who would cause concern: Manny
> Fernandez (.920 in 24 games as Belfour's back-up), Marc
> Denis (.917 in 23 games), J.S. Giguere (.914 in 7 games) and
> Stephen Valiquette (.949 in 6 games for the Islanders). And
> even in that group, Denis and Giguere look like they can
> legitimately maintain that level of play over the course of
> a season.

Since Giguere and Valiquette would be out with our system, I curious
about Denis and Fernandez. Also, doesn't it depends on which team's
backup they are? Can't we assume that Denis and Fernandez would get a
hell of a better support from their teamates? Doesn't that account for
something also... And how can this be calculated?

> Most of the leagues I see have a min game requirement of 12 NHL games
> to have been played for them to be used in your APBA season. So a

Now that's interesting, we choose 20 gms, and I forget why we chose that
number. Here we have 12 gms as a limit. Obviously, higher the limit,
more significant the stats become, but shallower is the pool of players...
Justin, how many teams in your league is there?

Well, that was my .02\$, your comments brought theses questions to my
mind. Curious to see where it goes from that.
Your message has been successfully submitted and would be delivered to recipients shortly.