Why do computers hate the Mavs?

[noob cake]

So during the NCAA tournament, Sagarin, RPI, and Pomeroy (aka computerized rankings) pretty much raped every one.

So today I was checking our Sagarin and noticed how low the Mavs are rated. Then I remembered Trollinger's power ranking from a few weeks back; to my surprise, the Mavs are ranked even lower now.

Hollinger has Mavs at 15th
http://espn.go.com/nba/hollinger/powerrankings

Sagarin has Mavs at 12th
http://www.usatoday.com/sports/sagarin/nba0910.htm

:rolleyes
Anyone know why?

edit:
One more rating here, Mavs 15th
http://sites.google.com/site/computerratings/nba-basketball/ratings

[Findog]

point differential. Mavs don't have an impressive one.

[D-Wade #3]

It doesn't like chokers

[noob cake]

point differential. Mavs don't have an impressive one.

3rd best record in the NBA, 12+ on all three rankings.....

[mogrovejo]

3rd best record in the NBA, 12+ on all three rankings.....

"Point differential". How hard is this to understand? Not W/L record; "point differential".

[Shank]

The game is measured in wins and losses, is it not?

[mogrovejo]

The game is measured in wins and losses, is it not?

So?

[noob cake]

"Point differential". How hard is this to understand? Not W/L record; "point differential".

I understand, you dumbshits.

Point differential does not make up 99% of any ranking. Shit offensive or defensive efficiency probably has something to do with this.

[Mel_13]

3rd best record in the NBA, 12+ on all three rankings.....

You ask why, are given the answer, and then argue with answer?

Those computer rankings give great weight to point differential. Dallas has a point differential consistent with a middle of the pack team, so those computer rankings place Dallas in the middle of the pack. It's that simple.

That doesn't make Dallas a middle of the pack team, just that the results of their games produce a certain outcome from the computer ratings you referenced.

[mogrovejo]

I understand, you dumbshits.

Didn't look that way. He said "point differential" and you answered by mentioned the W/L record.


Point differential does not make up 99% of any ranking. Shit offensive or defensive efficiency probably has something to do with this.

You're clueless. The gap between Off. and def. efficiency is exactly the point differential (adjusted to the pace).

[Juanobili]

It doesn't like chokers

[cheguevara]

Computer's also don't like gonorrhea

[mavs>spurs2]

who really gives a damn? the playoffs will come down to wins and losses, not some shitty computerized formula. obviously the mavs have been coasting all year, we're old as hell afterall.

last night was the beginning of crunch time if we want that 2 seed, and you saw how we demolished a decent grizzlies squad that has given us problems in the past.

[Rummpd]

The computers are smart - Mavs are the Sac Kings of the 2000s talented but still missing something vital every year

[coyotes_geek]

I beleive the computer rankings put more weight on the recent games than they do on the early games. Dallas hasn't exactly been playing all that great recently, plus their schedule has been relatively soft. Looking at the mavs schedule, it looks like they've only beaten one playoff team since february. I'd think the computers notice that as well.

[Findog]

I beleive the computer rankings put more weight on the recent games than they do on the early games. Dallas hasn't exactly been playing all that great recently, plus their schedule has been relatively soft. Looking at the mavs schedule, it looks like they've only beaten one playoff team since february. I'd think the computers notice that as well.

How hard did you look?

Since the trade, they've beaten the Lakers, Magic, Hawks, Bobcats, Suns, Heat and Nuggets.

[coyotes_geek]

How hard did you look?

Since the trade, they've beaten the Lakers, Magic, Hawks, Bobcats, Suns, Heat and Nuggets.

Lakers, magic, hawks, suns and heat were all in february. I was only looking at March and April, i.e. "since February". I probably should have just said "march and april", so apologies for the confusion. I did miss the Bobcats, so there's them and the nuggets.

[21_Blessings]

Why do the Mavericks like sucking pantomime cock in the playoffs?

[sribb43]

Mavs will fail in the playoffs because there defense is garbage and they only play hard for 1 quarter per game

[Obstructed_View]

I understand, you dumbshits.

