By Vahagn Go To PostPer my count, so far 22 teams have won or lost more than 1 game over and above the Projected wins prediction. That number may shrink tonight or it may go up. Lots have teams have had a +/- differential of 3 of more.
And that's based on data leading up to and from the penultimate night of the regular season. I imagine the end of season projected wins from data 40 games ago included variances that were far higher because of smaller sample size issues.
1 game isn't the cutoff. Learn was root mean-square error is first. There are no season projections from 40 games ago involved, not sure why that was brought up. That's not how this works.
Is any case, my point is, if you're going to put all your faith on the predictive capabilities of a stat, don't claim luck when you're wrong. These aren't coin flips. Which we know will land in 50/50 splits given a large enough sample size.
If a team outperforms your projection, own up and try to assess why. Or ignore the topic. Calling it luck is lazy.
It's not a predictive stat in the way you're arguing. It is saying a team with this point differential generally wins X games. That's the true quality of the team given the players that played and when. I hope to god we all agree point differential is a better indicator of team strength than win/loss record so if you accept that, then you agree with my point. The Houston Rockets have played as good as a 50 win team but they just happen to have 55 wins.
You say these aren't coin flips because we know it's 50/50 with a large enough sample size. But we do have a large enough sample size already (thousands and thousands of games) and have found that a team with a PD of X wins Y games over a large enough sample size.
Second, and something that really demonstrates you don't know what you're talking about, you don't have to assess why a team significantly outperforms their Expected W/L and adjust your model. Why? Because every model must have instances of teams significantly outperforming or underperforming the projection. It is mathematically impossible for it not to happen. And I don't mean in basketball, I mean in every model in existence. If it happens a lot, then you have to adjust, but if it is a rare occurance, then you don't. In fact, every season, there should be 1-2 teams either significantly outperforming or underperforming Expected W/L. If that wasn't happening, something very wrong would be going on. This is how math works.
So, every year there should be about 1-2 teams that are either unlucky or lucky. This year, the Rockets are one of those teams.
http://www.insidethebook.com/ee/index.php/site/comments/true_talent_levels_for_sports_leagues/#2
Read the article first, then the comment, which I quote:
what the heck… NBA var(obs) is around .145^2. var(rand) just like hockey, or .055^2. So var(true) = .134^2To get an r of .5, you need only 14 NBA games! Sheesh. This is a huge problem here. 14 NBA games tells you as much as 36 NHL games.On top of which, 16 NBA teams make the playoffs.NBA games need to be cut down from 48 minutes to something alot less.
SRS is usually a very good predictor in the NBA. If you can figure out how to improve on this substantially over publicly available statistics (which adjust for strength of schedule, for example), you have plenty of options available to you: publish a paper, sell your statistic to an NBA team, make a lot of money betting on basketball, gloat about it on SLAENT... but if you can't, saying "it wasn't luck!" really doesn't add anything of value to the conversation. Because you have no idea that it wasn't luck.
By Dark PhaZe Go To PostAnother part of management that I think is underrated is just knowing when to fucking let go. The Wizards have had projects that we damn well knew weren't going to work out, but we let them play through their contracts. Why? I get the argument of, ok, lets see if he improves, but it's the laziest thing to always just wait for a guy to just walk out of his contract. That time is valuable. Those minutes can go to another guy you want to take a crack it in your system.
For example Kevin Seraphin almost certainly won't be back. For chunks of the season he was getting almost all of the backup C minutes. What in the fuck is the point? Waive him. See if you can package him in a deal. Bring in a couple of interesting D-League prospects and see if you can luck into the bench tiered version of Whiteside.
I get certain GMs being really hands-off in regards to coaching and rotations but I agree. Gary Neal was shooting something like 22% and calling ISOs for himself before we got Mo Williams. And on top of that.... it should've been obvious to everyone that Mo would be our backup (or starter) - but instead this coach decides that Mo is better off playing shooting guard next to Kemba who has no clue how to play off-ball.
Like... that's when you need to go down to the coach's office and throw hands.
By IWMTB19 Go To Post The Rockets have either been winning a bunch of close games randomly or Harden has transformed into 2011 Dirk/2013 Joe Johnson this year in crunch time and I don't think I buy the second one.
He's been that good in crunch time for the second straight season now.
By thekad Go To PostHe's been that good in crunch time for the second straight season now.
Except in the 2014 Playoffs.
@HornetsPR
The @hornets 5 injured players: Hairston, Jefferson, MKG, Stephenson & Zeller all remain out for tonight's game. #CHAvsTOR
LFK pls respond
(real shit that lineup is better than most of the rotations we ran during the season LOL)
Steph Curry has a 64TS%.
Steph Curry is shooting 44% on threes
Steph Curry takes 8 threes per game
Steph Curry is shooting 91% on free throws
Steph Curry has broken his own single season record from Three pointers made in less minutes.
Steph Curry is shooting 53% on two pointers
Steph Curry is shooting 52% on threes post ASB
Steph Curry has missed 6 free throws in the last two months
I know he's not at the 50/40/90 mark but considering the amount of threes he's taking, this has to be one of the greatest shooting season ever correct?
