Pelton, With a More Agreeable Projection

by Ziller on Oct 23, 2008 10:30 AM PDT Comment 15 comments

Three weeks ago, John Hollinger's ESPN.com projection hit the web. It wasn't pretty.

The Kings are going to be substantially worse than they were a season ago, and I'm not sure they're prepared for how hard a fall this is going to be. Looking up and down their roster, they have only one above-average player (Martin) and several positions that shape up as major question marks.

Hollinger predicted 23 wins for the team, and 14th place in the West (ahead of only Oklahoma City).

ESPN.com's full Daily Dime previews came out this week. Every last one of the .com's pundits -- folks like Ric Bucher, Scoop Jackson, Marc Stein, Jemele Hill, Jalen Rose -- put the Kings in last place in the Pacific. No one had the Kings above 12th place in the conference, and two picked the Kings 14th.

Kevin Pelton of Basketball Prospectus has a background similar to that of Hollinger: he's a writer with a statistical bent. He's long been one of the top writers of this mold, and really deserves a bigger audience. And no, I'm not just saying this because his Kings projection makes more sense.

Don't get me wrong: Pelton still projects the Kings to be awful (33 wins), worst in the division and close to worst in the conference. But all the little ticks he tosses into his dispatch actually make sense.

The Sacramento Kings will likely take a slight step backwards this season, but in the wake of the Ron Artest trade the team has a clear direction. The rebuild is officially on, though with an All-Star-caliber shooting guard in Kevin Martin and some other solid pieces, the Kings are probably not going to be one of the league's absolute worst teams. This year, the goal is not so much measured in terms of wins as developing Spencer Hawes and Jason Thompson and continuing to sort through pieces to figure out who will be part of the next great Sacramento team.

To be fair, Hollinger admits this is a rebuilding season and things appear headed toward the right track. But looking at each player's projection comes off more likely to be reasonable than do those of Hollinger.

And it seems incredibly weird to me that two guys with similar methods could come up a full 10 games apart on a projection for a team in which only one major player moved.

0 recs | Comment 15 comments | Share

Story-email Email Printer Print

Comments

My beef

with Hollinger’s prediction is that he puts the Kings at -15 due primarily to losing Artest, who only played in 57 games last year. He does not take into account the litany of injuries, the fact the Kings go into this year with the same coaching staff, etc.

Of course, this is like arguing how bad is bad. Will the Kings miss the playoffs by 10 games or 25 games? At the end of the day it probably makes little difference to most folks that are not Kings fans or go by the name of Reggie Theus.

SACTOWN ROYALTY - Try our thick creamy shakes!

by section214 on Oct 23, 2008 11:44 AM PDT reply actions 0 recs

Hmmm

And it seems incredibly weird to me that two guys with similar methods could come up a full 10 games apart on a projection for a team in which only one major player moved.

Well, in the eyes of many, Bibby is a major player too. I disagree with that, but the know it all national media, well, they still think it.

No mistakes in the tango, darling. Not like life. It's simple. That's what makes the tango so great. If you make a mistake, and get tangled, you tango on

by pookeyguru on Oct 23, 2008 12:04 PM PDT reply actions 0 recs

Yeah, shows how even more idiotic they were

Bibby missed most of the games. Beno is shaping up to provide the same level of production at half the price. Yes, there are question marks of health, but we’re not going into season knowing he’ll be out for three months and contemplating starting Bobby Brown, are we? No, no we’re not.

We’re losing star power, yes, but there replacements are going to be there to play most of the time this time around. And oh yeah, the guys replacing them? they were there for all those minutes the other guys missed with injuries and suspensions.

If I’m getting paid to know basketball, I’m not relying just on my previous experience and presumptions to influence my opinions that people are expected to pay for the oppurtunity to read. Same thing as any job – if someone’s paying me to be an IT admin, I’m not exactly proposing we invest more in ten-year-old technology; I’m reading the latest info to determine the best course of action, not relying on the random headlines I remember from the last two years.

Donte? Donte'! Donté?!?!
'spect da 'xtra E'

by iashwash on Oct 23, 2008 12:18 PM PDT up reply actions 0 recs

Losing name power & star power aren't the same thing

Bibby is mostly a name from the past even today on the Hawks.

No mistakes in the tango, darling. Not like life. It's simple. That's what makes the tango so great. If you make a mistake, and get tangled, you tango on

by pookeyguru on Oct 23, 2008 10:04 PM PDT up reply actions 0 recs

previews

Maybe we might conclude that previews are irrelevant generally and the purely-quantitative previews are even less relevant.

Hence, we might preserve nothing but the pure and flagrant optimism for the future of the Kings (as crazy fans)

by KingsFanfromCentralEurope on Oct 23, 2008 1:36 PM PDT reply actions 0 recs

I like the way you think

Let's go home.-Kevin Martin

by LeaguePassAddict on Oct 24, 2008 10:11 AM PDT up reply actions 0 recs

33 seems about right to me

What I don’t get is the Cuttino Mobley comparison for Kmart. The dude’s peak was 21.7 per and because he dropped off the next season, we should expect Kevin to drop off too?

I guess what I really disagree with is using a past player’s career to predict a current player’s career based on their 1st couple of seasons being similar. Too many other variables to make any correlation valid.

What I do agree with is when they do a statistical analysis of all past players to make general assumptions about a current player’s career. ie historically players tend to peak at 26-27 so we should expect Kevin to peak over the next season or two.

