Hollinger's funny math bumps the C's up to 2nd this week

[Tr1boy]

Hollinger is an idiot period. Don't care much about his calculation crap. I guess celtics winning it all last year and having only 2 losses this year means nothing.

Boston has already beat the cavs earlier this year, have a better record, have a winning streak of 14(compared to clevalnd losing to atlanta) and we are still second??

this is y i guess espn is fun having retards like steven smith and hollinger around

[nickagneta]

As I have said in other threads, I think Hollinger's whole system should be called into question as he is using a lineal equation that stresses one area that he saw was common amongst former world champions(point differential). Sorry, but, no kidding the champions usually had a good point differential?! Really, I would never have thought that. But a lineal equation stressing point differential to judge the strength of a team is absolutely, ridiculously flawed. A proper algorithmic equation would be very flawed as determining the importance of each variable would be extremely difficult.

Simply, algorithms used to determine tendencies of huge amounts of people are extremely advanced and are still not reliable when the number of the people the algorithm is being applied to gets smaller. To apply such an mathematical principle to 12 person groups in variable conditions is, at this current time in history, impossible.

But Hollinger isn't even using that type of math. He's using a lineal equation with fixed constants that he himself determined based on stats of past teams. But the game changes every decade. Everything is different and using those stats as if they came off of a fixed set of parameters is just plain silly.

I give no credence to his numbers and never have, even last year.

[Sweet17]

I don't think using point differential is all bad. Is it worse then record which is how teams are generally judged? Obviously linear equations are not perfect but there is no easy solution to this problem. It's like the BCS in football.

But several posters weren't impressed with Hollinger last year. I hate this kind of revisionistic nonsense. It's always things like "everyone hated Walker when he was a Celtic' or "everyone hated Jim Obrien's coaching" or perhaps "Everyone knew K. Brown was a bust." Just speak for yourself..

Hollinger kinda turned anti Celtic after some of his pets were traded (Jefferson, West) and didn't think they would do well..

[Brendan]

As I have said in other threads, I think Hollinger's whole system should be called into question as he is using a lineal equation that stresses one area that he saw was common amongst former world champions(point differential). Sorry, but, no kidding the champions usually had a good point differential?! Really, I would never have thought that. But a lineal equation stressing point differential to judge the strength of a team is absolutely, ridiculously flawed. A proper algorithmic equation would be very flawed as determining the importance of each variable would be extremely difficult.

Simply, algorithms used to determine tendencies of huge amounts of people are extremely advanced and are still not reliable when the number of the people the algorithm is being applied to gets smaller. To apply such an mathematical principle to 12 person groups in variable conditions is, at this current time in history, impossible.

But Hollinger isn't even using that type of math. He's using a lineal equation with fixed constants that he himself determined based on stats of past teams. But the game changes every decade. Everything is different and using those stats as if they came off of a fixed set of parameters is just plain silly.

I give no credence to his numbers and never have, even last year.

"I award you no points and we are no dumber for having read that"

Seriously - its a simple approach that he developed using standard techniques like regression and re-running previous seasons data to see how well it works. People take this way to personally - or expect perfection - neither makes sense.

[nickagneta]

I don't think using point differential is all bad. Is it worse then record which is how teams are generally judged? Obviously linear equations are not perfect but there is no easy solution to this problem. It's like the BCS in football.

But several posters weren't impressed with Hollinger last year. I hate this kind of revisionistic nonsense. It's always things like "everyone hated Walker when he was a Celtic' or "everyone hated Jim Obrien's coaching" or perhaps "Everyone knew K. Brown was a bust." Just speak for yourself..

Hollinger kinda turned anti Celtic after some of his pets were traded (Jefferson, West) and didn't think they would do well..

If your comments here are directed at me then perhaps you should peruse my past comments regarding Hollinger and his mathematics and find out exactly just how much of what I have to say is revisionist before accusing me of making false accusations just to prove a point.

http://forums.celticsblog.com/index.php?topic=17209.msg275760#msg275760

Check out what I think about Hollinger in this article I wrote in May. It's not last year but is last season. I have never liked Hollinger or his self contrived reworking of stats to make a basketball statement.

[LarBrd33]

point differential should be put in there, but I swear there is a bug in Hollinger's system.  How can the Cavs be ranked ahead of us if they have lost two more games, won three less games, LOST TO BOSTON and the Celtics have been able to beat the three other teams that beat them (Atlanta, Detroit, New Orleans)...

And even if it factors in "recent success"... we have a 15 game winning streak and they have a 1 game losing streak...

what?

huh?

Cuz they blew out the Thunder or something?  I don't get it.

