Relative productivity of "big" and "small" lineups

[Kedsy]

This seems to be somewhat of a hot topic, so I did some basic math, using pfrduke's +/- calculations. Since the debate started as a discussion of Kyle's role on the team, I calculated our +/- in three groups -- two bigs plus Kyle; two bigs without Kyle; and one big plus Kyle. Here's what it looks like so far:

PRINCETON
-----------
2 bigs + Kyle: 10 minutes; +22 (2.2 plus differential per minute)
2 bigs, no Kyle: 14 minutes; +6 (0.429 pdpm)
1 big + Kyle: 16 minutes; +9 (0.563 pdpm)

MIAMI (OH)
-----------
2 bigs + Kyle: 16 minutes; +9 (0.563 plus differential per minute)
2 bigs, no Kyle: 11 minutes; -3 (-0.273 pdpm)
1 big + Kyle: 13 minutes; +28 (2.154 pdpm)


TOTAL (2 games)
----------------
2 bigs + Kyle: 26 minutes; +31 (1.192 plus differential per minute)
2 bigs, no Kyle: 25 minutes; +3 (0.120 pdpm)
1 big + Kyle: 29 minutes; +37 (1.276 pdpm)


So, after two games, all three configurations have been used for roughly the same amount of time. And the two configurations containing Kyle have had roughly the same aggregate success. It shouldn't be a surprise that our team isn't as good when Kyle isn't on the floor.

Of course, this doesn't take into account other factors. For example, Kyrie almost never played in a "2 bigs, no Kyle" configuration against Miami, which probably has something to do with the poor performance of that configuration, as does the fact that it was that configuration that played the last 6 minutes of the game while we appeared to be coasting. But it's a start. If people like this analysis, I can continue to update it after every game.

[COYS]

This seems to be somewhat of a hot topic, so I did some basic math, using pfrduke's +/- calculations. Since the debate started as a discussion of Kyle's role on the team, I calculated our +/- in three groups -- two bigs plus Kyle; two bigs without Kyle; and one big plus Kyle. Here's what it looks like so far:

PRINCETON
-----------
2 bigs + Kyle: 10 minutes; +22 (2.2 plus differential per minute)
2 bigs, no Kyle: 14 minutes; +6 (0.429 pdpm)
1 big + Kyle: 16 minutes; +9 (0.563 pdpm)

MIAMI (OH)
-----------
2 bigs + Kyle: 16 minutes; +9 (0.563 plus differential per minute)
2 bigs, no Kyle: 11 minutes; -3 (-0.273 pdpm)
1 big + Kyle: 13 minutes; +28 (2.154 pdpm)


TOTAL (2 games)
----------------
2 bigs + Kyle: 26 minutes; +31 (1.192 plus differential per minute)
2 bigs, no Kyle: 25 minutes; +3 (0.120 pdpm)
1 big + Kyle: 29 minutes; +37 (1.276 pdpm)


So, after two games, all three configurations have been used for roughly the same amount of time. And the two configurations containing Kyle have had roughly the same aggregate success. It shouldn't be a surprise that our team isn't as good when Kyle isn't on the floor.

Of course, this doesn't take into account other factors. For example, Kyrie almost never played in a "2 bigs, no Kyle" configuration against Miami, which probably has something to do with the poor performance of that configuration, as does the fact that it was that configuration that played the last 6 minutes of the game while we appeared to be coasting. But it's a start. If people like this analysis, I can continue to update it after every game.

Hmm, applying actual data to a hotly contested debate . . . that's no fun!

All kidding aside, I think this is excellent. By the time we get into the meat of the ACC schedule, these cumulative stats will really start to paint an accurate picture of what lineup is most effective. Early results indicate that the most accurate conclusion so far is that any lineup with Kyle = good while most lineups without Kyle are not as effective. I will be curious to see if this stat may change a bit as a lineup with Andre plus the starters (minus Singler, of course) plays a bit more together in competitive games. My hope is that such a lineup won't suffer as much as one might think so that Kyle can get some rest or sit because of foul trouble and the team can continue to function at a high level. More possessions equals more fouls per game for many of the players and so far Kyle has been the one most affected by this.

[MChambers]

It would be great if you could continue to update it.

[airowe]

I appreciate the work Kedsy but I caution people from drawing too many conclusions from this data because its insufficient. Claiming one side is right or wrong based on this data you'd have to rely on the assumption that every opponent we play is the same exact height, has the same talent distribution, and is playing the same style of defense.

Why was the bigger lineup more effective vs. Princeton? Maybe it had something to do with the Tigers having eight players over 6'8". Or maybe because they packed it in ona 2-3 zone?

