This proposal uses the existing Goalie calculation but instead of using the DC to determine the base GAA the bandwidths determine the base GAA. Here are the proposed bandwidths.
The base GAA is determined by the previous season overall GAA.Thus
4.00 - 4.4999 use 3.00
4.50 - 4.9999 use 3.25
5.00 – 5.4999 use 3.50;
5.50 – 5.9999 use 3.75;
6.00 – 6.4999 use 4.00;
6.50 - 6.9999 use 4.25;
Subscribe to:
Post Comments (Atom)
11 comments:
Before we vote on this proposal (which I at least in concept fully endorse) I want to see how it actually plays out statistically.
For example, this system locks goaltenders into a certain range of points dependent on the number of goals scored in a season - BUT it may be locking them in at a level too high.
Given that I don't endorse using 3.75 for the reason that it treats goaltenders as roughly 20% more valuable than even #1 centers, I would object if all this system did was lock goaltenders into that position permanently.
You would have solved the issue of the DC's involvement by making their mistake permanent - which is hardly an adequate solution IMO.
Just for clarification, I beleive Rob means GPG, or the total goals scored per game, should determine the baseline GAA for the goaltending formula.
As for Cam's concern, this system is very, very close to saying that it should be 3.50 so it is not surprising that he thinks 3.75 is too high - it almost is.
At 5.54 GPG, which is what the NHL was at last season you are only 4/100 of a goal per game over the line. Had there been 50 less goals scored in the 1230 total NHL games last season, the baseline would have been 3.5 under this system so clearly even the proposal acknowledges that we are very close to the edge of an adjustment - a problem we will always get whenever we are near a clear line but not yet over it. If next year the NHL scores 550 more goals next season (driving down GAA and increasing the pts of forwards with more goals and assists) it still wouldn't be enough to flip us back up to the 4.0 baseline. The dial is so close to the 3.5 boarder right now that even an offensive outburst of that magnitutde wouldn't be enough to tip it the other way.
If we are retaining the current formula system (as opposed to going with Mike and Richard's W/OTL/SO system) count me in favour.
With respect to the concerns noted:
The system I proposed is similar in concept, but based on actual GAA for goalies that are meaningful in our pool (say the top 24-36).
Using league avg GPG may only approximate changes in GAA in that group and may statistically lead us astray.
If the concern is how high goalies might score depending on the sliding scale, the system I proposed actually calculates in reverse - that is, we set what we want goalies to average FIRST, then adjust the calculation to give s the desired result based on the most recent year.
That way if we want to set a system whereby (if we applied the scale retroactively to the prior year) the average playing goalie (ie top 12) or average drafted goalie (ie top 24) would have received say 90 points, we can fix the formula to yield exactly that.
In my mind this may yield some odd numbers (like 3.62) in the formula, but will lend much greater stability to goalie scoring from year to year and not be at the mercy of variations in league-wide GPG (which clearly are not reflective of GAA, else the bandwidths (0.5) and the Goalie calculation base increments (0.25) could be the same which (as per this proposal) they aren't)
Dan,
I'm usually able to follow but I'm a little lost here. Will I know what the baseline is before the start of the season for that season or will it only be set at the end of the season?
Hi Doug et al,
Sorry, it's one of those great FuNHL traditions - complex as hell to explain, but actually fairly simple in execution.
I'll use our current situation as an example for timing.
The stats we would use would be 07-08 season, to determine the formula for 08-09. Each year the formula would be updated so the stat would be tailored to reflect last season's performance (same as the GPG sliding scale, where the formula for this year would be based on stats from last year)
The number of goalies used is slightly arbitrary and open to discussion as to what is most appropriate.
So to illustrate:
Let's say we want to base the stat on performance of draft-worthy goalies. Since we are working in hindsight, we can use the stats for the top 24 pointing goalies from last year (assuming they all played a decent # of games which they must have to be in the top 24 for last year).
Now we decide in advance how much an average drafted (ie top 24) goalie should get for points. For argument let's say we feel the AVERAGE drafted goalie (active and benched) should earn 75 points.
We look at the stats for top 24 goalies last year to see the avg GAA, and the avg games played, and the following formula:
Goalie points (avg) = 75 (our choice)
so
75 = (X-avgGAA) times avgGP
If the average GP last year by a top 24 goalie was 60 games, then we have:
75 = (X-avgGAA) times 60
so
1.25 = X-avgGAA
The rest is easy - if avg GAA last year for top 24 goalies was 2.60, plug that in and we get
1.25 = X - 2.60
X = 3.85
And for 08-09 our goalie formula would be:
Goalie points = (3.85-GA) x GP
If we apply the formula retrospectively to last year's stats, the average top 24 goalie will come out at 75 points.
