Week 17 - WARNING 1:00PM EST START

[Robin Hood]

That's because you're equating a weekly "win" to a game in the NHL.

Yup. A week is the closest analog to an NHL game in my opinion. This opinion does not seem to be shared with the majority of the BBKL. Oh well.

See it all depends on how you define "game". If you define it as a week, that is okay. But the problem is that there are just 21 games then <--- not a good emulation of the real world.

If you're a stats guy (which I think you are), a game is literally anything you define it as <-- as long as the odds in scale don't change. Maybe not the best way to phrase it so I'll try a different way:

If you take everyone's record in terms of categories. And then divide each column by 82 (W/82 - L/82 - T/82), that is just as good of a way to create a W-L-T record as a 21 week record. The parameters used to define what a game is are subject to change on whichever statistical model you use.

My record as of last week was 140-84-32 over 16 weeks. If we divide each column by 16, you get roughly 8-5-2. You could also divide by 32. Or by 64. Or whatever.

Also what determines a week? CBS defines it as Monday to Sunday. But a week in terms of stats could be split up Tuesday to Monday. Or Saturday to Friday. And how you define a week could significantly change the outcome of match-ups we use.

Heck, I'd almost support dividing out full record by 21 (W/21 - L/21 - T/21) over a straight W-L-T system. It would weed out the week by week variance.

But our category win system is still better.

[shooker]

that is how I originally read it but then reread it and was like, wtf, he agrees.

makes more sense now.

Yes. The statement is meant to be incoherent if you equate week in bbkl to game in nhl as i do.

I equate it to a 7 game series and not a single game. When two similar teams play there really are massive swings. each day can create a different matchup. It usually comes down to 4 or 5 actual days when gp are many vs the 2 or 3 days a week that are lacking with the quantity of gp. Still more of a series then a single game to me. Highs, lows and big swings reflect more of a series then a single game. If every player on played only one game each then I would be entirely on board with it being a one game comparison but they play more.

just my opinion.

[Robin Hood]

I equate it to a 7 game series and not a single game.

Exactly how I feel

[Nick]

1. It lowers variance. A Top 3 team in the league could have its stats dispersed week by week in such a way that its weekly record shows 10-6-5 while it would actually be a a Top 3 team when you count each category as a win. You want there to be more "events" so as to lower variance. E.g. flipping a coin 21 times versus could have high variance and Tails may come up 16 times. But flipping it 21 (game weeks) x 16 (categories) would likely result in a 50-50 split between heads and tails based on the sample size.

you've crossed more variance with more useful variance. By crossing levels of comparison we've made a measure that has tons of noise and wasted 'magnitude' added.

more data would be removing the h2h component, totalling each category over the years, compressing into a single score using all available data (measured in a meaningful way in this case, and valid for comparison).It would be boring.

Right now

LAK VS TOR with like 33 vs 33 points, is a tie, 1/2 category each
PHX vs TBL has me up 25 vs 22, thats 1 category for me

Our current system says, that I earned more points that either of them, because we compare category wins (not the data measured).

By repeating this 16 times, we've making a ton of false comparisons, because our standings are not *just* measured against the h2h matchup.

Our schedule is made to be as comparable as our divisional partners as possible, but that doesn't mean it suddenly makes this cross-level data more meaningful, or a better measure of variance between teams.

2. Magnitude matters. It matters how much you beat each team by. In terms of difficulty, each team has a fairly equal schedule compared to other team's in its conference. So if one team accumulates more category wins over the course of a season, it implies that said team is better than other teams who did not accumulate as many wins. This also matters for teams at the bottom. Think about it for a second. Right now DET and DAL are 2-14-0 on the season. CAR is next at 3-13-0. But CAR has 38 more points that DET. If we are going to do a lottery, the worst teams should get the highest odds at the top pick. Another example? DAL's 2-14-0 record still has it with more points than CAR with a 3-13-0 record. You will see these discrepancies everywhere.

magnitude matters [in a win] ? why? how is that an accurate comparison between teams? repeating it again certainly doesn't make it true.

Our schedules within divisions are most similar - as in we face the same teams but (obviously) during different weeks. As you always point out, teams change a lot over the season. Facing a team without a goalie gives you 6 points, these category win points are than compared to tight goalie battles (say a split) and you're saying this is a good comparison. Trying to say it's accurately measured (meaningful) variance. Your examples do not support what you are saying they do.

Typically, winning more categories will win you more weeks. However, it's all too common that Big Wins have more impact than multiple wins. Facing a team that flops, is tanking, should not have a bigger impact on your standing in the season than a normal win.

