No Time to Play Tin Soldiers with the Starting Unit...and Other Observations

This post is redundant with one I made on the "Slippery Slope" thread. Sorry about that, but I sort of wanted to highlight it because I think it may be worthy of some debate.

I've had a love-hate relationship with the +/- stat since the 60s, when I used to keep it myself during games. (What a nightmare to do it manually.) I love the stat in looking at player combinations and hate the stat when it comes to evaluating individuals.

In a mood to explore the more "romantic" aspects of the game, I just accessed 82games.com, which (among other things) presents +/- stats for the five-player combinations playing the most minutes for each team. It's early in the season, and the sample sizes of minutes are not large, so these stats should be viewed with caution. But one has to start somewhere.

Following are the number of points each listed combination of players gains or loses versus opponents per minute of time on the floor:

Rondo-Daniels-Pierce-Williams-Wallace......+1.31 (points gained per minute)
House-R.Allen-Daniels-Williams-Wallace.....+0.55 (points gained per minute)
Rondo-R.Allen-Pierce-Garnett-Perkins........+0.20 (points gained per minute)
Rondo-Daniels-Pierce-Garnett-Perkins........+0.03 (points gained per minute)
Rondo-R.Allen-Pierce-Garnett-Wallace.........-0.19 (points lost per minute)
House-Daniels-Pierce-Williams-Wallace........-0.31 (points lost per minute)
Rondo-Daniels-Pierce-Garnett-Wallace.........-0.35 (points lost per minute)

Despite the small sample sizes, there are some interesting general indications:

• Only three of the combinations produce anything approaching a significant net gain in points over the competition. (Actually there are three "garbage time units" that had modest plus figures, but I didn't report them because of relatively tiny minutes and the misleading effects of garbage time minutes.)

• The top two combinations include Daniels, Williams and Wallace. This fact may be overly influenced by the "euphoric" first few games, but the combination intuitively makes sense: slashing, rebounding, spacing/perimeter threat.

• Integrating Daniels with four starters (replacing Ray) seems to create a unit that basically stays level with the competition.

• Integrating Wallace and Daniels with three starters appears less productive.

• One gets the feeling (more of a hunch at this early juncture) that Sheed can be a particularly influential factor (pro or con) in the fortunes of whatever combination he's playing with; and it may take time for Doc to determine how to employ his services most judiciously.

• So far, the starter playing the most minutes with the bench has been Pierce, and that doesn't necessarily make for great results.

• The starters rank third among the combinations. Generally speaking (and based on a larger sample than for the rest of the combinations), it seems that the starters average a one-point advantage over opponents roughly every five minutes on the floor. I'd love to compare this stat with that of the starters in 2007-08, but I can't find the appropriate archive on 82games.com. However, if one looks at the current Celtics starting five versus the five-man combinations currently playing the most minutes for other selected teams, the Celts are in the middle of the pack:

Orlando...............+1.15
Indiana...............+0.73
Portland..............+0.42
Boston................+0.20
Atlanta................+0.16
L.A. Lakers..........+0.15 (without Gasol)
San Antonio.........+0.10
Cleveland.............-0.04

I've seen suggestions that the Celtics start monkeying around with the starting lineup. But this is no time to play "toy soldiers" and experimenting with the team's strength. It's time to make that strength more imposing because, without more difference-making from the five of KG/Perk/Pierce/Ray/Rondo, how far does one realistically think the Celtics will go this year? Among other things, elevating Daniels, Williams or Sheed to the starting lineup could negate the contributions of some other fairly productive combinations.

I'll finish with my earlier caveat. Small sample sizes to date. But it could be revealing to keep track of this stat over time.

Sam

Sam,

Another explanation for the effectiveness of the "Small 3" (Daniels, Williams and Sheed) could be that they're often playing against the other teams' bench. Two of those three were starters on other teams (and the third was a high draft pick out of a good college program) and know what it takes to play against higher quality talent. Weaker competition, better results. Would we see the same if we sat KG, Perk and either RA or PP and started Daniels, Williams and Sheed in their stead?

