Hollinger Projected PER Rankings

[PapaG]

Some data that stuck out at me.

Roy at #11 overall - 22.56 - 3rd for SG

Oden at #25 overall - 19.11 - 5th for C

LMA 8th for PF - 18.55

Outlaw 18th for SF - 14.78
Batum 28th for SF - 13.66
Hedo 29th for SF - 13.56

http://sports.espn.go.com/nba/teams/hollinger?team=Por - Blazer player link

http://sports.espn.go.com/nba/hollinger/rankings - NBA rankings

 

[Natebishop3]

Some data that stuck out at me.

Roy at #11 overall - 22.56 - 3rd for SG

Oden at #25 overall - 19.11 - 5th for C

LMA 8th for PF - 18.55

Outlaw 18th for SF - 14.78
Batum 28th for SF - 13.66
Hedo 29th for SF - 13.56

http://sports.espn.go.com/nba/teams/hollinger?team=Por - Blazer player link

http://sports.espn.go.com/nba/hollinger/rankings - NBA rankings

I think Batum is going to be significantly higher than that.

 

[Minstrel]

I think Batum has a decent chance to reach 15.0 PER, that of an average player.

Hollinger's projections for Roy and Aldridge expect a bit of regression, which tends to be a safe assumption in projections for players above the mean unless they've been producing at that level for several years. That said, I think Aldridge's strong finish to the season suggests some remaining upside and I'd project a 19.0 - 20.0 PER. Roy's projection is probably fair, but there's always the chance that he takes another step forward and cements himself as a truly elite player (pushing past the 25.0 PER level).

 

[huevonkiller]

These projections are all incomplete without an accurate measure of time.

Oh well, I expect Roy to be just as good, and 60-something wins maybe too.

 

[andalusian]

The interesting part is this:

For each player, I use as a comparison the players from the past 20 years who are the most similar, based on age, height and stats over the past three seasons. Some players will have more comparables than others, depending on how unusual they are -- guys with freak heights (Zydrunas Ilgauskas, Nate Robinson), freak ages (Shaquille O'Neal) or freak stats (Andrei Kirilenko) will have relatively few, while a more generic player like Al Harrington or Devin Brown could have over a hundred.

I wonder if he lists the "most similar players" for each one - it would be interesting to see.

 

[drexlersdad]

He's way off every year remember how batum was supposed to be a bust?

 

[number 10]

He's way off every year remember how batum was supposed to be a bust?

He made several mistakes with Batum - firstly, taking his Euroleague stats (IIRC, his team was 0-16 in these games against much richer teams) and not league stats. Secondly, statistics are less important for young players. Finally, statistics don't take into account Batum's greatest strength, his defense.

On the other hand, his warnings of regression from guys like Blake or Prz should be taken seriously, it's a very real possibility that they both had career years last season.

 

[MrSelfDestruct]

Interesting that he has Udoka listed but doesn't have Collins.

 

[STOMP]

so of the 11 Blazers he projects, his crystal ball says 9 of them are going to regress and only 2 will improve on their 2008-9 season??? Correct me if I'm wrong, but don't most players either improve or stay relatively the same until age 30 or so?

STOMP

 

[Boob-No-More]

so of the 11 Blazers he projects, his crystal ball says 9 of them are going to regress and only 2 will improve on their 2008-9 season??? Correct me if I'm wrong, but don't most players either improve or stay relatively the same until age 30 or so?

Interesting that the only two he sees improving are two 20 - 21 year old second year players who are likely to improve through natural growth, and likely to both play more minutes this season. Not exactly a huge reach there.

But, what about all our other young players? Have they all really peaked at 22, 23, 24 and 25years of age. ALL of them? Are the ALL on the downward side of their brief NBA careers?

I think Hollinger tends to be cautious in these projections - with most players regressing toward the mean. But, for every player that sees their production decrease, shouldn't there also be comparable increases? If the mean is the mean, shouldn't that be the case? By definition, a PER of 15 is league average. By definition, you can't really have 11 out of every 13 players with decreasing PERs. So, does he have other teams with predominantly increasing PERs?

BNM

 

[andalusian]

I think Hollinger tends to be cautious in these projections - with most players regressing toward the mean. But, for every player that sees their production decrease, shouldn't there also be comparable increases? If the mean is the mean, shouldn't that be the case? By definition, a PER of 15 is league average. By definition, you can't really have 11 out of every 13 players with decreasing PERs. So, does he have other teams with predominantly increasing PERs?

