APBRmetrics

[Mike G]

As Dean says in the ever-expanding "Isiah" thread, one topic has branched every which way.

Similarly, many of our topics come back to recurring themes. Since none seems to start and finish on this theme, I'll try to get one going.

Bob C wrote:

"people look at the 04-05 hawks and think walker and harrington are the two best players on the worst team. wrong - the hawks are the worst team because walker and harrington are getting so many minutes. they'd be far better off giving the bulk of those minutes to players like josh smith and josh childress... "

Well, Antoine's gotten big minutes from every coach he's played under (5 of 'em ), for whatever reason. I think one reason has to be his versatility. His production "resembles" a scaled-down Larry Bird.

Josh Smith looks like a PF but is listed as a SG -- ?? I guess that's where they can find him some minutes. He should be able to replace Harrington now, and be a star in the near future.

Childress shoots as badly as Walker, half as often. And he doesn't seem to do anything else as well.


Bob C wrote:
" ...manu ginobili, as of the stats of this past weekend, is clearly - CLEARLY - the best SG in the nba today.... "

Dan R wrote:
" ...Manu Ginobili is rated #7 overall this season and is the highest rated SG ..."

Kevin P wrote:
"I have Ginobili 10th in the league overall and second to Dwyane Wade amongst shooting guards. He, like Duncan, is hurt to some extent or another by the Spurs giving their starters so much rest."

If I'm not mistaken, all these rankings make the assumption that Ginobili's production in 30 minutes could be scaled up however one wishes, with identical results. Or, conversely, that his team is so good they'll never need his help beyond his prescribed 30 min.

If you've played ball, you know you can't keep up defensive intensity just as well, no matter how long you play. You also can't shoot when you're tired. These are fundamental limitations on human beings, even if they're not quantifiable in our spreadsheets.

Ginobili seems to be getting better by the week. He started the year strong, and has carried ever-bigger offensive loads for his team. He just might be the best SG. (I still have McGrady well ahead, and he's approaching the level of Kobe and Wade; 3 guys who play a lot more minutes.)

I've never seen a guy as good as Ginobili get so few minutes. On previous versions of the Spurs, he'd surely get more out of necessity. In the playoffs, if he has to go 40, I look for his numbers to drop (per-minute, and eff%).

So my position is that you can't assume a 20- or 30-MPG guy will do the same in an additional 10-20 minutes. Nor can you assume he won't. But Manu looks like (what I've seen) a guy who goes all-out, and is likely optimized right about where he is.


Hoiberg is another all-out type player. I think he expends too much energy just covering his ass on defense. This leaves him very little for offense, where his biggest obstacle is getting open. He might contribute more if his team was better defensively. I wonder if McHale can find him some minutes.

[Dan Rosenbaum]

But I thought that the attempts to look at this issue, most notably by John Hollinger, have found that players per minute averages usually increase as their minutes increase. Surely there is some limit, such as above 40 minutes per game, but I just don't know if there is good evidence for this often quoted complaint about projecting minutes.

[Mike G]

It may well be the case that a team finding itself out of the playoff chase can give some late-season minutes to guys who have had very few. And in such cases, (giving minutes to see the effect of giving minutes) may indicate that even marginal players do better with 20-30 minutes than with 5-15.

However, I don't know of any other instance when the independent variable would be "MPG". Usually, it's the other way around: when you play better, you get more minutes. Some guys actually thrive on 40 minutes or more. These are generally ego-driven types that are looking for a scoring title, etc. They may look at time on the bench as "wasted time".

Didn't several members of last year's Grizz have career-best per-minute marks? This would be the opposite effect, having minutes reduced not due to ineffectiveness, but by a coaching style.

I believe John H's study(s) looked at injuries to major players, and how others on the team increased their minutes and their per-minute stats. If an individual's rebound rate goes up when his team's rebound % goes down, I don't count that as necessarily an improved rate. If scoring rate goes up, but efficiency drops, that too might not be seen as "better" scoring.

In any event, such a study still doesn't differentiate between guys who "step up" and those who don't -- except that in the end, players who do the job get the minutes. Once again.

Surely no one needs to be directed to the garbage-time players who put up a point per minute but never can do that as a prime-timer. They are too numerous to mention. When a Redd or a Randolph busts out, it's more likely because they just improved their game, rather than that they finally got the minutes.