How about Miami? Maybe the Redhawks only playing one guy over 6'8" impacted the smaller lineup's effectiveness?

I'm going to predict that this all ends up in a wash but that both sides of the argument claim victory...

[Kedsy]

I appreciate the work Kedsy but I caution people from drawing too many conclusions from this data because its insufficient. Claiming one side is right or wrong based on this data you'd have to rely on the assumption that every opponent we play is the same exact height, has the same talent distribution, and is playing the same style of defense.

Why was the bigger lineup more effective vs. Princeton? Maybe it had something to do with the Tigers having eight players over 6'8". Or maybe because they packed it in ona 2-3 zone?

How about Miami? Maybe the Redhawks only playing one guy over 6'8" impacted the smaller lineup's effectiveness?

I'm going to predict that this all ends up in a wash but that both sides of the argument claim victory...

I agree with your prediction airowe, but once we get more datapoints it will hopefully smooth over the one-game anomalies. And if not, at least we'll have some numbers to cite while we're arguing.

[airowe]

At least we'll have some numbers to cite while we're arguing.

That would make it worth it right there. Carry on!

BTW, I'll stake my claim in this debate now. Coach K will play whatever lineup he feels gives his team the best chance to win. We might disagree with it and not understand his rationale, but it will always be his opinion and more often than not, the right decision.

[Jderf]

BTW, I'll stake my claim in this debate now. Coach K will play whatever lineup he feels gives his team the best chance to win. We might disagree with it and not understand his rationale, but it will always be his opinion and more often than not, the right decision.

Ahh, come on, Airowe. That's no fun! It's like going into a scientific controversy saying, "I'll take whatever side Nature ends up on." You can't wait and pick your horse after the race is over.

[airowe]

Ahh, come on, Airowe. That's no fun! It's like going into a scientific controversy saying, "I'll take whatever side Nature ends up on." You can't wait and pick your horse after the race is over.

In this instance, I'd rather be accused of being unfair than of being wrong.

[bluedevil2012]

That would make it worth it right there. Carry on!

BTW, I'll stake my claim in this debate now. Coach K will play whatever lineup he feels gives his team the best chance to win. We might disagree with it and not understand his rationale, but it will always be his opinion and more often than not, the right decision.

Well now you're starting to sound like John Fox, airowe.

[hq2]

Given the versatility and depth of this team, I don't think K will stick with one configuration or the other; he'll simply use whatever the matchups and/or hot hand calls for. What he's trying to do now is not only discover which combinations are best, but use lots of different ones so he can have different ones ready when he needs them. I would guess he could use as many as four or five different starting lineups, depending on matchups and how individuals are playing.

[Kedsy]

Eh, there are ALWAYS games (at least once or twice a year) when nobody can seemingly buy a basket. In those games it should be interesting to see what we do.

It shouldn't matter.Duke really has too much talent simply to be a sharpshooting team from the perimeter...with Irving and the bigs supposedly being better, we should be able to get a lot of good looks inside as well. Assuming we improve on that.

I agree on your latter point. As far as games when nobody can shoot, it does usually seem like we have one or two games like that every year. But I'm not sure we've ever had as many high-quality perimeter threats as this year. Not even counting Ryan (because I'm not sure if he's really a good outside shooter or just a decent shooter for a big man), we have five guys who should take a lot of three-point shots and could hit 40%+. Last year we had four guys in the 38% to 40% range; the year before that we had two (and, yes, those two combined to go 3 for 16 from 3-point range in the loss to Villanova). Not saying it won't happen, but unless it's snowing or something inside the gym it's unlikely all five of this year's sharpshooters will go cold on the same day.

Given the versatility and depth of this team, I don't think K will stick with one configuration or the other; he'll simply use whatever the matchups and/or hot hand calls for. What he's trying to do now is not only discover which combinations are best, but use lots of different ones so he can have different ones ready when he needs them. I would guess he could use as many as four or five different starting lineups, depending on matchups and how individuals are playing.

Well, I suppose four or five starting lineups is possible, but I don't think it's all that likely. On the other hand, I think you're spot on when it comes to minute distribution. The configurations that are working that day are going to be on the court more frequently.

And I think you're right about touting our versatility and depth. I expect K to vary this up a lot during games because it will keep our opponents off balance. They won't be able to get comfortable with so many different looks and styles at K's disposal. E.g., the opposing 6'6" SF is trying to get himself in the mindset to stop a bigger Kyle and all of a sudden he has to guard Andre or Seth; an opposing PF is trying to jump screens and guard the alley-oop and suddenly he has to chase Kyle around the perimeter; an opposing SG gets comfortable laying back to stop Nolan's drive and all of a sudden he has Seth or Andre taking threes from NBA range. Game preparation for opposing coaches will be a nightmare.