Ball movement ... is like jogging for most people: They do it occasionally, and it makes them happy. Then they go back to not doing it. - Henry Abbott

by Kfan in Korea on Oct 23, 2008 1:53 PM PDT reply actions 0 recs

Comparisons

Comparisons have worked marvelously in baseball. Nate Silver’s PECOTA (which is the inspiration for Pelton’s system) did excellent this year and most years. It’s completely up in the air whether they work in basketball, precisely because the data set is a) noisier (basketball outcomes being far more interdependent than baseball outcomes), and b) smaller (baseball has a longer history, and a longer history with the same set of rule).

That said, it remains to be seen whether comparison-based projection works in pro basketball. Are there molds you can build, or is each player a unique snowflake?

by Ziller on Oct 23, 2008 2:32 PM PDT up reply actions 0 recs

Ah, makes some sense

I stopped following baseball about 15 years ago when it became clear to me that small market teams really had no chance other than a perfect storm year of developing youth, and Free agency had ended any long term team identification for me. So I missed the whole statistical revolution.

I’m not a mathematician, but my argument against that method working for basketball but being successful for baseball would have to do with they type of sport and the number of players. Baseball is essentially an individual sport. Your teammates don’t affect your batting avg, on base percentage etc. There are some teammate effects regarding RBI, if there are runners on base you might see better pitches and some other categories, but in general it’s Pitcher vs. Batter.

With basketball one player’s offense is very dependent upon his teammates. As we’ve seen over the past couple of seasons. Kevin may be efficient but if Bibby and Artest won’t pass him the ball he’s not going to get the most shots. A team’s offensive system, pace, and teammates all have an affect on a player’s offensive stats.

Kobe’s stats last season are another example. Some of his individual stats went down, was this because his skills declined or because Bynum improved and Gasol arrived? Did Rondo improve last season because his skills improved or because KG and Ray Allen arrived or both?

My point is two players could match very closely for 6 straight years and you think I’ve got the perfect model for predicting, but in their 7th years, player A had a hall of fame teammate retire and player B has a hall of fame teammate join his team. Were the players statistically similar for 6 years because they are the same kind of player, because player A was being held back or improved by his teammate while player B benefited by playing on a poor team with no other scoring option or a faster pace and their respective abilities are really nowhere near similar? Is this accounted for in the statistical models? I don’t think so.

So, IMO until they can find a way to statistically match players with some of these “team/teammate similarity” factors included, I don’t see how a statistics based historical precedent prediction formula could possibly work in basketball.

Ball movement ... is like jogging for most people: They do it occasionally, and it makes them happy. Then they go back to not doing it. - Henry Abbott

by Kfan in Korea on Oct 23, 2008 4:40 PM PDT up reply actions 0 recs

Hmmm interesting points

The biggest problem I see, as you say, is using baseball models to start the basketball one’s. Guys like Kevin Pelton, as best as I can tell, and am hardly anything of a judge, is trying to start a system. I think even Pelton would have to agree that using the baseball system and transferring it to the basketball side of things would be a waste of time.

On the other hand, if you start with a very flawed model, you can work your way up with models that are less flawed and with better design to show the type of information you’re seeking to show.

No mistakes in the tango, darling. Not like life. It's simple. That's what makes the tango so great. If you make a mistake, and get tangled, you tango on

by pookeyguru on Oct 23, 2008 10:10 PM PDT up reply actions 0 recs

It's nice to see

a prediction that takes into account that there is a multiyear plan and one season does not spell absolute doom and gloom for the team. Also great to see KMart get a little love.

Napear: "Going in to the final timeout what did the coach tell you?"
Bibby: "I asked the coach if I should go and he said, I don't really know what he said."

by kangsfan on Oct 23, 2008 2:28 PM PDT reply actions 0 recs

The Key is

We gotta play with the same intensity we did last year especially at Arco. Arco Arena has got to be very, very good for the Kings this season. The early signs in the preseason aren’t as hot this year as they were last year.

BOOK IT!

by kingme18 on Oct 23, 2008 2:29 PM PDT reply actions 0 recs

Similarity

Kfan, not sure whether you read my introduction to the system, but the model is based on ideally more than 50 comparable players. Over that large a group, the kinds of factors you mention, which are totally reasonable, should tend to even out.

As Tom said, we don’t really have a lot of evidence yet that this method works in the NBA (I’ll be in a much better position to know after this season), but I think it’s much better than saying that a point guard and a center of the same age will develop the same way.

by kpelton on Oct 23, 2008 6:16 PM PDT reply actions 0 recs

I didn't read the introduction

I didn’t see it. I just scrolled down and checked the Kings write-up. I can see where including more than 50 players would be much more viable. I just saw the Cuttino Mobley comparison and went to look at his career numbers. And while I can see similarities, I didn’t see conclusions.

I just looked through the introduction. No matter how well your system turns out, I’d say some info gleaned is better than no info. It’s always good to get new and better ways to asses and predict. Thanks for the response and good luck refining your system.

Ball movement ... is like jogging for most people: They do it occasionally, and it makes them happy. Then they go back to not doing it. - Henry Abbott

by Kfan in Korea on Oct 23, 2008 7:14 PM PDT up reply actions 0 recs

I also liked the point myself

About Johan Petro. As Pelton noted, there aren’t that many 20 year old 7 footers to choose from.

No mistakes in the tango, darling. Not like life. It's simple. That's what makes the tango so great. If you make a mistake, and get tangled, you tango on

by pookeyguru on Oct 23, 2008 10:20 PM PDT reply actions 0 recs