will be ridiculous in 3 years. No mercy massacres on a frequent basis.

Get ready to break our your rec stick.

(you can delete this evidence comment)

I’m glad he saved me the time and did it himself.

The only comments that I have (and they related to the way I analyzed the issue of overall strength given age) are that the age-defense relationship is probably not linear and might be slightly different than the age-experience relationship. I’d bet you a $100 that a team of 60 year old NBA veterans would not dominate the league on defense. Thus, I think he’s adjustments probably over penalizes San Antonio (or, implicitly, the San Antonio coaching staff). See the post below for examples of non-linearity in the age-success relationship.

http://www.blazersedge.com/2008/12/12/690628/ages-experience-and-win-in

I agree with his general conclusions. I’m not at all surprised that the Blazers would be above average if you adjusted for age, but not necessarily the best.

Extra credit problem: how many years will it take the Blazers to reach the top five, if the Blazers were to make no changes to their roster. This question can be answered.

is my favorite part of Blazersedge. It speaks my stat geeky soul.

I was impressed that 18% of defensive efficiency is due to age… that is far more than I expected. My next question would be how age affects offensive efficiency. Does it have similar results? I would tend to think that it has less of an impact, but it would still be a very interesting study.

Advanced stats have revolutionized the way we look at baseball, and now basketball is coming around. We wouldn’t have dreamed of this kind of stuff ten years ago.

As someone who makes these kind of plots for a living, I call shenanigans on those trendlines. A correlation of 0.434 is terrible and simply says that age has very little to do with defensive prowess.

But I’ll echo torso and say that the correlation seems pretty low, though the # of plot points isn’t particularly fantastic (and that’s part of it)

John Hollinger’s Defensive Efficiency rating measures the number of points a team allows per 100 possessions. To see his ratings, you must have ESPN Insider, but to give you an idea: Boston rates 1st, Portland rates 18th and Sacramento rates 30th.

Maybe I shouldn’t post this because if ESPN sees it they’ll lock it down, but if you go to Hollinger Power Rankings and then under any team click on Hollinger Stats it will take you to his efficiency ratings. You can sort all you want just as if you were really an insider. And you’ll see that Portland is now 17th in defensive efficiency (2nd in offensive).

Great analysis here, I love this kind of stuff. I do marketing analysis in my job so i am actually able to follow along really well.

Also goes to show that us Blazer fans really to forget how young the team is and that we do need to be patient.

Also if you like this type of stuff you will love the book Freakonomics which applies this same principles to social issues, great book.

A quick question of clarification. In the charts, a higher defensive rating is better, but in the formula and in the charts explaining AADR, lower numbers are better? Did a miss when and why he went back and forth between the two metrics?

i’ll be wearing mine on Thursday: 7…i won’t be CB’ing anyone!!!

You should. This means that 19% of the Blazers defensive “woes” can be explained and corrected automatically by simply aging. We still need to improve in other areas. According to KP2, once we hit the leage average age we will be 11th in DR. That’s not bad, but not what we want.

Right now we are 4 pp100 out of being a top 5 defensive team according to Hollinger. Accoring to KP2 we will improve by 2 pp100 just by aging a couple of years. This means that Portland just needs to improve their defensive rating by 2pp 100 in other aspects of the game.

That is ONE bucket per 100 possessions that we need to prevent to be elite. Down from the 2 buckets that we all thought it was. In other words, this age stuff means that the other aspects of our defense only need to improve by half of what we thought they did.

KP predicts that coaching/players are part of this. The coaching probably won’t change, that just means the players need to. Anyone think Oden playing more (not getting older since we already adjusted for that) and Webster getting healthy and replacing Outlaw could cover this?

Being as I’m an engineer, and have taken more calculus and statistics classes than normal humans should, I can attest to the fundamental mathematical soundness of these equations.

But that sure is a lot of effort to quantify the somewhat obvious axiom that the more players play in the league,, the better they get at being good defenders.