I do not think he is being cautious - I think it was clearly stated what I quoted earlier - that the projections are based on similar players from historical data. What this does not take into account is how well KP and company drafted and how good a development staff that Blazers have on their coaching roster.

Remember that when it came to his team projections - that tend to judge the team by it's own performance as a unit without that much emphasis on "other players" - and he considered that the Blazers will improve to be sole owners of the 2nd place in the west.

 

[STOMP]

I think Hollinger tends to be cautious in these projections - with most players regressing toward the mean. But, for every player that sees their production decrease, shouldn't there also be comparable increases? If the mean is the mean, shouldn't that be the case? By definition, a PER of 15 is league average. By definition, you can't really have 11 out of every 13 players with decreasing PERs. So, does he have other teams with predominantly increasing PERs?

Comparing last year's results to 2009-10, he has these 11 Blazers to go -6.75 down for their collective PERs... thats a big number. I'm sure it's mostly math and not that subjective but it seems to suggest this particular player projection system might have a wider then desired level of expected error.

The Blazers are stacked with a big healthy deep and talented team going into the 2009-10 season... we shall see

STOMP

 

[Draco]

I think Hollinger tends to be cautious in these projections - with most players regressing toward the mean. But, for every player that sees their production decrease, shouldn't there also be comparable increases?

Hollinger's numbers don't make a lot of sense. Looking at the top 15 PER players, I have highlighted the three with a projected increase;

1. CP3
2. Wade
3. Lebron
4. Howard
5. Durant
6. Jefferson
7. Nowitzki
8. Bynum
9. Bosh
10. Kobe
11. Roy
12. Duncan
13. Parker
14. Ginobili
15. Gasol

So he is saying 12 of the top 15 players in the league, measured by PER, will have a decrease in PER? Only Durant, Bynum and Bosh will have an increase? Yet even their increases are a modest 2.65, 2.79, and 0.62 respectively?

I usually enjoy Hollinger's analysis but his projections do not look feasible.

 

[blazerboy30]

What makes his PER projections even more interesting what he already said about the Blazers as a team:

2008-09 Recap
The Blazers racked up the point differential of a 59-win team and led all NBA teams in point differential after the All-Star break -- which, as longtime readers know, is a better predictor of future success than win-loss record.
...
Certainly, I'd expect them to get past the first round this year. After that it's a bit more of a crapshoot, but if the Blazers accomplish the same things they did a year ago and aren't beset by bad luck, they're a 60-win team. Chances are they'll experience an injury or two along the way and fall short of that standard, but they'll still rank among the last teams standing out West.

Prediction: 55-27, 1st in Northwest Division, 2nd in Western Conference

So he expects 11 of 13 players to be worst than last year, but expects us to be better as a team.

note: Yes, I realize that these statements are not necessarily contradictory. But it is still interesting that so many players could get worse, including our best players, and the team gets better.

 

[number 10]

So he expects 11 of 13 players to be worst than last year, but expects us to be better as a team.

pretty sure "expects" is the wrong word here. His formula projects said 11 players to be worse. The man himself might (and based on his writing, probably does) have a different opinion altogether.

 

[blazerboy30]

pretty sure "expects" is the wrong word here. His formula projects said 11 players to be worse. The man himself might (and based on his writing, probably does) have a different opinion altogether.

You don't think he believes in his own model that he publishes? Interesting.

 

[MARIS61]

Hollinger has never been good at projections. He should stick to stat recording.

 

[PtldPlatypus]

What makes his PER projections even more interesting what he already said about the Blazers as a team:



So he expects 11 of 13 players to be worst than last year, but expects us to be better as a team.

note: Yes, I realize that these statements are not necessarily contradictory. But it is still interesting that so many players could get worse, including our best players, and the team gets better.

Well, the two expected to improve are going to see significantly more minutes, ditched Frye and Sergio (low PER players) and added Miller (projected 1.9 above average). It would make sense that most of the main players' PERs might dip a little.

 

[Minstrel]

So he is saying 12 of the top 15 players in the league, measured by PER, will have a decrease in PER? Only Durant, Bynum and Bosh will have an increase? Yet even their increases are a modest 2.65, 2.79, and 0.62 respectively?

I usually enjoy Hollinger's analysis but his projections do not look feasible.

Statistically, there's nothing unusual about "outliers" (the data points at the extreme top end and bottom end) showing a bit of regression to the mean in general. It's quite common in a lot of places.

