Hollinger can't be wrong...he can only be 'unluckily inaccurate'

[guava_wrench]

No matter how good a statistical analysis is, there will be outliers. If an analysis is 95% accurate, there are still the 5%.
 

[jackson_34]

Hollinger is just like an economist;

"Sure that's what happens in real life, but will it work in theory?"

[footey]

Indeed Proceed is spot on, pointing out that Hollinger just can't admit it when he is wrong, or rather when his numbers fail to correctly project results.  The fact of the matter is, this Celtic team was coasting, experimenting, getting healthy, doing anything but trying to win every stinking regular season game. We were confused because the 2008 team were animals in the regular season; since this team was mostly the same guys, I just assumed that they were still trying but could not get the job done. The incredible turn around, as soon as the play-offs started, proves me a fool (gladly).  Hollinger's ability to make predictions based on empirical data is completely undermined by these Celtics. His ego won't permit him to admit as much.

[wahz]

ah, statistics!

I read an article yesterday am and I scanned so many different articles that Im not even sure where it was from but the author concuded that since KG is 13-38 or something like that so far, its not him that has made a difference in the Orlando series.

The numbskull who wrote that apparently has never heard of defense and or thought KG might be the reason Rashard has disappeared off the face off the earth.

Ah, statistics. You see, the problem is picking and choosing the ones that you want to use and interpret because the reality is that whole of life is the outcome of every possible statistic all working with and inconjunction and oppossed to each other.

Here are some stats for you, my son said this yesterday: "so, 3-0?"

[Boris Badenov]

I don't see why anyone would view that article as fodder for Hollinger-bashing. He's just saying that what is happening right now is pretty unusual by historical standards, and that one cannot easily find a common factor shared by the other surprise finalists/championships (a factor, for example, clear enough to incorporate into a statistical model).

And let's keep one thing in mind: the failure here is not just of Hollinger's models. EVERY single member of the ESPN expert team picked the Celtics to lose BOTH series. It's not as though Hollinger's stats-based approach was telling us one thing, while the non-stats-based experts were all picking the Celtics to win based on their hunches. That would be a much stronger indictment of his model.

Bottom line: everyone was wrong, from the stat geeks to Vegas to Barkley. At least Hollinger has the good sense to go back and take a closer look at the historical record, to place things in perspective. I learned something from the article.

[wbones1]

Aren't all statistics based upon past data?

Hollinger extrapolates an outcome based upon past statistical evidence.

The human factor doesn't remain constant (like for the most part as numbers usually do).

The past Celtics are not the same Celtics.

Statistics start with a null hypothesis, and then you look at your collected data. That data is often used to justify, whatever you are looking to confirm. You have a 5% probability, in which you want your data to fall. If that data falls within the critical region or 5% probability, then one can draw a conclusion.
One can lower the threshold for confirmation by simply putting the critical region together, and not splitting it between the opposite ends of the spectrum. That will make it easier to reject your null hypothesis, which allows you to draw a conclusion. That is why people can argue about statistics.. Data can be made to confirm, what you are attempting to prove.

[Mr October]

Any "expert" who picked the Magic in series after the Celtics depants the Cavs during the last 3 games of the semis, should have their expert card revoked.

Just look at Boston and Orlando, and who they beat. I like seeing them pick the other team though. It keeps our guys extra hungry.

[TangRen]

I think this is a rather useless article.  Everyone already knew that the teams performing poorly in the regular season usually didn't make it to the conference finals, often.  That's why a lot of the "experts" picked against the Celtics.  Then, Hollinger proceeds to use second half record and point differential to separate this year's from the rest of the bunch.

Second-half record and point differential statistics are another form of statistics just like regular season records.  They are an estimation of true team strength.  They might be a valid estimation of strength.  They might not be a valid estimation of team strength.  In this case, Holliger does the really stupid thing of considering postseason seeding.  Seeding is a terrible stat to use.  Point differential is a much better stat to use, but it's meaningless for individual teams until you corroborate it by watching the actual games.

Frankly, I've read some of Holliger's stuff and his numerical methodology is usually pure garbage.  I wouldn't put a lot of weight on his conclusions, especially when it involves complicated stochastic events. (probablity)

Nonetheless, our takeaway is the same. The Celtics' ability to suddenly dial it up for the playoffs has been remarkable, but we shouldn't start expecting this as a normal occurrence. Out-of-the-blue conference champions come along about once per decade, and out-of-the-blue title teams appear with even less frequency. It would be a mistake to glean from this one example that those will now become annual events.

The closing line is pure garbage.  The only lesson is that tanking the regular season has never been really tried before.  It might set an example for future teams to emulate - to put a greater importance on getting healthy rather than getting good seeds.  Cars weren't that common before 1910, but there were lots of them a decade later, and they are all over the place now.

[Neurotic Guy]

It seems that the big complaint here is the presumption of an over-active ego and/or an all too transparent attempt at self-protection.   