[nickagneta]

As I have said in other threads, I think Hollinger's whole system should be called into question as he is using a lineal equation that stresses one area that he saw was common amongst former world champions(point differential). Sorry, but, no kidding the champions usually had a good point differential?! Really, I would never have thought that. But a lineal equation stressing point differential to judge the strength of a team is absolutely, ridiculously flawed. A proper algorithmic equation would be very flawed as determining the importance of each variable would be extremely difficult.

Simply, algorithms used to determine tendencies of huge amounts of people are extremely advanced and are still not reliable when the number of the people the algorithm is being applied to gets smaller. To apply such an mathematical principle to 12 person groups in variable conditions is, at this current time in history, impossible.

But Hollinger isn't even using that type of math. He's using a lineal equation with fixed constants that he himself determined based on stats of past teams. But the game changes every decade. Everything is different and using those stats as if they came off of a fixed set of parameters is just plain silly.

I give no credence to his numbers and never have, even last year.

"I award you no points and we are no dumber for having read that"

Seriously - its a simple approach that he developed using standard techniques like regression and re-running previous seasons data to see how well it works. People take this way to personally - or expect perfection - neither makes sense.

I don't take it personally or expect perfection. I just don't think his numbers or how he comes about them are not very good or correct. I may not always agree with everything you say, Brendan but I don't call your comments dumb, no matter how conservatively warped or blatantly wrong I think some of them may be.

[Roy Hobbs]

The problem (if anybody cares about this enough to be a problem) is that losses don't appear to weigh heavily enough.  That's why you have a four loss team healthily ahead of a two loss team.  One wonders what Hollinger's ratings would look like if, say, a team lost three games by one point and won a fourth by 63.  Would they be the #1 team because they have a +15 point differential?  Probably.

It's a flawed system, which would be fine, if Hollinger wasn't using it to predict the teams likelihood of winning 70 games or advancing in the playoffs.

[Kwhit10]

Well they beat the Jazz and their Hollinger ranking went down...

But we shouldn't be getting worked up at all.  These rankings have no say in how the games will work out.

[Sweet17]

If your comments here are directed at me then perhaps you should peruse my past comments regarding Hollinger and his mathematics and find out exactly just how much of what I have to say is revisionist before accusing me of making false accusations just to prove a point.

It wasn't really directed at you. One poster was saying no one criticized Hollinger last year when in fact some posters did. That's hardly a "false accusation."

[nickagneta]

If your comments here are directed at me then perhaps you should peruse my past comments regarding Hollinger and his mathematics and find out exactly just how much of what I have to say is revisionist before accusing me of making false accusations just to prove a point.

It wasn't really directed at you. One poster was saying no one criticized Hollinger last year when in fact some posters did. That's hardly a "false accusation."

Sorry, Pete, I misread your post. TP 4 straightening me out on that. Now that I have reread it and have a cup of coffee in me I see what you were saying.

[cordobes]

The problem (if anybody cares about this enough to be a problem) is that losses don't appear to weigh heavily enough.  That's why you have a four loss team healthily ahead of a two loss team.  One wonders what Hollinger's ratings would look like if, say, a team lost three games by one point and won a fourth by 63.  Would they be the #1 team because they have a +15 point differential?  Probably.

It's a flawed system, which would be fine, if Hollinger wasn't using it to predict the teams likelihood of winning 70 games or advancing in the playoffs.

Sure, I don't entirely disagree with your points, but what are the chances of that particular occurrence happening? And the formula doesn't have any kind of validity for such small sized samples.

Hollinger's ratings are based on plenty of empirical evidence that shows that scoring margin, adjusted with a few weights, is a better predictor of future success than the W/L record. Surely they don't have full predictive power - there are teams that can go unlucky and lose a historically bizarre amount of close games, for example; and there's lots of noise: injuries or teams running up the score, for example; but Hollinger reckons that (unlike many others sports staticians, Hollinger is relatively humble - albeit only relatively - about the explanatory power of stats). There are always statistical oddities, and in this case they are very prone to happen.

Personally, I don't even look at power rankings, including Hollinger's, but I don't see the problem with it. I think his power ranking formula is way more solid than many other of his linear weight formulas, like PER.

People should try to see them as what they are and understand what's Hollinger is exactly trying to measure. If they don't have any interest in what he's measuring - I do, but I don't check it because I already have an idea on how are the teams doing - just don't read them. Hollinger's method is flawed (I don't even like to call it Hollinger's method, because he didn't discover it at all), but it's the best objective method of predicting success, at least for now. "Best" and "objective" being key words.

Hollinger is predicting the likelihood of play-offs success because this is the best metric to do it; but it's only the likelihood. The same way I can predict the likelihood of not raining at the next June 7th as large, although sometimes it rains in June.

If you want a ranking that weights the W/L record, check Sagarin's ranking.