Portland at .500?
Another thing from the Simmons NBA preview. He predicts Portland will go 41-41, citing the theory that every year there’s a “You Thought We’d Take Another Leap, But Instead We Went Backward Because Expectations Were Too High, We Tinkered With Our Chemistry And Our Young Guys Tuned Out Their Coach.”
Maybe. But not only did last season’s Blazers go 54 – 28, they had a point differential that would have predicted 58 wins. And point differential is a better predictor of future performance than win-loss record. To merely go 54-28 again would require the team to regress somewhat. To win 41 games would involve a regression as big as the step forward that would be require to win 75 games and become the greatest team of all time. And that would seem like a strange thing to happen to a young team that basically lost nothing of importance during the offseason.
23 Responses to “Portland at .500?”
Register
RSSLogin
Topic Cloud
ArchivesBlog RollAbout Matt YglesiasContact Me
Donate
October 24th, 2009 at 4:06 pm
Bill Simmons can sometimes be a funny writer, but any predictions he makes should be taken with all the salt in the ocean. In any sport.
October 24th, 2009 at 4:44 pm
Simmons has also decided to become anti-bandwagon with the Blazers. I’m not even sure he’s aware of it, but I think he’s not reacting to the team so much as so many other people predicting they are the good up-and-coming team. It’s more of a I-liked-Nirvana-before-they-were-mainstream-now-they-suck kind of thing.
So after talking them up and loving them a year, year and half ago, he’s moved on, trying to pick the new ‘best young team’ just before it becomes consensus. Hence his crush on the Thunder, who he predicts will make the playoffs, but who are more likely to win 30 games than 45.
October 24th, 2009 at 4:52 pm
To win 41 games would involve a regression as big as the step forward that would be require to win 75 games and become the greatest team of all time
Is this argument based on the idea that 58-41=17 and 58+17=75? I know nothing about basketball, but I would guess this is a seriously flawed argument. Let’s say the unconditional historical distribution of NBA season wins is approximately Normal. Then the probability of drawing 75 from that distribution is substantially lower than the probability of drawing 41 conditional on the last draw=58 (no matter what the serial correlation). So to go from 58 to 75 would be an outlier much more than 58 to 41 would be.
October 24th, 2009 at 4:53 pm
I don’t know anything about basketball, but one part of Matt’s analysis is wrong (or at least misleading). The marginal difficulty of improving increases dramatically as the number wins increases. This means that it is far harder to improve from a 70 to 71 win team than from a 41 to 42 win team. Therefore, it is far (far) easier to imagine a team underperforming by 15 wins than overperforming by 15 wins. They are not equal.
October 24th, 2009 at 4:54 pm
People whose only knowledge of basketball is stats should go talk about baseball.
October 24th, 2009 at 4:57 pm
Love to. But that’s over at 538.
Besides, the logic holds. It’s hard to go from the best team in the game to the best team ever. It’s far easier to go from .500 to a few games above .500. This holds true for every sport.
October 24th, 2009 at 5:27 pm
Marshall and wrigleyviller are correct. Matts assertion that “To win 41 games would involve a regression as big as the step forward that would be require to win 75 games” is fatally flawed.
Ironically, someone who mentions point differential as an accurate assessment of win loss record should understand how bell curves work. In order for an expected win% of 100%, for example, the team would need either 0 points allowed or an infinite number of scored points. For non infinite numbers, you can use an estimated 81.5 wins (99.4%) instead of 100%. The formula is at http://www.rawbw.com/~deano/articles/basics.html for anyone who is interested.
October 24th, 2009 at 5:33 pm
To win 41 games would involve a regression as big as the step forward that would be require to win 75 games and become the greatest team of all time
Perhaps you can see the flaw in your reasoning via this extrapolation: if +17 has the same likelihood as -17, then +25 has the same likelihood as -25. This means their chance of winning 83 games is the same as of winning 33 games.
October 24th, 2009 at 6:07 pm
Any predictions in any sport before the season starts are meaningless because injuries determine all in team sports. Thus, any prediction as to a teams’ predicted performance must assume no injuries to key players. The Wizards could only hope for such an occurrence. Don’t believe it, just ask Celtics fans about their elimination in the conference finals last year with Kevin Garnett among the missing.
October 24th, 2009 at 6:39 pm
Ok, for some stupid reason I decided to do the math. Assuming a teams score an average of 100 points a game, I calculated the following point differentials:
wins = 40: Mult = 1.00296, diff = -0.148
wins = 41: Mult = 1.00000, diff = 0.000
wins = 42: Mult = 0.99705, diff = 0.148
wins = 1: Mult = 1.30517, diff = -13.238
wins = 2: Mult = 1.25053, diff = -11.132
wins = 80: Mult = 0.79966, diff = 11.132
wins = 81: Mult = 0.76619, diff = 13.238
So roughly 50% teams only need to improve their differential by about 0.15 points per game for each additional win. At the extremes, however, you need to improve by over 2 points per game for each additional win.