[tele]

Not to complicate things, but here I go anyway, it is interesting to see what lineups do well, especially early in the season when things are getting tried out and minutes spread around. However, to me, it is more telling to see what lineups get used when the game is in the balance, when your team has to have a bucket or a stop.

I'm also waiting to see the 4 guard lineup play, that may have to wait until Duke has the lead and is letting the air out of the ball at then end of a game. Kyle, Kyrie, Nolan, Seth, and Andre could also put some points up on a zone or apply a full court press. May not see it, may not need to use it.

[Newton_14]

Great thread Kedsy. I just caution folks to not get so hung up on debating the Big/Small/Medium lineups that we miss the wonderful ride that started this past Sunday.
This team is an absolute dream to follow and root for.

Last year K was so limited in what he could do. Outside of rotating the bigs, there was just nothing there to change. This year he has as diverse team as he has had in years. Since 1998 in my view. He can throw so many different looks at opponents it almost isn't fair.

Regarding the debate, I count Mason/Miles/Ryan/Josh as the true "bigs" and that is looking at both size and skillset. Kyle is the hybrid 3/4, who I do think will spend more time at the 3 but still see good minutes at the 4 and a few more than I had originally anticipated.

Andre at a big 6'4 is a solid hybrid SG/SF. Nolan, Kyrie, Seth are the shrimps. My main point of contention in this spirited debate was that we would try to play the majority of the time with 2 of the true bigs on the floor, and we would not play the majority of the time with the Kyrie/Seth/Nolan/Kyle/1 True Big, lineup that many were pining for. I do feel we will occasionally see that lineup, but only when we are dictating terms, not reacting, and not for large stretches. So far, we actually have not seen that lineup very much. The lineups with Andre at the 3 and Kyle at the 4 have been very effective so far though. To me that is one of our "medium" lineups.


But again, lets not get so hung up on any of those points that we miss what is happening. This team has great potential. Andre's play has been a blessing to me. I wanted so bad to see him succeed after what he went through last year. So far he has not disappointed. His play has been well beyond what I "feared". Kudos to him for raising his game and not letting the presence of Seth and the future presence of Austin cause him to hang his head. He is staking his claim at minutes.

Still need Miles to get going, and I still feel he will. Mason is much improved as is Kelly.
Tyler is proving he can handle the pressure as well. Josh at this point, has further to go imo, than the rest, but even he has shown glimpses and there is upside there.


Based on the scrimmage and the 4 games I have seen thus far, come conference time I expect we will be very strong 1 thru 8, with a solid 8 man rotation, with Tyler's minutes going down a little bit, and Hairston's minutes resembling Kelly's from last year.

I have my Colgate and Mich St tickets in hand and cannot wait to see the boys in action again.

[Kedsy]

Not too much you can glean from a 50 point win, but once again the small lineup was by far the most effective (2 out of 3 games now) and the "middle" (no Kyle) lineup was by far the least effective (all three games).

Note that the big lineup was best against the biggest team we played (Princeton) and the small lineup was best against the two smaller teams. It will be interesting to see what happens when we play BCS teams with legitimate size.

COLGATE
-----------
2 bigs + Kyle: 11 minutes; +15 (1.363 plus differential per minute)
2 bigs, no Kyle: 18 minutes; +12 (0.667 pdpm)
1 big + Kyle: 11 minutes; +25 (2.273 pdpm)


TOTAL (3 games)
----------------
2 bigs + Kyle: 37 minutes; +46 (1.243 plus differential per minute)
2 bigs, no Kyle: 43 minutes; +15 (0.349 pdpm)
1 big + Kyle: 40 minutes; +62 (1.550 pdpm)

PER GAME AVERAGE (3 games)
-----------------------------
2 bigs + Kyle: 12.33 minutes; +15.33 (1.243 plus differential per minute)
2 bigs, no Kyle: 14.33 minutes; +5 (0.349 pdpm)
1 big + Kyle: 13.33 minutes; +20.67 (1.550 pdpm)

[Kedsy]

Well, if today's game is a harbinger of things to come, then the small lineup people may ultimately be able to claim victory. We went small 65% of the game. On the other hand, although we are dealing with smaller numbers (which makes a one game sample even less reliable than usual) the big lineup was more effective tonight.

It's worth noting that the "middle" lineup barely played, as Kyle was on the court for 38 minutes.

