http://sports.espn.go.com/nba/news/story?id=3469696
Great, great news. Paul is the best player in the league and the Hornets would be nothing without him. If CP3 were to leave, they’d not only lose his production but a percentage of everyone else’s. He is absolutely essential to that franchise – and he’s a character guy too. Pay him whatever he wants for as long as he wants.
The National Hurricane Center, which is the best in the world at what it does, employs two sorts of models in predicting the path of hurricanes.* Dynamical models actually analyze the atmosphere, taking into account the real position of weather systems, wind speeds and directions, surface temperatures, pressure gradients, and the like to predict where the storm will go. Statistical models, by contrast, simply predict where a storm will go based on where storms like it went. There’s no analysis of the atmosphere – it’s just the most likely case based on past storms.
Let’s apply these concepts to forecasting, say, the success of NBA draft picks. A dynamical model might say “well, this guy’s got a much better three point shot than most power forwards, but he has some significant character issues as well. These roughly cancel out, so he’s as valuable as the league-average power forward of his age, and can be expected to produce statistically as though that were true”. A statistical model would say “power forwards at age 20 with a PER of 12 are, on average, league-average performers over their career. This guy is a 20 year old PF with a PER of 12, so he’ll be league-average”.
The NHC uses a statistical model called CLIPER5 as a baseline for its own predictions. “CLIPER” stands for CLImatogy and PERsistence. Basically, they look at which way the storm is going now, which way other storms in that location at that time of year have gone, and combine the two into a track prediction. In order for the NHC to consider its forecasts as “skilled”, they have to outperform the CLIPER model. In other words, the NHC guys aren’t adding any value if they do no better or worse than the CLIPER forecast, which is basically an Excel sheet that needs updating. The CLIPER5 model helps the forecasters to understand the usual situation and apply the unique circumstances of the present to get a more accurate forecast. However, its most important function is to serve as a way to know if the NHC’s forecasts are any good.
The NBA Draft needs a CLIPER5 model. Some basic statistical prediction, which is designed to be dumb and to ignore context, whose sole job is to filter out the undesirables and get at the truth. I will design such a model, and we will see if Kevin McHale has any business being an NBA GM. It’ll be several days before I can assemble the data, but it will be fun.
Any comments on the statistical model are welcome. Remember, it’s going to be all numbers, with no subjective judgments of any kind. The model is supposed to be dumb, mostly so we can see if NBA GMs are dumber.
* Of course, this oversimplifies weather modeling to an alarming degree. Just trying to make a larger point here.
http://www.midmajority.com/2008/04/fearlessness-and-failure-epilo.php#more
Scroll all the way to the bottom and play the video. I seriously couldn’t watch all of it, it’s so touching.
OK – so ESPN has embraced this damnable trend of encouraging you, the viewer/reader/pork rind consumer, to comment on the world of sports, and to put this stuff on their front page. This prize pig comes from “zags352″, who is clearly an objective, qualified member of the viewing public, who has a considered opinion that differs from ESPN’s resident bracketologist. This opinion is……
Lunardi gives no love to the Zags. A No. 4 seed? Do you see all the talent they have?
Where to start? Well, first of all, Lunardi’s column makes no value judgments about how “good” a team is. He’s simply predicting the field. Now, doing this in June involves a fair bit of assuming what teams will do, as Duke is currently as qualified as Texas A&M-Corpus Christi to enter the tournament. In fact, Lunardi regularly disagrees with the committee’s choices, and often indicates these disagreements in his column. However, his bracket is a prediction of the committee’s actions. He missed no teams last year (in his final bracket; I’m quite sure he missed some in June). Could “zags352″ get that done? Unlikely. Anyway, “zags352″ may have lots of inadequacies, but none of them affect the key issue: Lunardi is making a prediction of seed, not a value judgment about seed.
Second, Gonzaga only once received a #2 seed, and has only twice received a #3 seed (obviously, no other WCC team has ever come close to that in the modern tournament). For “zags352″ to have any real complaints would require he/she/shim to maintain that Gonzaga will earn at least a #2, as a one seed difference can easily be explained by a regional shift. This happened one time, back in 2004, which was a season with a notable dearth of qualified top seeds. What’s going to happen this year to at least duplicate those conditions, combined with a top notch season from the Zags? They would need to go unbeaten or very, very close (two losses max) to get a #1 seed, and not much less would be required for a #2. Ain’t happenin’.
Now, let’s look at the twelve teams above the Zags. UNC returns a devastating roster (and plays a #1 seed’s schedule). Memphis brings back a good bit of talent and has their usual devastating recruiting class. Pitt has a battle tested team that plays more D than anybody and brings back two of the top players in the country (and plays a #1 seed’s schedule). Duke is loaded with shooters and shouldn’t lose more than a few times, most of which to UNC (and plays a #1 seed’s schedule). Arizona…..well, I quibble with this one, but if you’re just talking “talent” as the chosen criterion of “zags352″, Arizona’s problem has never been talent, and won’t be this year either. Notre Dame has perhaps the best combo forward in the country in Harangody and brings back an offensive arsenal that will likely rate as one of the best offenses in America (and plays a #1 seed’s schedule). UCLA has been there, done that, and brings a very strong defensive lineup, although one that struggles to score (and plays a #1 seed’s schedule). UConn has tons of talent, although I doubt they’ll get these results (but again, talent is our criterion of choice). Louisville, MSU, and Texas are loaded, while I see Purdue as being at around the same level of Gonzaga. Oh, what else is true of those five teams?