October 24th, 2009 at 6:47 pm
the great thing about sports is that the games actually have to be played. after real things happen, stat guys can say the stats caused the results. they may very likely be wrong, but we can ignore them and still be impressed by their math skills and ingenuity. when they say before the games are played, that such and such will happen because of stats, stats which they make stand for the different players on this year’s team or the same players in different physical and mental circumstances, we can ignore them and disdain them and just see what happens in the real-time action.
October 24th, 2009 at 10:04 pm
I’m not really into sports, but Matt’s suggestion (whether flawed or not) poses an interesting question: Assuming teams generally speaking aren’t trying to lose, does that mean significant drops in win percentage are as unlikely as significant increases in win percentage (for average teams)? In principle it’s as unlikely to become one of the worst teams of all time as one of the best, right?
October 24th, 2009 at 11:29 pm
I don’t think it’s harder to win victory number 2, 22, or 72. A team has X% to win each night with some variability for other conditions like travel, injuries, and schedule so I don’t see how the distrubution of records makes things harder or easier.
October 25th, 2009 at 12:00 am
JoeJoeJoe,
You misunderstand. I’m not saying the 1 game is hard to win. I feel very confident that exactly 50% of all first games are won! (Assuming both teams are playing their first game, of course, for the number to be exact.)
I’m saying it is much harder to improve your record from 70 wins to 71 wins then it is to improve your record from 40 wins to 41 wins. It all comes down to percentages and errors. To win all 82 games, for example, you have to be perfect. You have to have enough margin of error to survive the freak game where the opposing team hits all of their three pointers and yours aren’t going in.
Statistically, the margin needed to win 82 games is infinite. However, the margin to win 81.5 games is 15.3, so margins 16 and higher can be thought of as nearly perfect.
October 25th, 2009 at 12:30 am
The game does have to be played, but that doesn’t mean that stats can’t illuminate some of it’s nuances for us.
I’d also mention that the long-haul NBA season creates a situation where a team’s record is as much about it’s personality or culture as it is about the talent level. At some point, of course, a dearth of talent creates a ceiling on win potential, but a bad team that’s putting in a “playoff-level” effort can easily beat a very good, very talented team that is slacking. With that in mind, one of the reasons those last 10-15 wins to get to 70, 71, 75 wins (or what have you) are so difficult is because the team lacks mental urgency after distancing itself from the pack and creating a big enough cushion to have assured in essence all the advantage the regular season avails a team: home court adv throughout the playoffs. Starters then rest, players don’t push as hard, etc etc.
You could go on about this at length, but its an interesting phenomena that is very specific to basketball. In football, there are too few games to ever “coast” and in baseball there are too few teams making the playoffs… basketball, however, has a far-too-long grind of a season and a significantly high % of teams going to the postseason, so you regularly see good teams take it easy against bad teams, rest their stars when they could feasibly play, etc etc.
October 25th, 2009 at 12:32 am
Oh, and I totally forgot to add:
…and this idea of personality can easily be measured by things like win % at home vs +.500 teams, win % on the road vs -.500 teams, etc etc. You can also see it in individual players who always seem to “get up” for marquee match-ups with their opposing player, or nationally televised games, etc etc
October 25th, 2009 at 6:27 am
Who cares about also rans? The Lakers will win again.
October 25th, 2009 at 10:02 am
I love this topic and the analysis it has inspired. I love reading Simmons. But, over the years, I’ve learned not to trust a prediction he makes.
October 25th, 2009 at 12:50 pm
What everyone else said about the +/- 17 games thing.
October 25th, 2009 at 12:50 pm
Scott Herbst says: “But, over the years, I’ve learned not to trust a prediction he makes.”
Whose predictions do you, uh, trust? There isn’t a single sportswriter anywhere who offers reliable predictions.
October 25th, 2009 at 6:12 pm
Brandon Roy is the unquestioned leader of the Trailblazers.
Brandon Roy reveres Nate MacMillian.
Bill Simmons is an idiot.
October 25th, 2009 at 8:04 pm
You, sir, and other Blazer fans must be unfamiliar with Andre Miller.
October 26th, 2009 at 12:04 am
i think stats can illuminate some nuances of games played and i think they can’t begin to capture others. i grew up watching the celtics 69 though 87 and there were too many players, havlicek, white, silas, bird to mention a few, whose actions just couldn’t be quantified. i’ve spent the last decade with tickets to the spurs and duncan and ginobili defy quantification—there is just no accounting for the effect the mere presence of certain players on the court has on what other players do or don’t do. you can see that watching the game, but it’s very hard to put a number to it and use it as a predictor or systematic explainer. basketball is not nearly so easy to explain as politics