Another aspect of this conundrum that hasn’t been raised yet: How much does money play into the desire to play defense. With offensive juggernauts with suspect defensive abilities such as Amare Stoudamire and Rashard Lewis getting huge jack, and defensive stalwarts like Bruce Bowen scraping by on 4 mil a year, why would a player invest the time and effort to develop said defensive mentality and ability when it clearly doesn’t pay to do so?

I know that’s a very conservative “money is the motivating factor for everything” way of putting it, but since the concept of team defense and actual accountability for individual defenders being so opaque, how can anyone reasonably expect NBA players to commit to becoming better defenders on their own.

we shoulda traded RLEC or not?

Get ready to break our your rec stick.

Get ready to break our (your) rec stick? NO

Get ready to break out your rec stick!

REC’D

Those of us who have felt that Portland really doesn’t need anything more than time playing together were not standing out in left field.

It would be interesting to me to see how that team looks on the same chart, how they progressed from when Pip was a rook until MJ retirement #1.
With Batum, Bayless and LA, I see a chance that this team could pressure and trap like the Bulls did then…leading to lots of 10-0 runs that break the back of the other team.

Or do we want that can of worms to go away?

Warning…do not read if you’re totally obsessed with stats and their truth.

It’s pretty colorful and all Ben, but I can take statistical figures and shove them into a formula that can prove that jfk shot himself with that magical bullet from Louisiana (or whatever state he was in) 4 days prior to his ever being shot.

We knew beforehand by the performances that the team had, that we kinda are bringing up the rear as far as Team D and the league goes.

This type of analysis in my opinion takes away from what is really happening….our team is just getting better as they should…and places an emphasis on….#s instead of what’s really bringin the goods…..intensity.

I think that Our defense is shady to ok enough to get us by the crap offensive teams of the league, but if we’re to use it against the super offensively talented one player, or entire team (Boston/L.A.) we’ll be shredded….so team D isn’t really all it’s made out to be, unless!!! It’s played by great OFFENSIVE players who can put the ball in the basket more than you can….and then in that case…yeah…they play great D because they won. Had the losers lost, would we still be praising their lockdown D?…of course not. Find me a great defensive teams LOSS that’s defensive stats match their winning efforts to the T on multiple nights…and maybe…maybe I’ll listen to the argument that d wins stuff….cuz as far as I can tell, great teams (defense and offense) lose when they have off nights…and are not putting out their usual (offensive/defensive) effort.

History along with the title of winner goes along with actually winning games….

I don’t need a figure or a fact other than the score to tell me that I score more than you I win…. that’s apparent by the fact that…we won/ or lost….. I do need stats and figures to tell me how and why I lost but they always point back to the offense…uh…..we just didn’t get the ball through the hoop more than the other guys?

It’s willingness out of each individual player to bring their O game night in and night out…..

When that’s happening, our D follow suit…..but make no mistake, offense comes first…..

IN MY OPINION…….it’s always easier to be ahead and stay ahead than it is to be behind and need to catch up….the later requires you to give more than you would otherwise.

So when our offense is getting it done…like it did last night (san Antonio) ….Our defense has a much easier time given we don’t HAVE to stop anyone…..nerves are at ease, mistakes are not magnified by the fact that another 2 points got added to the pile we have to climb…..last night we were the ski lift on the mountain with no seats and the spurs couldn’t find a way to keep up….

Our offense broke down their D…..and our D kept with the intensity of the offensive output. We didn’t shut down San Antonio….they couldn’t keep up.

Duncan still got his shots…parker still drove the lane, and baseline, and broke a few ankles……his shot wasn’t falling. Neither was Duncan’s… Lamarcus had hands all up in his face…but it was defense that made his shot fall?….D had Brandon driving with reckless abandon? D was getting Blake his assists and his 3 pt shots?….

I’ say na….that was just great offensive execution….and our energy followed on the opposite end….

All subjective, but…I’m an offensive minded fellow……don’t need a great Defense if your offense keeps the opposition on its toes for 48 min.

Lets put it this way,….if rudy was able to shoooooooot the lights out at an unheard of 90%…..would the oppositions D be credited for him having a bad night and only hitting 18 out of 30 3’s?