COL is 12-3-1 (w/292) and MIN is 9-6-1 (w/304), COL also beat MIN (10-6). But you are saying that Hong's 13-2-1 Win over Vancouver, 13-2-1 over PHX, when Bryma lost to SJS (7-9) and LAK (4-11-1) is a fair comparison. I'm not saying ignore the categories won, it would make sense for that to be the first tie-breaker, and/or the measure for home-ice, but to have more value on these extreme scores (that don't even accurately measure dominance, can explain that later if needed) than on actually winning the head to head match ups, is a glaring error to me.

over those 3 weeks, MIN has 66 points, COL has 43 for a 23 point spread. In a W-L-T system it would be COL 2pts, MIN 4 points, for a 2 point spread. FWIW I didn't even have to hunt to find these glaring examples. Fairly sure FLA/OTT would show the same.

To discuss 'our schedules make this comparison more fair' - point:

COL VS VAN: 7-7-2, vs: PHX 10-5-1
MIN vs LAK (1st): 5-7-2, vs SJS: 10-6-1

meaning in the 5 weeks of same opponents, COL is 2-1-2, MIN 3-2-0, COL with 5pts, MIN with 6. Our system says Hong has a 19 point lead (99 vs 80). Very easy to close a 1pt gap in a W-L-T system (in fact he does close it and would surpass with 25pts for COL, 19 for MIN) however in our system despite more wins in the other weeks, a 12 point gap still exists. we've actually said that despite losing more weeks, his wins were by a bigger margin, and therefore count for more.

Now there is a counter argument - teams that are built to win 9 of 16 categories weekly can be argued to be just as strong as a team meant to be strong in 16 of 16. Because when they face off, the 9 category team could actually win more often. But we are not trying to weigh which team would win more vs other teams 1 on 1.

Will stop there so things don't compound themselves too far into a circular argument.

We are an h2h league, that means you are being measured against ONE OTHER TEAM at a time. If we aren't doing that, our measure is even less valid.

We also have a playoff seeding based on records. A 9-6-6 team in the East vs a 9-6-6 team in the West in the final --> who gets home advantage?

that's easy to sort out, same situation could arise in our system with 400 point vs 400 points.

The last point I will make: Our scoring system makes our weekly schedule, playoffs and standings soooo entertaining. I don't think you guys appreciate how valuable that is. It keeps shit interesting for teams at the top, middle and bottom. A playoff race/division title race is far more likely with a category win system than a W-L-T system. I want to see if a team can squeeze into the playoffs in the last week. I want teams to have that chance. I don't want the 9th place team to give up because its W-L-T record isn't close even though it is close on category wins.

No. All races, presidents trophy, playoff ranking and draft picks are more live in a W-L-T system. More change is possible and likely without an error of magnitudes added. If you value 'interest in standings' that point goes to a W-L-T system.

Let's not screw up a good thing.

It's one of very few undesigned errors/fallacies that we've never fixed. I cannot wait for the season where we do.

[Nick]

it doesn't matter much.pretty sure the discussions been had and idont think itwill ever happen.

misunderstanding is best settled with ongoing conversation to remove myths and incorrect assumptions.


We say W-L-T is sufficient for playoffs and determining the 'winner', yet we (I'll say it again, by mistake) sum categories wins in the season.

The only reason this is our system, is CBS. Not designed to be this way.

[Mike]

Not designed that way no, but it is the incumbent.

[The BBKL Insider]

lol

Lee starts a shit-storm

[shooker]

on a fun not, aka actually something worth talking about. Buffalo is here. tied shiv last week and is going to lose 8-7-1 to me while putting up 30 pts (assuming nothing else changes). pretty impressive turn around happening there. Hats off to Bryuns, putting on a rebuilding clinic.

[The BBKL Insider]

matt is a GM genius

[shooker]

on a shitty note, my injured list is now Datsyuk, H Sedin, Bieksa, Spurgeon, Dupuis and Chi's number 2. things are really starting to compound.

[Robin Hood]

Huh ? so many false statements to begin it's worthless to continue.

Thank you for providing me with a reason to discontinue writing down my thoughts. Your entire post basically said "wrong because my way is better" in more ways than one without any solid proof towards any of what Shook and I said.

I was going to write a long post like I would have a few months ago when you wrote stupid shit like this but I don't have the energy or the drive to argue like this with you any more. I'll be CAM's age before I want to be haha.

If you have the votes, you'll get W-L-T. Don't think the CC ever votes it in though. Leaving this thread now.

[Lee]

on a fun not, aka actually something worth talking about. Buffalo is here. tied shiv last week and is going to lose 8-7-1 to me while putting up 30 pts (assuming nothing else changes). pretty impressive turn around happening there. Hats off to Bryuns, putting on a rebuilding clinic.

You mean it's easy to rebuild when your team has desirable assets?

Not to take anything away from what Dave has done.

[Mike]

"wrong because my way is better"

That's the entirety of this argument on both sides. It's purely subjective. I like Coke, you like Pepsi.

[shooker]

on a fun not, aka actually something worth talking about. Buffalo is here. tied shiv last week and is going to lose 8-7-1 to me while putting up 30 pts (assuming nothing else changes). pretty impressive turn around happening there. Hats off to Bryuns, putting on a rebuilding clinic.