MARQUETTE
------------
2 bigs + Kyle: 12 minutes; +4 (0.333 plus differential per minute)
2 bigs, no Kyle: 2 minutes; -2 (-1.000 pdpm)
1 big + Kyle (or 0 big + Kyle): 26 minutes; +3 (0.115 pdpm)


TOTAL (4 games)
----------------
2 bigs + Kyle: 49 minutes; +50 (1.020 plus differential per minute)
2 bigs, no Kyle: 45 minutes; +13 (0.289 pdpm)
1 big + Kyle: 66 minutes; +65 (0.985 pdpm)

PER GAME AVERAGE (3 games)
-----------------------------
2 bigs + Kyle: 12.25 minutes; +12.50 (1.020 plus differential per minute)
2 bigs, no Kyle: 11.25 minutes; +3.25 (0.289 pdpm)
1/0 big + Kyle: 16.50 minutes; +16.25 (0.985 pdpm)

[mepanchin]

I think we probably saw more of the smaller line-up because of Marquette's personnel. They are quick and small and this let us match up really well with that. There will be bigger teams where the 2 big line-up may be better.

[Saratoga2]

Maybe I didn't read the thread carefully enough, but I did see two bigs plus Kyle seemed to work better than other lineups. What I didn't see is differentiation of which bigs. Clearly Mason is one of those bigs but is the 2nd Kelly, Miles or Hairston. Based on what I saw of the Marquette game, Kelly looked more effective than Miles and Hairston DNP.

Miles may have a sore finger and than may have impacted his play. He is about the same size as Mason, has lots of hops and strength, but he seems not to catch the ball well and finish on rebounds around the basket. Kelly has less jumping ability and lateral quickness but seems to react better to the speed of the game. Miles has a lot of upside potential as does Kelly. It would be interesting to differentiate performance between the two.

[Kedsy]

Maybe I didn't read the thread carefully enough, but I did see two bigs plus Kyle seemed to work better than other lineups. What I didn't see is differentiation of which bigs. Clearly Mason is one of those bigs but is the 2nd Kelly, Miles or Hairston. Based on what I saw of the Marquette game, Kelly looked more effective than Miles and Hairston DNP.

It's way too much work to break down minutes on specific lineups. But going through pfrduke's +/- figures, in the Marquette game the Kyle/Mason/Ryan lineup was +3, the various Kyle/Mason/Miles lineups were +3 and the various Kyle/Ryan/Miles lineups were -2. Through four games, the various Kyle/Mason/Ryan lineups are +25 and the various Kyle/Mason/Miles lineups are +21. The other big combinations didn't play enough to bother calculating, although presumably they all combined for +4.

[Kedsy]

Incidentally, and this is neither here nor there but it is sort of relevant to the big/small debate, I was very interested in this quote from Buzz Williams after the Marquette game (quote from GoDuke's quotes page): "They are very talented team particularly when No. 12 (Kyle Singler) is at the three."

The fact that he believes we are more talented when Kyle is at the 3 says to me he fears our big lineup more than he fears our small one. Of course that may be because his team plays a three-guard lineup and his bigs aren't very tall, so they match up better with our small lineup than our big lineup, but I thought it was an interesting quote nonetheless.

[Kedsy]

Interesting that in a game where are opponent had a big front line the small lineup was the dominant one. Through five games, the small lineup has been on the floor 44% of the time (17.6 minutes per game) and has overall been the most effective.

It's also interesting that the reason for the small lineup's predominance has not really been Seth (as most of the small lineup advocates predicted) but rather the emergence of Andre as a viable option at the 3.

Still lots of season to play before this debate is settled.

KANSAS STATE
---------------
2 bigs + Kyle: 12 minutes; -2 (-0.167 point differential per minute)
2 bigs, no Kyle: 6 minutes; -6 (-1.000 pdpm)
1 big + Kyle: 22 minutes; +22 (1.000 pdpm)


TOTAL (5 games)
----------------
2 bigs + Kyle: 61 minutes; +48 (0.787 point differential per minute)
2 bigs, no Kyle: 51 minutes; +7 (0.137 pdpm)
1 big + Kyle: 88 minutes; +87 (0.989 pdpm)

PER GAME AVERAGE (5 games)
-----------------------------
2 bigs + Kyle: 12.20 minutes; +9.60 (0.787 point differential per minute)
2 bigs, no Kyle: 10.20 minutes; +1.40 (0.137 pdpm)
1/0 big + Kyle: 17.60 minutes; +17.40 (0.989 pdpm)[/QUOTE]