(nope, rudy would just be the choke artist of all choke artists if that happened)

Even the far and inbetween steal is offensively orientated…… you steal the ball to put it in your basket…. OFFENSE! GOOD team D helps, but Offense wins basketball games.

I did enjoy the game lastnight, and did appreciate the work in the post, it’s just fools gold for me. Given in my opinion stats can be made to say anything that serves anyones purposes, given they want to put in the work to get them to say what they want.

Pure unmolested numbers speak the truth……after they’ve been ran through any gizmo with an intended outcome they are now ment to serve a purpose…and are not pure anymore after that, but are twisted and presented to prove something…..that may or may not be true…tha’d be up to the decision of the audience…and in this case…. We were pretaught to expect what we were being shown.

But believe what ya want ya know….I just think overanalyzing our team takes some of the wind out of the sail that is the wins over these great teams.

First, my head hurts. This has a TON of good info, and I can find no faults with it, but trying to track it all = brain owie.

Second, take THAT Dwight Jaynes. He has ragged on the Blazrs about their defense all year, and has stated several times that the Blazers need to quit using their age as an excuse.

and its certainly cause for optimism regarding our D. I’d caution, however, that our age-adjusted D is still WAY below that of Cleveland and Orlando— teams we’re going to need to beat if we want to be champions. I’m now much more optimistic that our D will improve with time. However, I have high expectations for this team— ie championships— and I’m still not sold that we ever play championship D under Nate.

I wonder: If you do the same thing for offense, is their an equally strong trend? Is there a trend, but a weaker one? Is there no trend at all?

Really great stuff. Thanks Kevin and thanks Ben for getting Kevin to do this.

For example, have the Blazers stepped up this season from last?

I have only one thing to add.
I’d guess that the expected age effect is probably higher for us, than the estimated one.
Quality of coaching and age(experience) are probably complementaries in the production of defensive efficiency( my guess). If we assume that Nate is an above average defensive coach ( I do), than we’d expect that the age effect for the blazers to be higher.

What it would take to also isolate coaching out of this metric. It would be really hard I think but you could come up with a Defensive score for each coach based on how their teams perform but I am not sure that would vet out again because of all the variable involved. Ideas?

A lot of people are really interesting in projecting the future. I think the analysis above, give us some intuition for how to project into the future, but it’s not the way that I would do it, if I had to make a wager.

If I had the time, I would look at the following:

E(Defensive efficiency_t+1 | Defensive Efficiency_t, Age_t)

(The expected defensive efficiency in year t+t, given defensive efficiency in year t and age at year t).

This comes closer to analyzing the question in a lot of Blazers fan’s heads: how good are we going to be next year? I would be shocked if the relationship across years weren’t weaker than the cross-sectional (or within year analysis) because of roster moves made in the off-season, but I think it would be really interesting to see the results.

I know KP2 has done some projections for players across years, so I am sure he could produce something useful along these lines.

link

KP2’s system had us finishing 51-31and 3rd in the West. The system was too bullish on our D and underestimated our O, but overall its a good looking projection.

Of course, KP2’s system also had the Spurs going 42-40 this year (though he noted subjectively that this was too low).

It seems fair to say that we can draw from this that our defense will improve as we get older and that we don’t need to freak out about it.

But it raises a number of other questions, from the perspective of wanting to maximize our odds of winning a championship. Staying on the defensive end, I would like to account for other factors in the variation of Defensive Ratings.

- weighted years of NBA experience (that should be easy)
- years of coach’s experience (you could even add assistants)
- team record over previous seasons (having a "winning culture)
- length of tenure of coach and/or GM and/or owner

If anyone else can think of anything, add it to my list! Maybe KP2 or someone else will have a shot at this.

You have Miami at -3, they should be at +3 when adjusted for age. They move from 11th to 8th. Their change should be tied for 5th best, not the 6th worst.