You mean it's easy to rebuild when your team has desirable assets?

Not to take anything away from what Dave has done.

He may have had some assets but what he did with them is more then commendable. A lot of rebuilding teams have/had similar asset pools as buffalo did to start this season. I stand by my comments, I think he is doing an awesome job and making the most out of the hand that was dealt to him. He isn't going to win it all this year but is constantly improving. That is "pat on the back" material.

[Nick]

Huh ? so many false statements to begin it's worthless to continue.

Thank you for providing me with a reason to discontinue writing down my thoughts. Your entire post basically said "wrong because my way is better" in more ways than one without any solid proof towards any of what Shook and I said.

I was going to write a long post like I would have a few months ago when you wrote stupid shit like this but I don't have the energy or the drive to argue like this with you any more. I'll be CAM's age before I want to be haha.

Apologies for me taking the easy road there and not explaining why, I just spent a few minutes and attempted to explain it some more. FWIW, what I wrote is not stupid shit, it's factually accurate and would stand up the test of a proper statistical analysis. We do not have a valid measure, the error is our measuring system is creating 'standings' which do not represent our head to head league, and are doing our very in depth league a disservice.

[The BBKL Insider]

Mike Won
6-8-2 with no stat corrections overnight.

good matchup mike, not the matchup i was hoping for with our goalies shitting the bed, but a fun week.

[bma]

Thanks for the reply Shiv, your response and opinions are very much appreciated and well stated. You are one of the few in this league that will express your opinions like this in a passionate and well thought out matter. Stuff like this is what makes this league different and better than all the other fantasy leagues. I encourage everyone else to speak their minds like Shiv has.

Anyways, heres some of my rebuttal thoughts:

1. It lowers variance. A Top 3 team in the league could have its stats dispersed week by week in such a way that its weekly record shows 10-6-5 while it would actually be a a Top 3 team when you count each category as a win. You want there to be more "events" so as to lower variance. E.g. flipping a coin 21 times versus could have high variance and Tails may come up 16 times. But flipping it 21 (game weeks) x 16 (categories) would likely result in a 50-50 split between heads and tails based on the sample size.

Your variance argument is a good point. But I don't think 21 "games" is a small sample size at all. The NFL is the best ran league in the world and they play 16 games a year. Do the "best" 6 teams based on the most advanced statistics always make the playoffs in each conference? Heck no they dont. Thats the beauty of sport.

2. Magnitude matters. It matters how much you beat each team by. In terms of difficulty, each team has a fairly equal schedule compared to other team's in its conference. So if one team accumulates more category wins over the course of a season, it implies that said team is better than other teams who did not accumulate as many wins. This also matters for teams at the bottom. Think about it for a second. Right now DET and DAL are 2-14-0 on the season. CAR is next at 3-13-0. But CAR has 38 more points that DET. If we are going to do a lottery, the worst teams should get the highest odds at the top pick. Another example? DAL's 2-14-0 record still has it with more points than CAR with a 3-13-0 record. You will see these discrepancies everywhere.

Does magnitude matter in any sport thats played? A win is a win. I'm going to simplify this and equate goal differential in the NHL to what category wins and losses are in BBKL. The leafs have a -13 in goal differential yet they are 4th in the conference in points. Do you think they are actually a top 4 team based any advanced stats or talent or whatever you want to base it on? Heck no, but they are 4th right now. Again, the beauty of sport.

We also have a playoff seeding based on records. A 9-6-6 team in the East vs a 9-6-6 team in the West in the final --> who gets home advantage?

The tiebreaker would probably be categories wins and losses. Basically use what our scoring system is right now.

This argument is so annoying because it is the definition of a "let's try to fix something when there is nothing to fix" situation. The tanking issue brought up by Lee not dressing goalies is far more important.

I don't think its annoying at all. I want BBKL to succeed and thrive. Just become something isn't broken doesn't mean improvements cant be made. Continuing conversations like this will do nothing but further the success of BBKL.

[bma]

Also, I think as a group we can be a little less stubborn. Be a little bit more open to what others have to say and really take the time to appreciate each others points and arguments. You might think you have everything figured out, but you probably don't.

[The BBKL Insider]

Also, I think as a group we can be a little less stubborn. Be a little bit more open to what others have to say and really take the time to appreciate each others points and arguments. You might think you have everything figured out, but you probably don't.

solid point!

everyone has their own opinion, we should be open to the views of everyone.

[Bruyns]

on a fun not, aka actually something worth talking about. Buffalo is here. tied shiv last week and is going to lose 8-7-1 to me while putting up 30 pts (assuming nothing else changes). pretty impressive turn around happening there. Hats off to Bryuns, putting on a rebuilding clinic.

Thanks, happy to see some progress with back to back solid weeks against you and shiv. Looking at my google docs week one lineup compared to now is quite the change.