The Ryan Suter has no value aka value for puck movers thread

[Nick]

607 players with 20 or more GP. That's a lot of copy/past + formula edit.

Mike or someone wanna remind me how to tell excell how to slide the forumla? (i only want x to slide, not mean or std dev).

[shooker]

that's fair but if you remove those players, wouldn't they drop down in percentile?

Ie 839 current players. player is top 15% aka top 126 players wouldn't he still be top 126 players but now be 126 out of 539? which now makes him top 23%?

Maybe I am missing something here but I don't really see how dropping the worst players in the league would make someone rise in percentile.

[Nick]

that's fair but if you remove those players, wouldn't they drop down in percentile?

Ie 839 current players. player is top 15% aka top 126 players wouldn't he still be top 126 players but now be 126 out of 539? which now makes him top 23%?

Maybe I am missing something here but I don't really see how dropping the worst players in the league would make someone rise in percentile.

It changes the mean and the standard deviaition, a bunch of these scrubs under 20gp might have big hits and pims but not much shtoi or points, making the 'impact' from points or shtoi larger, and changing the rank/percentile.



It's changing what we're comparing each individual player. Instead of distance from the average production of all 900 'nhlers' it's impact from the average full time NHlers.

[Robin Hood]

607 players with 20 or more GP. That's a lot of copy/past + formula edit.

Mike or someone wanna remind me how to tell excell how to slide the forumla? (i only want x to slide, not mean or std dev).

Can you be more specific? If I understand you correctly, you need to use a $ sign with the appropriate Row/Column/both.

[shooker]

that's fair but if you remove those players, wouldn't they drop down in percentile?

Ie 839 current players. player is top 15% aka top 126 players wouldn't he still be top 126 players but now be 126 out of 539? which now makes him top 23%?

Maybe I am missing something here but I don't really see how dropping the worst players in the league would make someone rise in percentile.

It changes the mean and the standard deviaition, a bunch of these scrubs under 20gp might have big hits and pims but not much shtoi or points, making the 'impact' from points or shtoi larger, and changing the rank/percentile.



It's changing what we're comparing each individual player. Instead of distance from the average production of all 900 'nhlers' it's impact from the average full time NHlers.

k that makes sense. are you going off last year or the last 82 game season. Last year would have some rather skewed values as it was such a short season. If that's all that is available fine, but otherwise use the last year because it would be a truer sense of who is valued in a full year.

This definitely isn't the cure for arguments as it is flaw for this, but its probably the best measure we have.

[Nick]

the forumla for a z-score in excell is : STANDARDIZE (x, mean, stdDev).

X changes with each player, so Backes's Goals are G7, Backstrom's are G8, however both are being compared to the same Mean and standard deviation (in this case G609 and G610).


Basically I want a list from (G3, G609, G610) to (G607, G609, G610). From their I can slide it across to all 12 skater categories (could grow it to 16 and include goalies for that 'complete league value' study, but...


I'm just using last year's 48game season, the actual number of games isn't a big deal with this many individual stats, 604 players x 14 catgeories we relaly only need like 7gp (events) to have powerful stats, of course each additional repeat makes the data more reliable, but it should not really have an impact in the outcome.

[Robin Hood]

the forumla for a z-score in excell is : STANDARDIZE (x, mean, stdDev).

X changes with each player, so Backes's Goals are G7, Backstrom's are G8, however both are being compared to the same Mean and standard deviation (in this case G609 and G610).


Basically I want a list from (G3, G609, G610) to (G607, G609, G610). From their I can slide it across to all 12 skater categories (could grow it to 16 and include goalies for that 'complete league value' study, but...


I'm just using last year's 48game season, the actual number of games isn't a big deal with this many individual stats, 604 players x 14 catgeories we relaly only need like 7gp (events) to have powerful stats, of course each additional repeat makes the data more reliable, but it should not really have an impact in the outcome.

Easy - change it to (G3, $G$609, $G$610) in the formula I mean. The $ signs hold the cell location to a specific cell.

[Nick]

thx,

[Robin Hood]

That list is great Nick - it only confirms what Shook and I have been saying:

1. There are several finesse players on that list whose BBKL value does not correspond with their actual value due to their cap hits
2. Many top d-men are on that list but their values are lower due to cap again. Eg Yandle

[shooker]

For me this not only reassured my opinion but makes me want to go back and find a solution to the Dman thing. OEL at 100! Keith at 93! By bbkl value Keith was worse then cody freaking hodgson last year.. That to me is the bigger issue then finesse overall.

[Nick]

It puts dmen into the top 10, but it puts too many dmen too high, ...luke schenn in the top 100?

Subban above Crosby, Stamkos & Giroux ? MSL even lower on the list...

[shooker]

My argument to that isn't who is on the list but it went from 19/100 to 34/100. You are telling me that there are only 19 dmen in the top 100? I would say there is more than 34 dmen on that list. I agree it isn't perfect but Subban last year was lights out, don't forget that. AGAIN I REPEAT MY INTENT WAS TO CHANGE DMEN VALUE, DEF WAS JUST ONE SUGGESTION. I never said we had to use that. It was just the best solution I cam up with at the time.

I would be content with PPTOI as it would at least help a bit if you found DEF was too extreme as well.

I'm not out to prove DEF's worth, my objective was and is only to prove a flaw in our scoring. I don't see the need in continuing to try to discredit that single stat when my goal isn't to get it put into effect but address and issue I see. I also don't see that list discrediting it but supporting it. Just saying

[Nick]

I didn't try to discredit it? I showed what change it would have on the impact these players are having. 1dmen in top 10 probably isn't enough... but 40 in the top 100 seems like a lot.

I can see if PPTOI would have a better impact. This is how we can actually tell when these stats do to the impact on our league.

[shooker]

I didn't try to discredit it? I showed what change it would have on the impact these players are having. 1dmen in top 10 probably isn't enough... but 40 in the top 100 seems like a lot.

I can see if PPTOI would have a better impact. This is how we can actually tell when these stats do to the impact on our league.

it just felt like you were as I haven't brought up that stat in a while, I just reiterated my feelings towards dmen being undervalued. I do not feel that 34/100 is too high of a number as far as dmen are concerned. Dmen make up 40% of the skaters on a teams top 5 player unit that can be on the ice at each point in time, so 34% of the top 100 seems pretty appropriate to me. It might be a smidge high for some, but I view it as pretty accurate.

[Robin Hood]

Great stuff again Nick - your Excel work has produced a mathematical reason to look into this. Try doing it with PPTOI and let's see what kind of list we get.

[The BBKL Insider]

hey look paul, there's brent burns

[shooker]

We cant do estoi no? goalies!

[shooker]

DEF + PPTOI is definitely something none of us would have pushed for, its going way too far. It is the further away from balance then we currently are.

[shooker]

interesting to see that list though LOL

[Nick]

It's not biggest impact list, it's who is impact by the stat change. list is 3 posts up or prev page