Point differential does not make up 99% of any ranking. Shit offensive or defensive efficiency probably has something to do with this.

Since you don't seem to want to accept the answer to your question, go look up stats from previous years, and you'll find that point differential is the best stat for predicting which team is going to be the best. The Mavs have been very clutch this year, which has resulted in a high percentage of close wins, but statistically it doesn't carry much weight.

[Shank]

So?

So what does it matter if someone runs numbers through a formula and thinks they're the 15th best team in the league? Wins and losses say they're hovering around being the 2nd best in the West right now. Point differential or not, wins are wins.

[mogrovejo]

So what does it matter if someone runs numbers through a formula and thinks they're the 15th best team in the league? Wins and losses say they're hovering around being the 2nd best in the West right now. Point differential or not, wins are wins.

Yeps. Wins are wins, point differential is point differential and reg. season point differential is a better predictor of playoffs success in the NBA than reg. season wins. It's not "running through a formula", it's like NBA history, scores you can check in the records.

What's so difficult to understand about this? For my money, I have no idea why there's such a confusion about this issue.

[Findog]

The Komputors sure do love them Spurs. Here's what winning teams are doing against other +.500 teams. I figured what playoff teams do against the Knicks and Nets is irrelevant, so here it is:

LA Lakers: 27-19 (+1.3 PD/.586), 16-6 Home (+6.22 PD/.727), 11-13 Road (-3.2 PD/.458). 9.42 points better at home than on the road.

Utah Jazz: 26-21 (+1.25 PD/.553), 18-5 Home (+7.86 PD/.782), 8-16 Road (-3.66 PD/.333), 11.52 points better at home than on the road.

Dallas Mavericks: 26-19 (+.24 PD/.577), 14-9 Home (+1.82 PD/.608), 12-10 Road (-1.4 PD/.545), 3.22 points better at home than on the road.

San Antonio Spurs: 21-23 (+1.54 PD/.477), 12-9 Home (+4.38 PD/.571), 9-14 Road (-1.04 PD/.391), 5.42 points better at home than on the road.

[timvpimp]

How hard did you look?

Since the trade, they've beaten the Lakers, Magic, Hawks, Bobcats, Suns, Heat and Nuggets.
still you can only get a negative answer when the question comes to a series in playoffs. Maybe a series against the Suns will be one battle the Mavs have a chance in, but there's no way the Mavs can surmount the likes of Thunder and Blazers both of whom dominated the Mavs harshly in regular season meetings not to mention Lakers and Nuggets.

[timvpimp]

Mavs will fail in the playoffs because there defense is garbage and they only play hard for 1 quarter per game
:tu

[Findog]

Utah Jazz: 11.52 points better at home than on the road

They're not winning a series w/o HCA, and they'll be a bitch to beat if they have it.

[ElNono]

point differential. Mavs don't have an impressive one.

Sagarin's ranking is actually a mix of point differential ranking (PURE POINTS) and w/l based ranking (ELO CHESS). It obviously takes into account SOS, SOO, etc.

As explained in the actual preface text of those rankings, ELO CHESS is the most politically correct method, while PURE POINTS is the better predictor of future games.

[mogrovejo]

The Komputors sure do love them Spurs. Here's what winning teams are doing against other +.500 teams. I figured what playoff teams do against the Knicks and Nets is irrelevant, so here it is:

LA Lakers: 27-19 (+1.3 PD/.586), 16-6 Home (+6.22 PD/.727), 11-13 Road (-3.2 PD/.458). 9.42 points better at home than on the road.

Utah Jazz: 26-21 (+1.25 PD/.553), 18-5 Home (+7.86 PD/.782), 8-16 Road (-3.66 PD/.333), 11.52 points better at home than on the road.

Dallas Mavericks: 26-19 (+.24 PD/.577), 14-9 Home (+1.82 PD/.608), 12-10 Road (-1.4 PD/.545), 3.22 points better at home than on the road.