(they play a #1 seed’s schedule)
Look, I’m on midmajority.com as much as anybody. I grew up in Tulsa following the Golden Hurricane in the ’90s. They were a win from the Final Four, and should have been able to take care of business and get there. I love the basketball that’s played in the Patriot League, the MEAC, and the WAC (oh snap!). But the Zags don’t deserve any seed based on talent. And neither does anyone else. Lunardi was predicting what they’d get based on the committee’s behavior and his expectations for Gonzaga’s season, and they’d need to have a historically great season to qualify for a #1 or #2.
Edit: And, let’s be honest. If we’re picking pickup teams off of Lunardi’s top 4 seed lines, Pargo and Heytvelt are the only two that would be picked before the vast majority of the other seed lines would be decimated. I can think of at least eight Tarheels, eight Dookies, and seven Longhorns I’d take before I’d choose a third Zag. It’s not that they’re not talented: it’s just that this standard is a lousy reason to say they’ll get a better seed than Duke, UNC, or even Texas.
Well, I’m here to find out. The league average three point shooting percentage was .362 on an average of 1485 attempts (that’s an average of eighteen per game). Thirteen teams attempted more than 1485 threes on the season. Of these thirteen, seven shot better than league average and six shot worse. We cannot assume that those who shot better than league average were making a good bet, as they might have been even more efficient from two point range. However, let’s look at those six teams who shot more threes at a worse rate (Cleveland, Denver, Washington, Houston, Memphis, and Golden State).
Cleveland took sixty more threes than you’d expect and posted a percentage almost exactly league average, so really nothing to see here. However, the biggest story is that LeBron James hurled up 359 threes at a 31.5% clip, which is well below the league average. Anytime your most frequent three point shooter is shooting well below league average, something needs to change. Bron-Bron is a terrific player, but this element of his game either needs to improve or become less prominent. On possessions where LeBron shoots a three, he becomes a league average player at best.
Denver took 120 more threes at a rate just below league average. So, on the whole, not much to see here. Kleiza and Iverson took the most threes on the team and shot worse than the team from long range. This is not acceptable: too many shots chucked up there, too many wasted possessions. This again is a team that shot better than the league from the field, indicating that better options were available.
Washington is a poor shooting team anyway. Antawn Jamison took too many threes at a bad rate.
Houston had the worst shooting percentage from three of any of the teams who shot more threes than league average. Tracy McGrady hucked up 300 threes at a 29% clip, while guys like Rafer Alston shot 400 at a rate worse than the league. Shot selection is terrible on this team. Their non-three shooting percentage is pretty average, but they’re taking too many bad shots. This is one reason that a team that boasts two offensive superstars in Yao and McGrady can nevertheless be a mediocre offensive club.
Memphis shot three hundred more threes than you might expect. Kyle Lowry’s 140 three point attempts resulted in just 36 makes. Rudy Gay was also below average with nearly 400 threes. Young guys, and you’d expect them to improve.
And then…..there’s Golden State. They proudly hoisted SEVEN HUNDRED more threes than the league average – that’s nearly nine extra threes per game, all of them shot at a below average accuracy. Who are the culprits? Baron Davis, for one, took 525 threes and missed over two thirds of them. Stephen Jackson shot 503 at just about league average. Matt Barnes shot 181 at under 30%. Their other primary three shooters were average to good. But Baron Davis is killing them from three, and he’s doing it six and a half times per game.
This raises another question: all of these teams feature a star player who’s shooting a lot of threes at a bad percentage. Is this a leaguewide trend? Let’s look at the top ten guys in the league:
1) Jason Richardson, 599 attempts, 40.6%
2) Rashard Lewis, 533 attempts, 40.9%
3) Baron Davis, 525 attempts, 33.0%
4) Peja Stojakovic – 524 attempts, 44.1%
5) Stephen Jackson – 503 attempts, 36.3%
6) Jamal Crawford – 494 attempts, 35.6%
7) Ray Allen – 452 attempts, 39.8%
8 ) Joe Johnson – 444 attempts, 38.1%
9) Raja Bell – 439 attempts, 40.1%
10) Juan Carlos Navarro – 432 attempts, 36.1%
Thankfully, all but three of these guys are at least league average. Navarro is only a tenth of a percent below, so that isn’t really an issue. But Crawford and Davis stand out as real chuckers. Davis, for instance, cost his team a point per game from three when compared to the league average three point shooter. That doesn’t seem like much, but keep in mind that championship teams typically only outscore their opponents by an average of seven to nine points per game. These shot selection issues matter, and a mere swing of five points per game can be the difference between making and missing the playoffs, or making the playoffs and being a factor there. We’ve got a few teams whose shot selection is costing them a significant portion of that number. Many of the culprits are at least established NBA players, and many (including LeBron, McGrady, Baron Davis, and Allen Iverson) are bona fide stars.