[bchaikin]

So my position is that you can't assume a 20- or 30-MPG guy will do the same in an additional 10-20 minutes. Nor can you assume he won't. But Manu looks like (what I've seen) a guy who goes all-out, and is likely optimized right about where he is.

well then let's not assume - why not just look at league history? where's the best examples of players going from few minutes to major minutes at the nba level? bench warmers replacing injured starters is a good place to start, but there are extenuating circumstances, i.e. bench players playing alongside established starters and the argument that they would help his numbers...

so to avoid that argument why not just go even further and look at the cases where a number of bench sitters become starters at the same time - like on expansion teams? there have been plenty in recent past decades...

as just one example remember when the minnesota timberwolves first came into the league in 1989-90, and a coach named musselman believed you should play your best players 40 min/g even if your team, err....stunk? take a good look at the per minute and per minute tendex or whatever type ratings you wish (other than per game) and compare the before (the expansion team) and after (on the expansion team) numbers. players like tony campbell and tyrone corbin and sam mitchell and pooh richardson and doug west went from playing non-starter minutes per game to major minutes in just a season or two....

i've done this for every expansion team since the 1960s and i can't find a definitive pattern one way or the other. some players play better with extended minutes, some worse, but the majority seem to play just about the same on a per minute basis, once you account for their touches/min being adjusted based on who their new teammates are...

so i don't have any problem projecting stats out over 40 min/g. is it realistic? possibly not - maybe 34-36 min/g would be better, as most starters tend to play these kind of per game minutes and not 40 min/g. but the simulation will show the maximum impact a specific player would have on a team if he plays maximum minutes, and so i use 40 min/g because that appears to be about the maxmum that starters will play consistently (a few play upwards of 42-44 min/g, but each season i think it's just a handful)...

But I thought that the attempts to look at this issue, most notably by John Hollinger, have found that players per minute averages usually increase as their minutes increase. Surely there is some limit, such as above 40 minutes per game, but I just don't know if there is good evidence for this often quoted complaint about projecting minutes.

again look at expansion teams, they appear to me to be the best statistical evidence for this scenario of major minutes for players not thought of at the time as deserving them...

If you've played ball, you know you can't keep up defensive intensity just as well, no matter how long you play. You also can't shoot when you're tired. These are fundamental limitations on human beings, even if they're not quantifiable in our spreadsheets.

imho nba players are probably the best conditioned athletes of the four major sports (hockey being close). there are your tractor traylors and danny fortsons of the league but i feel the vast majority of players are in tip-top shape and could play 36-40 min/g if they landed on an expansion team and had to...

there may be fundamental limitations on human beings, but if there are they didn't seem to affect those early minnesota t-wolves (and other expansion teams') players, whose stats i do have in my spreadsheets, unless the limitation you are speaking of is - winning....

Hoiberg is another all-out type player. I think he expends too much energy just covering his ass on defense. This leaves him very little for offense, where his biggest obstacle is getting open. He might contribute more if his team was better defensively. I wonder if McHale can find him some minutes.

there have been plenty of players in this league - even guards - with low touches/min like hoiberg that have played more min/g for their respective teams than he is playing now. unless you can prove to me that hoiberg is simply an atrocious defender that hurt his team critically if he played major minutes, i don't think there is any reason he couldn't play major minutes for the t-wolves or any other team...

i'm sure there were those people that said this same thing about brian cardinal last season, but in 03-04 he had no business playing behind the likes of clifford robinson, yet there he sat because the team that had him had no clue how valuable he was at that time...

Well, Antoine Walker's gotten big minutes from every coach he's played under (5 of 'em ), for whatever reason. I think one reason has to be his versatility. His production "resembles" a scaled-down Larry Bird.

he resembles a "scaled down" larry bird all right - like scaled down right to the bone...

larry bird's statistical production was one of the best the league had ever seen, and statistically he impacted more wins for his team during the time he played than all but less than a handful of players. antoine walker is worse than your average player in this league on a per minute basis, and in all seriousness has no business taking minutes away from better bench players - again unless you can prove to me he is a superb defender, which i do not believe he is (not even close imho)...

and for "whatever reason" is right - might these be the same coaches and GMs from around the league that have signed players like jon koncak and stanley roberts and jim mcilvaine and calvin booth and allan houston and jalen rose and numerous others to outrageous contracts - for "whatever reasons"? nba coaches and GMs are not nba statistical analysts as we all in this group claim to be, and if they were walker (and others) would not be getting big contracts and major minutes...

[Kevin Pelton]

[bchaikin]

Surely no one needs to be directed to the garbage-time players who put up a point per minute but never can do that as a prime-timer. They are too numerous to mention. When a Redd or a Randolph busts out, it's more likely because they just improved their game, rather than that they finally got the minutes.

in 0203, 0304, and 0405 zach randolph's pts/min were 0.50, 0.53, and 0.54, his rebs/min 0.26, 0.28, and 0.28, his fga/min were 0.40, 0.46, and 0.47, his fgm/min 0.20, 0.22, 0.20, his bs/min, 0.011, 0.013, and 0.010, his Scoring FG% 55%, 52%, and 50%. yet his min/g were 17, 38, and 35, and while his touches/min were 0.7, 1.0, and 1.0, his shots per touch have been right around 50% all three seasons. his turnovers per touch have been at 6%, 8%, and 7%. there was improvement in some areas, but for the most part his per minute numbers have been similar...

in 0102, 0203, 0304, and 0405 michael redd's pts/min have been 0.54, 0.54, 0.59, and 0.59, his ast/min 0.6, 0.5, 0.6, and 0.6, his fga/min 0.43, 0.42, 0.48, and 0.49, his fgm/min 0.21, 0.20, 0.21, 0.22, yet his min/g have been 21, 28, 37, 38. his touches/min have been 0.9, 0.8, 1.0, and 1.0, and he too has always shot the ball with right aroiund 50% of his touches. his turnovers per touch have been at 4%, 4%, 4%, and 5%. his Scoring FG% was at 57%-58% his first two seasons, and at 53%-54% his last two, so again there was improvement in some areas, he got worse in some other areas, and in several areas he's been the same player...

but in both cases their per game stats look much better later in their craeers because both went from playing bench minutes to playing starter's minutes...