Nevertheless, I think your point is well-taken. The whole starter vs bench issue is one that should become more blurred as the players become more role-identified and offensive-and-defensive-set integrated. As the season goes on, we should be able to experiment more without it being much of an experiment at all.

Let's hope Doc sees it that way too.

bob

Bob,

I agree completely. We're talking about how starts do vs. other starters and the same for the bench. So far, it appears (in general) that the latter pairing is more to the Celtics' advantage than the former. And that bench plurality could well be the key to a championship...but only if the team continues to have a reliable starting unit as its centerpiece. The Celts won't win if the bench improves a lot and the starters go in the opposite direction.

As usual, I'd support more situational substitution, for example with selected bench players (my vote would go to Daniels first and Williams second) joining the starters sooner than they've heretofore been on the floor.

Sam

Thanks for the explanation Sam. I'm a natural born skeptic, so I'm uneasy about trying to capture the worth of a player with one number and I'm guessing that it may make you feel similarly but it is interesting, no doubt, and definitely of some worth.

Sam,
Thanks for the stats. I just hope Doc provides us with more samples of lineups with vets and younger players on the floor together, so we can see what works and what doesn't.
I have to disagree that now is not to the time to experiment with the team's strength (or weaknesses). IMO, there will never be a better time in this campaign to experiment. Improvise in November and capitalize the rest of the year.
KG appears to be tentative, as you note elsewhere, and that's only sensible. If he lands wrong on that leg, it could be bad. But why count on him for 30 minutes while he's making his way back? For me, it's enough to keep rehabbing, keep practicing, keep his minutes to 20 or so and let Williams play 30. Let PP and RA rest more, Daniels, Hudson and Giddens play more. If the vets need more rest and the young players more PT, I'd love to see it happen.
When I think of our vets, I think of that old saw that ``I'm not as good as I once was, but I'm as good once as I ever was.''
Let's not use our `once' up until we absolutely need it.
Tom

Sam,

Interesting numbers.

I think we're on the same page overall on this topic though;

Regardless of the stats on combinations, I can't see why toying with the starting unit would make any sense. The coaching staff puts that combo out on the floor first for many reasons.

I do think that of the bench rotation, Williams is making a strong case for more minutes due to his rebounding, especially at the offensive end.

Regards

Upforglory,

I may not have made myself clear. I never said now is not the time to experiment with the team's strengths (plural) and weaknesses (plural). I singled out only the starting unit as a core asset of the team on which I believe it's entirely premature to consider experimenting by diluting the bench while introducing new chemistry issues to the list of people who start the game.

One of the reasons I used the "tin soldiers" analogy is that I don't believe they just drop a player into a slot and have the unit magically perform better. There's usually a major learning period (often substantial), during which the performances of the new player's teammates may suffer so that people now blame them and want to experiment with them (a sort of chain effect).

I could reel of a number of experiments I'd like to see. In fact, yesterday, I mentioned one on the board—a sort of "swat team" consisting of KG, Williams, Daniels, Ray and Rondo—combining rebounding power, floor spacing and perimeter shooting, slashing and strength in the paint, and stiff defense. This would be a unit I'd consider inserting when things are falling apart—a sort of roster Gatorade.

Note that I didn't say they shouldn't SUBSTITUTE for a starter earlier in the game than they've been doing. I personally would like to see Daniels come in for Ray earlier in the first quarter, with Ray being saved to spend more time with the second unit. I just feel that replacing someone in the starting lineup right now would represent an experiment that, at the very best, would not pay immediate dividends and, at the very worst, could mess up both the starting and bench units big time.

Long-term, if the starting unit of KG, Perk, Pierce, Ray and Rondo consistently fails to set the tone of the game from the outset, the Celtics are cooked.

Sam