Speaking specifically about athletes, it actually does make sense: the top players in any given year are the set of very good players who had very good seasons. That specific set of players are likely slightly above their career averages and will regress slightly toward their career averages.

As has been noted, Hollinger is modeling players on the combined career paths of the player's most-comparable past players. So, in a sense, what you see in the projections is what has tended to play out through NBA history.

Baseball Prospectus' PECOTA projection system does something similar and generally expects a bit of regression from the past season's top performers also.

 

[Minstrel]

note: Yes, I realize that these statements are not necessarily contradictory. But it is still interesting that so many players could get worse, including our best players, and the team gets better.

Well, the best players on most teams are expected to show a slight decrease too. I don't think that actually changes much in a relative sense.

I think Hollinger mostly expects the team to do better because the team underperformed its points scored/points allowed, in terms of wins, last season. Assuming that was bad luck (and there have been plenty of theories on this forum about why it may not have been luck), one should expect the team to improve with neutral luck.

 

[number 10]

You don't think he believes in his own model that he publishes? Interesting.

Sure he believes in it, just not with 100% accuracy. I've noticed that in some of his player profiles, he'll comment on other factors (injuries, playing time, strange coaching decisions) that might be misleading the system.

 

[magnifier661]

Some data that stuck out at me.

Roy at #11 overall - 22.56 - 3rd for SG

Oden at #25 overall - 19.11 - 5th for C

LMA 8th for PF - 18.55

Outlaw 18th for SF - 14.78
Batum 28th for SF - 13.66
Hedo 29th for SF - 13.56

http://sports.espn.go.com/nba/teams/hollinger?team=Por - Blazer player link

http://sports.espn.go.com/nba/hollinger/rankings - NBA rankings

Well this is coming from a true "Blazer Homer", but I think LMA and Roy will be making the Hollinger top 10 next season. Those two look awesome and a year wiser. If the Blazers can make it to the WCF this season, I suspect respect will be deserved.

 

[oldmangrouch]

You don't think he believes in his own model that he publishes? Interesting.

Using a mathmatical model to project human performance is never an exact science. The model may be "right" for most players (eg his projection that Blake and Joel will regress) - but there will always be some players outside the norm.

Would you have more confidence in Hollinger if he didn't admit that?

 

[blazerboy30]

Using a mathmatical model to project human performance is never an exact science. The model may be "right" for most players (eg his projection that Blake and Joel will regress) - but there will always be some players outside the norm.

Would you have more confidence in Hollinger if he didn't admit that?

Obviously it isn't an exact science. And I know he would admit that. I never brought up anything expecting "exactness", or even being "right".

I was responding to the claim that using the statement "Hollinger expects" was so much different than "Hollinger's model predicts". If he won't standby his model enough to substitute the words "model predicts" with "expects", then we shouldn't even be wasting our time by discussing the model or its predictions.

 

[Boob-No-More]

Using a mathmatical model to project human performance is never an exact science. The model may be "right" for most players (eg his projection that Blake and Joel will regress) - but there will always be some players outside the norm.

Actually, based on your first sentence, I think he may be wrong about Blake (and possibly Joel). His PER formula makes adjustments for pace and minutes played. So, while Blake (and Joel) will probably see reduced minutes this season, I'm not so sure that his actual PER will decrease. If Blake plays most of his minutes against other back-up PGs, his PER may actually go up, even though his per game points and assists might go down.

Also, Hollinger uses historical data based on similar players at similar ages. If there's one thing Blake has shown it's that he's a MUCH harder than average worker. Most guys with Blake's natural talent are lucky to even get drafted, let around stick around the league as long as he has and become the starting PG on a 54-win team.

He posted his career high PER last season, and with his starting role in jeopardy, he's obviously worked hard during the off season to improve even more. Based on Nate's conditioning test, Blake is in incredible shape. He also looked great in the Fanfest scrimmage (I know, it's just a scrimmage, but his shot was in mid-season form, he played agressive and looked damn good out there).

I guess that's just one way of saying don 't sell Steve Blake short. People have been underestimating him his whole career and he just keeps proving them wrong.

I do think Andre Miller will end up being our starting PG, and playing more minutes than Blake, but I expect Blake to be extremely productive in the minutes he plays - quite possibly THE best back-up PG in the league. I also expect Joel to be THE best back-up center in the league this season.

BNM