So the formula is updated to reflect last year's ACTUAL goalie points and performance, and is PRESET to keep goalie points in the range we select.
It can work for whatever avg goalie points we pick - 75, 85, 100 - we can control exactly where goalies score with no risk that a "related" but not necessarily reflective stat (like league avg GPG) could skew the goalie stats drastically smply because league avg GPG changes may or may not be reflected in the GAA of the top 24 goalies (Maybe it's all those rotten goalies at the bottom letting in all the extra goals:-)
Point is, why use an approximation when we can use the exact number we are trying to adjust?
Clear as mud?
I should add that all this can be calculated as soon as the regular season is done, so all GMs would know the goalie stat formula for the coming year right away, no fussing with DCs or votes close to the ED!
The only problem I see with your proposal Dan is that we might disagree as to what a goaltender (top 24 or top 12) should average.
As it stands I like using the metric of comparing the average top 12 centers to the top 12 goaltenders - as IMO a #1 pivot and a #1 goaltender are of relatively equal value (though I should note that the NHL Entry Draft does not agree, as for a variety of reasons goaltenders aren't selected 1st overall with near the frequency of #1 centers - but I digress).
What it all boils down to is that we need to arrive at a common notion of what the average starting goaltender should produce.
Thanks for the clarification Dan.
Looks interesting but I will have to say it is comes across as being really complicated (perhaps unfairly, but if I'm only half following it and I'm doing the stats - it's not super simple).
I will have to vote no on Dan's proposal.
I fail to see where the 1.25 = X-avgGAA comes from. Out of a hat?
I do not like pre-determining player values, or what we think they should be.
A simple formula is still needed.
A must vote no to this proposal.
I think at the heart of this is the lack of agreement (or even solid opinion) on what a goalie SHOULD score.
Cam - the nice thing about your point is if we want to tailor the formula each year such that the top 12 goalies earn the same average points as the top 12 centres, we can do exactly that - in other words if we think those two should always be on par, we can set the formula that way all the time!
Mike - the 1.25 doesn't come out of a hat - it is simply the figure IF the average goalie plays 60 games and IF we think the average goalie should score 75 points. Strictly an example.
The dilemma is that goalie stats are calculated (unlike a skater's goals and assists) so no matter what formula we choose, we are determining roughly what a goalie should score in our pool because it is a figured stat.
We still need to know roughly what we think a goalie should earn in order to know whether a formula makes sense...
Dan said: "We still need to know roughly what we think a goalie should earn in order to know whether a formula makes sense..."
I agree that this is in fact the core issue of what we want to determine.
That said, I have suggested that whatever system we utilize the stats for the top 12 goaltenders should be roughly similar to the average stat for the top 12 centers.
One might argue that this over-values goaltenders vs reality, but I can't see how it would make sense to argue that this undervalues them.
Keep in mind I am not talking about the center position, or the active center position, just the top 12 centers - which represents what is in reality less than half of what would be considered the #1 centers in the NHL.
Given that centers are the top scoring position in the NHL (though Ovechkin is going to skew the LW upwards for the next decade I don't see him pushing it up to be par with Centers), it makes sense to me to consider the goaltenders in a similar light.
As such systems and formulaes that make the average goaltender 15-30% more valuable are in my mind suspect/flawed, and the closer we get to the average top 12 pivots the more comfortable I am with how the goalies 'look'.
That said, using the 3.5 calculation is my first preference, because while it moves the average slightly below 1st line pivots, it does so for an NHL that routinely sees a slide in goalscoring, and aside from one brief shining year post-lockout, seems to have no idea how to stop it - and because doing so is reflective of how we always have done things.
But something like the 'bandwidth' proposal intrigues me precisely because it gives us a non-controversial result (more or less) grounded in empirical data.
All in all, I reject the Mike proposal (completely reject it - whatever it is, it simply isn't reflective of goaltending), am curious about Dan's proposal and endorse the bandwidth proposal.
If I might corrupt the words of Baltar for a moment 'The pool is perfect just as it is. God loves the pool exactly as it is.'
As such, please consider my vote for all other changes to be 'no'.
Post a Comment