However, I don't care how large Hollinger's ego is nor do I care if someone admits they are wrong or gloats when they are correct.  Hollinger is an entertainer and I am either entertained by him or not.  It sounds like IP is entertained by Hollinger -- if defined as 'gaining your interest'.

That said, Hollinger is nothing more than a handicapper.  Like any handicapper, he uses data to assess likelihood -- not an exact science (which I am certain he knows).  Many highly skilled handicappers analyze the Kentucky Derby each year -- determining the favorite as the horse deemed 'most likely' to win.  I believe that particular horse finished 6th this year.  Does that make them wrong? Depends --  in fact, they are probably correct about likelihood, but with sports you just can't account for all variables with precision.   Thus, handicappers like Hollinger may be correct in their analysis, but not necessarily correct about who they 'pick'. 

I think most 'experts' have the goal of being correct more often than they are incorrect.  And as I said in another thread, most of us use part head, part heart and part hunch to make our predictions -- and with that we should only hope to get it right about half the time.  No one has to mea culpa for a wrong prediction in my opinion -- we should all take it for granted that every sports guess is just that -- a guess.

[Fafnir]

Any "expert" who picked the Magic in series after the Celtics depants the Cavs during the last 3 games of the semis, should have their expert card revoked.

Just look at Boston and Orlando, and who they beat. I like seeing them pick the other team though. It keeps our guys extra hungry.

Why is that? Orlando depants the Bobcats/Hawks in a similar manner.

I don't have any hate to throw at experts who picked Orlando.

[tb727]

He's lousy, but not as lousy as Adande and Plaschke- those 2 are the bottom of the barrel.

I pay Hollinger as much attention as the gecko from the Geico commercials.

[Cman]

I don't see why anyone would view that article as fodder for Hollinger-bashing. He's just saying that what is happening right now is pretty unusual by historical standards, and that one cannot easily find a common factor shared by the other surprise finalists/championships (a factor, for example, clear enough to incorporate into a statistical model).

And let's keep one thing in mind: the failure here is not just of Hollinger's models. EVERY single member of the ESPN expert team picked the Celtics to lose BOTH series. It's not as though Hollinger's stats-based approach was telling us one thing, while the non-stats-based experts were all picking the Celtics to win based on their hunches. That would be a much stronger indictment of his model.

Bottom line: everyone was wrong, from the stat geeks to Vegas to Barkley. At least Hollinger has the good sense to go back and take a closer look at the historical record, to place things in perspective. I learned something from the article.

TP. This is a very good point that deserves to be repeated:
Just because Hollinger was wrong doesn't mean his method & statistics are bogus.  ALL the experts were wrong.  What matter is whether Hollinger is right more often than the other experts.  I'd love to see a careful analysis of that.

[kozlodoev]

Any "expert" who picked the Magic in series after the Celtics depants the Cavs during the last 3 games of the semis, should have their expert card revoked.

Just look at Boston and Orlando, and who they beat. I like seeing them pick the other team though. It keeps our guys extra hungry.

I don't want to jinx anyone, but we haven't beaten the Magic yet.

[More Banners]

We all know that playoff success after a fairly mediocre regular season is a statistical anomality.  No surprise there.

The article does seem to spend more time/effort confirming that the current C's are a statistical anomality than in explaining it, possibly because that would reduce the importance of statistical predictions in the first place, which is counter to Mr. H's interests.

One of many inherent problems with predicting the playoffs by regular season records is that matchups (largely a luck-of-the-draw factor) matter tremendously in the playoffs.  In fact, one could argue (based on matchups) that the strongest team isn't always the champ, but that perhaps another team could have dispatched the champ if they would've met for 7.  That's why his stats don't have all the answers; he might predict the strongest team against a generic opponent, but the theoretically best team surely doesn't always win.

[TangRen]

I'd say that Hollinger has zero credibility on probability.  I remember reading an article where he was discussing his probability estimates of record breaking careers by players, and he was introducing an error or an assumption in every half sentence.  At the end, I was completely agog at the con job he was pulling on the readers - except I think he actually believe his bull.

While simple arithmetic like number crunching the point differential for the regular season is great,  I wouldn't trust a word coming out of his mouth when it comes to probability.  Don't let him tell you what his newfangled stat means or at least ask for a second opinion on it.  I'd have to analyze his PER stat more to understand if it's good at all.

The aggregate stats are far more complicated than people think they are.  Credit where credit is due though.  These new stats are usually better the simple ones available from a box score.  But their improvement is usually more out of luck and from trial and error than from understanding of the game.

Edit:  I just looked at his PER formula.  It really looks like someone decided to make up a whole bunch of numbers mash them up together and create a value.  Without reading his book, I'll never be able to know if his method is good or not, but I'd guess that he arrived at his numbers by trial and error, and it only values offense and doesn't account for jacking up dumb dumb shots.  Good defense outside of steals don't matter.  At least he knows stuff like clogging the passing lanes, getting good position in defending the post, and not getting blown by on the dribble aren't accounted for.  It's the prototypical statistic of dubious value.  I doubt even he knows why his stats works for offensive contributions - so he won't know when they stop working.