[Dan Rosenbaum]

[CrazyFromTheHeat]

It seems to me that a vast amount of knowledge sitting in the play-by-play logs could help with this. It wouldn't matter much over a short period, but if you measured Dwyane Wade's production in five minute intervals over the course of the season, you might find something there.

As for player exhaustion, my impression has always been that players do get rested for being tired during the course of a game, but also for these reasons:

- the season is so long
- to avoid foul trouble
- the player has a nagging injury that's not enough to keep him out of the lineup.
- to keep bench players involved in case they are ever needed.
- if the game is a blowout either way.
- if the game is decided before the last couple of minutes.

and probably a host of other reasons. In other words, a lot of player rest is to prevent trouble, not necessarily because his performance is suffering.
_________________
The Best Miami Heat Coverage
http://heat.mostvaluablenetwork.com/

[Nikos]

[Ben]

[Mike G]

Dan R wrote:
"I guess the burden of proof is on you to start mentioning them.."

And before that, Bob C wrote:
"...look at the cases where a number of bench sitters become starters at the same time - like on expansion teams..."

Alright, I like the expansion hypothesis. Many guys will definitely get more minutes, through no merit of their own, but because there is a demand for more player-minutes throughout the league.

Between 1966 and 1971, the NBA grew from 9 teams to 16, and the ABA came about and took a good number of NBA players. On average, NBA players might expect roughly twice the demand on their services.

Of course, a guy already going 40 minutes can't pick up many more. But a number of 5-10 minute players became full-time or at least platoon players.

Just using NBA rosters, I find 668 player-seasons, of players who played (at least 400 min.) in 2 consecutive seasons. It's pretty straightforward to add their per-48 (pts + reb + ast) rates, and to determine whether a rise in minutes is accompanied by a rise in P+A+R.

This says nothing about causality, of course. Improved production should lead to more minutes, on average. But anyway, here's what I find:

326 total instances of players' MPG correlating positively with their productivity. This means increases, from one year to the next, of both minutes and per-36 PAR; or decreases in both.

342 Negative correlations: One measure went up while the other went down.

I'm amazed. Even with all the players you'd expect to get more minutes just by their own improvement, there were more players whose productivity dropped when receiving more minutes.

There may be fundamental differences in the NBA of '66-72 and today's league that make the situations hard to compare. But a league-wide expansion situation isn't replicable today. In that era, even established teams had minutes available, as their bench players were drafted away.

[Nikos]

I'm still a bit confused as to how Ginobilis win % could drop so low? Is their a spreadsheet or program/calculation within excel you have that you could post online to test this out Kevin? (just wondering)?

Duncan is playing less minutes this year too, does that mean extrapolating his minutes to 40mpg would make him worse (despite the fact he has had great seasons playing 40mpg as well)?

[kjb]

nikos: In my reading of the other Kevin's post, I'm not seeing where Ginobili's win% actually drops that low. I think Kevin is saying that to reduce Manu's WARP to his current level with 5 additional minutes per game, Manu's win% would have to drop below .500. I don't think Kevin is saying Manu's win% actually does drop that low.

[Nikos]

I guess I am just not familiar with that WARP? I thought he was saying his Win % would drop to .466 with 5mpg. It just seems a little extreme considering his win % right now. Maybe if you added 10mpg and increased his usage rate I could see a HUGE drop, but 5mpg? In a system where even Duncan is playing 5mpg less than he really should be or would be if healthy?

If it is not troubling, could you briefly explain what the .466 really is or means?

[jkubatko]

I did a small study using player-seasons from 1978-2004. To be included in the study, a player had to (a) see an increase of at least 50% in minutes per game from one season to the next and (b) play at least 41 games in each season. These criteria gave me 465 player-seasons. In 346 of these seasons (74.41%), the player's PER increased with an increase in playing time. The mean change in PER was 1.55 and the median was 1.58. The range of changes for the middle 50% was -0.05 to 3.18.

Because younger players are more likely to improve (and see an increase in playing time) than older players, I decided to add an age requirement to the criteria above. Now players had to be at least 30 years old in the target season to be inlcuded in the sample. The results were similar: in 42 of the 58 seasons (72.41%) the player's PER increased with an increase in playing time. The mean change in PER was 0.98 and the median was 1.19. The range of changes for the middle 50% was -0.21 to 2.32.
_________________
Regards,
Justin Kubatko
Basketball-Reference.com