By thekad Go To PostHe's been that good in crunch time for the second straight season now.
Unlike IWMTB, I actually do buy that there might be players who were better in crunch time (because I suspect final shots in close games have a different distribution than in other games, e.g. contested twos may be more frequent, fouls less likely to be called; so players good at shooting those shots should be more clutch). But I don't know if there's any evidence for that, and Harden really doesn't fit the profile I'd expect (only 38% on 2s out of the paint this season).
By reilo Go To PostExcept in the 2014 Playoffs.
You could have at least pretended to research this before posting.
By Sharp Go To PostUnlike IWMTB, I actually do buy that there might be players who were better in crunch time (because I suspect final shots in close games have a different distribution than in other games, e.g. contested twos may be more frequent, fouls less likely to be called; so players good at shooting those shots should be more clutch). But I don't know if there's any evidence for that, and Harden really doesn't fit the profile I'd expect (only 38% on 2s out of the paint this season).
I'd say there must be players that are chokers or anti-clutch. Players who perform worse at the end of the game than we'd expect. There just has to be a few mentally. But I highly doubt there are players who perform better than we'd expect. To me, clutch would then be players who are not anti-clutch. That is, they perform about how we would expect (this doesn't mean shoot the same percentages as they do they rest of the game because defenses play harder, refs might hold their whistle, etc).
By Sharp Go To PostUnlike IWMTB, I actually do buy that there might be players who were better in crunch time (because I suspect final shots in close games have a different distribution than in other games, e.g. contested twos may be more frequent, fouls less likely to be called; so players good at shooting those shots should be more clutch). But I don't know if there's any evidence for that, and Harden really doesn't fit the profile I'd expect (only 38% on 2s out of the paint this season).
The biggest attribute in those situations (which are pretty much always ISO or pick and roll) is being able to break down a defender one on one. Harden might be the best at that.
By thekad Go To PostYou could have at least pretended to research this before posting.
im gonna bet he watched all of those games.
By diehard Go To Postim gonna bet he watched all of those games.
A year ago. I'm sure he remembers the last five minutes of 4th/OT in all six games vividly and without bias.
Harden's Clutch splits the past 3 years:
.486 FG% -- .522 3P% -- .569 eFG% (2015)
.438 FG% -- .458 3P% -- .506 eFG% (2014)
.402 FG% -- .280 3P% -- .445 eFG% (2013)
By thekad Go To PostYou could have at least pretended to research this before posting.
Oh, BUT I DID!
http://i.imgur.com/lc4ItaB.png"> #ClutchCity
By rodeoclown Go To Post The Timberwolves have 10 guys on the injury report and are playing only 6 or 7. This is the 2015 equivalent of Mark Madsen shooting 10 3s against the Grizzlies.
shameless tanking.
By unknown Go To Postshameless tanking.
brev
By unknown Go To Postshameless tanking.
For the last month. But Pekovic and Rubio did legitimately need surgery. Other guys (KG, Gary Neal, Kevin Martin's hamstring), well, I'm less convinced.
More importantly, Mike Malone met with Glen Taylor recently. Maybe we have a good coach to go along with Wiggins, Rubio and a high draft pick. I can dream.
By Sharp Go To Posthttp://www.insidethebook.com/ee/index.php/site/comments/true_talent_levels_for_sports_leagues/#2" target="_blank">http://www.insidethebook.com/ee/index.php/site/comments/true_talent_levels_for_sports_leagues/#2
Read the article first, then the comment, which I quote:
what the heck… NBA var(obs) is around .145^2. var(rand) just like hockey, or .055^2. So var(true) = .134^2To get an r of .5, you need only 14 NBA games!
Wow...somebody has finally managed to type an article in wingdings.
By DY_nasty Go To PostLFK pls respond
(real shit that lineup is better than most of the rotations we ran during the season LOL)
About time Raps got a break!
If the Jazz beat the Rockets tonight, I think that the Rockets should just concede the first round and go fishing early. In a must win game, surely there's no chance that they botch this one, right?
Also, be honest, does it look like I'm farting in my avatar? Because that's not a good look.
By psychintellect Go To Post So, even if we win, if Brooklyn beats Orlando, we get the pick. Don't fuck this up Brooklyn.
Eh no promises I'm not sure the players even want the playoffs.
By Sharp Go To Posthttp://www.insidethebook.com/ee/index.php/site/comments/true_talent_levels_for_sports_leagues/#2" target="_blank">http://www.insidethebook.com/ee/index.php/site/comments/true_talent_levels_for_sports_leagues/#2
Read the article first, then the comment, which I quote:
SRS is usually a very good predictor in the NBA. If you can figure out how to improve on this substantially over publicly available statistics (which adjust for strength of schedule, for example), you have plenty of options available to you: publish a paper, sell your statistic to an NBA team, make a lot of money betting on basketball, gloat about it on SLAENT... but if you can't, saying "it wasn't luck!" really doesn't add anything of value to the conversation. Because you have no idea that it wasn't luck.
i get the post and the comment but why is this a huge problem like he said?