San Antonio Spurs: 21-23 (+1.54 PD/.477), 12-9 Home (+4.38 PD/.571), 9-14 Road (-1.04 PD/.391), 5.42 points better at home than on the road.

The reason why they use Point differential and not "record against +.500 teams" or "point differential against +.500 teams" or "point differential against teams that are going to be in the playoffs adjusted to HCA" is not because nobody ever thought those could be better predictors.

Thinking that "what playoff teams do against the Knicks and Nets is irrelevant" isn't exactly innovative. But this isn't a subjective issue. To calculate predictive power one uses maths, doesn't pick what looks better just because.

[Findog]

The reason why they use Point differential and not "record against +.500 teams" or "point differential against +.500 teams" or "point differential against teams that are going to be in the playoffs adjusted to HCA" is not because nobody ever thought those could be better predictors.

Thinking that "what playoff teams do against the Knicks and Nets is irrelevant" isn't exactly innovative. But this isn't a subjective issue. To calculate predictive power one uses maths, doesn't pick what looks better just because.

Why is point differential league-wide considered a better predictive tool than when you reduce the sample size to just winning teams or playoff teams?

You're not going to see the Knicks or the Nets in the playoffs, you're not going to get caught napping against a bad team in the playoffs when you have to take your opponent seriously, and you're not going to be playing b2bs and four games in five nights. I don't know what sort of adjustment you come up with in regards to point differential as it pertains to the regular season when Team A was at home and had the previous two nights off, whereas Team B was on the 2nd night of a b2b and playing their fifth game in seven nights. It seems like there's all sorts of extraneous factors to account for.

[mogrovejo]

Why is point differential league-wide considered a better predictive tool than when you reduce the sample size to just winning teams or playoff teams?

Huh? Because it is? Why is 2+2=4?

This doesn't come from any philosophical or speculative process: just estimate the correlation between playoff success and various predictors - record, PD, record and PD in different samples, cumulative age, FT%, blowouts, etc. etc.

Then you'll see which one is a better predictor.

[mogrovejo]

The fact that something is counter-intuitive doesn't make it wrong.



As explained in the actual preface text of those rankings, ELO CHESS is the most politically correct method, while PURE POINTS is the better predictor of future games.

Maybe PD from games with playoff teams would be more politically correct. But it isn't the better predictor.

[Findog]

Huh? Because it is? Why is 2+2=4?

This doesn't come from any philosophical or speculative process: just estimate the correlation between playoff success and various predictors - record, PD, record and PD in different samples, cumulative age, FT%, blowouts, etc. etc.

Then you'll see which one is a better predictor.

It just seems counter-intuitive that games against lottery teams would be relevant. I know the data says otherwise.

[Findog]

If there's any hope for my Mavs (and personally I don't think there really is), the 2001 Lakers, 2006 Heat, 1995 Rockets and 1979 Sonics all won a 'chip with less than imposing PDs. I disregarded the 78 Bullets because of Walton's injury. Looking through the list of past championships, generally speaking you have to have a PD of 5 or up to be a title-worthy team.

[mogrovejo]

It just seems counter-intuitive that games against lottery teams would be relevant. I know the data says otherwise.

It seems. It increases the sample size, reducing the noise. Plus I've always heard that the mark of a good team is the ability to destroy bad teams. The data isn't static anyway; it changes every season, so maybe the better predictor today won't be the one for the future? It'd imply some kind of change in the way teams approach games. Never thought much about this.

[timvpimp]

If there's any hope for my Mavs (and personally I don't think there really is), the 2001 Lakers, 2006 Heat, 1995 Rockets and 1979 Sonics all won a 'chip with less than imposing PDs. I disregarded the 78 Bullets because of Walton's injury. Looking through the list of past championships, generally speaking you have to have a PD of 5 or up to be a title-worthy team.
I would scoop the nuts the fuck out of my scrotum if the Mavs made the century-miracle by winning the championship this season.