I asked John Hollinger a similar question in regards to age and statistical analysis in a Q&A today.

http://trailblazerscentercourt.blogspot.com/2009/03/no-longer-outliers-q-with-john.html

Also interesting that Hollinger noted that, all things equal, younger players actually perform better in the playoffs than veterans. Conventional wisdom be damned!

passing for a good correlation. In KP2 post about similarity scores, I assumed that some sort of regression was being used, and we were told that similarity scores that were above 90 were worth noting, but go sketchy after that. I may have wrongly assumed that this was the R value for the regression on a different scale, 0-100. If I am correct in assuming that then why are we supposed to believe that this R value of ~0.4 should be taken seriously. I am coming from a different field, but it has been expressed before that this correlation seems low.

If KP2 could clear this possible discrepancy for me that would be great.

First, the reason for the drastic drop in defensive rating for some teams is actually due to a flaw in the statistic. The stat plots the “line” of best fit, when logically a plot of defensive efficiency and age should be parabolic in nature. At a certain point, players regress defensively as a result of age. This stat, however, assumes that no such regression ever takes place.

The reason that the Spurs, Suns and Pistons are among the greatest in decline… The stat assumes Shaq, Duncan etc. are drastically better defenders at 35 than at 25. (Once again Bruce Bowen is the bane of basketball statistics.)

It is also possible that extremes on the low end are affected similarly.

As great as this is, I would like to see the data for average experience as opposed to age.

The second issue that I have is the age used. If I read it right (poor assumption) the statistic used the average roster age. A more accurate age would be the team age waited by minutes played (though it is possible that this was used and I missed it.)

Overall… great work.

Great post and great new way to look at this stat..

But whenever I dig too far into these types of metrics.. I get disappointed.. why? Because they are based on data that doesn’t tell enough of the story to fully answer the question.. in this case.. I am left with realizing that measurement error in the ‘defensive efficiency’ stat really makes this stat questionable. By measurement error.. I mean that there is a difference between defensive efficiency (or rating) as measured by opponent’s PER and the actual latent (unmeasured) construct of defensive efficiency. There is no point is going too far into detail with this thought, because it reflects aspects of the game that are nearly impossible to track. A couple of obvious thoughts include how one team can play ‘good defense’ and still end up with an opponent with an amazing offensive performance and another team can play little to no ‘good defense’ and get lucky with a poor offensive performance. The difficulty with measuring defense is that we are really measuring results that are a mix of defensive effort/skill, offensive effort/skill, and some luck.. In some ideal world, I would love access to the stats coaches measure to gauge efficiency in terms of effort and not in terms of results.

My other issue with this stats relates to how age is used. We are taking a team stat and the aggregate of individual stats. Nothing wrong with that necessarily (although it does suggest a couple of other stat models that may be better to model this than a correlation (linear or otherwise)), but the impact of the standard deviation of that aggregate weighted age isn’t being considered at all. Does it make a big difference if you have two teams with identical weighted ages but one has 50% rookies and 50% aged veterans and the other one has 100% average age players? This, of course, has a ton of impact for the Blazers. Many fans believe they need a veteran presence now to even out the rookies but does it work that way? Trying to create a sort of regression equation to predict the impact of adding age to the team is a bit misleading because trading one rookie for one aged vet could have the same impact on that aggregate age difference as trading three rookies for three close to average age guys but.. likely perhaps.. an entirely different impact on the team defense..

http://myespn.go.com/blogs/truehoop/0-38-260/Evidence-that-Age-Makes-NBA-Players-Better-Defenders.html

Henry is focusing a bit on the question of why older players could be better defenders, especially if they get fewer foul calls.

But have we considered that coaches and GMs select for good defensive players?

If 100 players enter the league at age 21, and only 20 are good at defense, those 20 players would be selected by their teams to continue their career much longer than their poor-defending counterparts. That would explain the density of good defenders at higher ages.

I have no idea if this is accounted for in the data (I couldn’t find it if it was), but it would mean that older players are NOT better defenders, but simply that better defenders have the unique opportunity to become older players.

Between KP2’s analysis and Ben’s ability to present it in clear, readable form, this was a joy to read. Thanks for your hard work, fellas.

-sw