+/- interesting stats so far in the season ( J.Sullinger )

[vgulab]

I was looking in the +/- stats and realize something interesting, see your self...

best 5 player combination for Celtics

K.Garnett, J.Terry, J.Green, C.Lee, J.Sullinger  +42  60min

best 4 player combination

K.Garnett, J.Green, C.Lee, J.Sullinger   +74   159min

best 3 player combination

K.Garnett, C.Lee, J.Sullinger    +81   217min

best 2 player combination

K.Garnett, J.Sullinger   +101   283min

best 1 player

J.Sullinger    +68    670min

So J.Sullinger is in all best combinations. Our best +/- player and when KG and Sully are together they are +101 and second best is +55. This is another pointer how good Sully is and he is getting better every day. Sullinger and Bradley my favorite Celtics, after KG ofcourse. Go Celtics

[Snakehead]

Interesting stats.

Sully has to be starting IMO.

[Fafnir]

Sample size for +/- to be meaningful must be much larger than what we have right now.

[vgulab]

I'm not saying this stats are all that matters and that we can learn everything about our team but one thing is very clear to me. When KG and Sully are together on the court this team is much much better.

[lightspeed5]

Sample size for +/- to be meaningful must be much larger than what we have right now.

40 games isnt meaningful?

[Fafnir]

Sample size for +/- to be meaningful must be much larger than what we have right now.

40 games isnt meaningful?

For +/- (and its better cousin adjusted +/-) your sample size of minutes needs to be pretty large, entire seasons and more large frequently. Thousands and thousands of minutes worth of data.

Otherwise its subject to so much noise to render it completely unreliable.

Unadjusted +/- is already unreliable no matter the sample size due to teammate and lineup effects.

[Boris Badenov]

Interesting stats.

Sully has to be starting IMO.

But his best lineups are all with other bench guys (Lee, Green, Terry). Doesn't that mean we want to keep him with those guys, if Terry/Lee/Green play better with him on the court than without?

They're not mutually exclusive, of course, but I think the best role for Sully is off the bench, but at around 30mpg.

[Chris]

Interesting stats.

Sully has to be starting IMO.

The only reason Sully isn't starting is because Bass currently needs to play as well, and by playing him with the starters, you can likely get more out of him, while Sully seems to be able to play well, regardless whether he is starting or coming off the bench. 

[pearljammer10]

Interesting. Ive said all along that Sully should be starting alongside KG and that Bass fits better with the second unit. But Doc has seemed to find some sort of rotation that is working so I dont want to mess with what he's got going on.

[2short]

Interesting stats.

Sully has to be starting IMO.

The only reason Sully isn't starting is because Bass currently needs to play as well, and by playing him with the starters, you can likely get more out of him, while Sully seems to be able to play well, regardless whether he is starting or coming off the bench.

agreed
to me if bass can play good defense and rebound then he doesn't need to score 10 its

[Boris Badenov]

Sample size for +/- to be meaningful must be much larger than what we have right now.

40 games isnt meaningful?

For +/- (and its better cousin adjusted +/-) your sample size of minutes needs to be pretty large, entire seasons and more large frequently. Thousands and thousands of minutes worth of data.

Otherwise its subject to so much noise to render it completely unreliable.

Unadjusted +/- is already unreliable no matter the sample size due to teammate and lineup effects.

While you're right in principle about small samples, it's unlikely that the randomness you're talking about is driving the overall pattern of all lineups, which shows that nearly every good lineup has Sully in it.

http://www.nba.com/statistics/plusminus/plusminus_sort.jsp?pcomb=5&season=22012&split=9&team=Celtics

(Think about the difference between the odds of getting heads on one coin flip, and getting heads on 17 or more out of 20.)

Also, I'm not sure what is the basis for your claim that you need "thousands" of minutes. The sample size necessary depends on the hypothesis one is testing.

[kozlodoev]

Interesting. Ive said all along that Sully should be starting alongside KG and that Bass fits better with the second unit. But Doc has seemed to find some sort of rotation that is working so I dont want to mess with what he's got going on.

Why, just to make sure we've got prolonged stretches of time with neither Garnett nor Sullinger on the floor? Not the greatest rotation plan.

[Roy H.]

I agree with Faf that it's too small of a sample.  However, it's definitely encouraging, and Sully should definitely be starting.

[BballTim]

Sample size for +/- to be meaningful must be much larger than what we have right now.

40 games isnt meaningful?

  Those pairings are for 5 or so minutes a game. It's probably something like near the end of the 1st and 3rd quarter, where KG comes back in and they're playing against the other team's backups.

[vgulab]

Sample size for +/- to be meaningful must be much larger than what we have right now.

40 games isnt meaningful?

  Those pairings are for 5 or so minutes a game. It's probably something like near the end of the 1st and 3rd quarter, where KG comes back in and they're playing against the other team's backups.

still the second best 2 player combination behind KG and JS is +55 and it's almost double less than +101 for KG+JS