Guyz, are you ready for the most awesome tank off of all time tonight?
Loser literally gets a draft pick.
By vire Go To PostGuyz, are you ready for the most awesome tank off of all time tonight?
Loser literally gets a draft pick.
Neither player will jump at the opening tip.
By Mamba Go To Post1 game isn't the cutoff. Learn was root mean-square error is first. There are no season projections from 40 games ago involved, not sure why that was brought up. That's not how this works.
It's not a predictive stat in the way you're arguing. It is saying a team with this point differential generally wins X games. That's the true quality of the team given the players that played and when. I hope to god we all agree point differential is a better indicator of team strength than win/loss record so if you accept that, then you agree with my point. The Houston Rockets have played as good as a 50 win team but they just happen to have 55 wins.
You say these aren't coin flips because we know it's 50/50 with a large enough sample size. But we do have a large enough sample size already (thousands and thousands of games) and have found that a team with a PD of X wins Y games over a large enough sample size.
Second, and something that really demonstrates you don't know what you're talking about, you don't have to assess why a team significantly outperforms their Expected W/L and adjust your model. Why? Because every model must have instances of teams significantly outperforming or underperforming the projection. It is mathematically impossible for it not to happen. And I don't mean in basketball, I mean in every model in existence. If it happens a lot, then you have to adjust, but if it is a rare occurance, then you don't. In fact, every season, there should be 1-2 teams either significantly outperforming or underperforming Expected W/L. If that wasn't happening, something very wrong would be going on. This is how math works.
So, every year there should be about 1-2 teams that are either unlucky or lucky. This year, the Rockets are one of those teams.
I told you there are 22 teams that are off by more than 1 win or loss from the projected win total. Which means I wasn't measuring it exactly, I was giving the model the benefit of a doubt. A Margin of error.
Youre saying a model that has a large variance from reality should be adjusted, but 22 teams out of 30 isn't large enough for you. We'll check tomorrow after all the games have been played, and my hunch is that fewer than 20% will be spot on, and about another 5-10 will have a variance of at least 3 games.
This isnt 1 or 2 territory.
your Vegas example is a problem. Unless we're discussing roulette (you didn't specify the game) there are a variety of games where skill level comes into play. Are you better than everyone else at the poker table, can you count cards in blackjack and stay disciplined to your betting ratios.
I'm telling you that Bill won money because of X. (Skill, tactics, luck) and you're telling me you don't need to know whether he's actually good or not to determine it being lucky. ANYTHING that falls outside of the average is, and can only be, luck.
i.e - I'm telling you to understand the substance of the game you're talking about. Why teams do better than expected or worse than expected. you're Telling me it's unnecessary once you have a formula.
Basically you're being lazy. Chalk it up to luck and move on.
By DY_nasty Go To Post Hubie sounds like he's in pain calling this game
Hubie is 124 years old, he's always in pain.
By Diprosalic Go To Posti get the post and the comment but why is this a huge problem like he said?
Basically, because it means the season is way too long, in the sense that most of the time you know how good a team is long before game 82. This year isn't really an exception--there's a lot of manuvering around in the rankings in the West, but only because all the teams are so close in record; basically all of the teams going to the playoffs in both the West and the East have been known for ages, except for maybe the 7 and 8 positions.
Personally, I'd be completely in favor of a shorter season, dissolving divisions, and a round robin playoffs format with only eight playoff teams and five game series. NBA wins with more playoff revenue, crap teams that don't have a chance of winning don't make the playoffs, all the matchups are good, regular season is more exciting, fewer player injuries, and best of all, makes it harder to do dumb counting stat comparisons between eras.
By matrix Go To Post When the fuck did this place become the Clutchfans forum?
Over my dead body.
By KumaJG Go To Postthere isn't a hornets fan left that wants to watch this gameWho idea was it to put Hornets on TV instead of the Pelicans
nice to see kemba gets out of bed to play on national tv. maybe he should go to philly or brooklyn
By Sharp Go To PostBasically, because it means the season is way too long, in the sense that most of the time you know how good a team is long before game 82. This year isn't really an exception--there's a lot of manuvering around in the rankings in the West, but only because all the teams are so close in record; basically all of the teams going to the playoffs in both the West and the East have been known for ages, except for maybe the 7 and 8 positions.
Personally, I'd be completely in favor of a shorter season, dissolving divisions, and a round robin playoffs format with only eight playoff teams and five game series. NBA wins with more playoff revenue, crap teams that don't have a chance of winning don't make the playoffs, all the matchups are good, regular season is more exciting, fewer player injuries, and best of all, makes it harder to do dumb counting stat comparisons between eras.
ah yes that makes sense, i thought it might be something different.
i'm also in favour of a shorter season after 50-ish games i have seen enough.
By knux-future Go To PostHeat holding out Dragic, Wade and Whiteside.
this game will be special
Chalmers unleashed! Free Chalmers!
By Moris Go To Post Knicks looking for that W tonight.
I can so see it. It brings back so many painful memories of the Bryan Colangelo Raptors "building a winning culture" all the way to the 8th pick year after year.