[mogrovejo]

If there's any hope for my Mavs (and personally I don't think there really is), the 2001 Lakers, 2006 Heat, 1995 Rockets and 1979 Sonics all won a 'chip with less than imposing PDs. I disregarded the 78 Bullets because of Walton's injury. Looking through the list of past championships, generally speaking you have to have a PD of 5 or up to be a title-worthy team.

Different ways to overcome a relatively small Pythagorean record:

1995 Rockets - would never win without Clyde who they only acquired at the trade deadline.

2001 Lakers - Fisher barely played in the reg. season; stepped up their defense big time for the playoffs

2006 Heat - much improved team after Pat Riley took over; carried the momentum to the playoffs

1979 Sonics - can't remember them; they were DJ's team, excellent defensive team - that probably explains part of it

Anyway, PD is just the better predictor, not a perfect one. I can perfectly see the Mavs winning it.

[Findog]

It seems. It increases the sample size, reducing the noise. Plus I've always heard that the mark of a good team is the ability to destroy bad teams. The data isn't static anyway; it changes every season, so maybe the better predictor today won't be the one for the future? It'd imply some kind of change in the way teams approach games. Never thought much about this.

That makes sense.

[Findog]

Different ways to overcome a relatively small Pythagorean record:

1995 Rockets - would never win without Clyde who they only acquired at the trade deadline.

2001 Lakers - Fisher barely played in the reg. season; stepped up their defense big time for the playoffs

2006 Heat - much improved team after Pat Riley took over; carried the momentum to the playoffs

1979 Sonics - can't remember them; they were DJ's team, excellent defensive team - that probably explains part of it

Anyway, PD is just the better predictor, not a perfect one. I can perfectly see the Mavs winning it.

I would theorize that a less than gaudy PD for a veteran team with a bunch of 30+ guys is less worrisome for your title chances than a young team going all out that can't overwhelm teams. A team like San Antonio or Dallas has to manage minutes very carefully over the course of an 82-game season than a team like the Zombie Sonics trying to establish themselves.

[ElNono]

The fact that something is counter-intuitive doesn't make it wrong. Maybe PD from games with playoff teams would be more politically correct. But it isn't the better predictor.

You're missing the point. What's being said there is that PD has statistically been a better predictor of future games than W/L record.
Most of all of these rankings adjust for strength of schedule, strength of opponent, etc.

[ElNono]

I would theorize that a less than gaudy PD for a veteran team with a bunch of 30+ guys is less worrisome for your title chances than a young team going all out that can't overwhelm teams. A team like San Antonio or Dallas has to manage minutes very carefully over the course of an 82-game season than a team like the Zombie Sonics trying to establish themselves.

This is correct, and affect statistics in more than one way. Duncan and Ginobili always had their minutes increased in the playoffs.
Also somewhat explains the 2006 Heat, with a Shaq that sleepwalks in the regular season, and stood out in the playoffs.

[mogrovejo]

You're missing the point. What's being said there is that PD has statistically been a better predictor of future games than W/L record.
Most of all of these rankings adjust for strength of schedule, strength of opponent, etc.

How am I missing the point? Have you read my posts?

[duhoh]

Didn't look that way. He said "point differential" and you answered by mentioned the W/L record.



You're clueless. The gap between Off. and def. efficiency is exactly the point differential (adjusted to the pace).

look at his name :lol

[nkdlunch]

LMAO at the title of this thread

[Ghazi]

Regarding the '06 Heat, point differential couldn't anticipate D-Whistle getting gifted 97 free throws. Even in defeat the Mavs appeared to be a better team. Mavs and Spurs were the best teams in the league that year :).

[ElNono]

How am I missing the point? Have you read my posts?

Yes, I was responding to this:

Maybe PD from games with playoff teams would be more politically correct. But it isn't the better predictor.

The text asserts that PD is normally a better predictor than W/L (nothing less, nothing more). We don't really know wether PD adjusted for SOO is not a better predictor than PD not adjusted for SOO. As a matter of fact, most of these rankings do adjust for SOO, which would point to it being a better predictor.