FanPost

Fanpost: How to figure out successful Lakers rotations - statistical analysis

Raj Mehta-USA TODAY Sports

Editor's Note: This story on figuring out what the numbers tell us about the best possible rotation combinations for the Lakers was originally published in our FanPosts section, but had such good info we wanted to bump it up to the front page.

Original story follows.

I've dug into the stats once more, this time focusing on lineups. There's several questions I'm trying to find an answer to:

· Which lineups work, which ones don't?

· Why do the bad ones fail?

· Why do the good ones succeed?

· Which backcourt, wing or forward combos work best?

· Which non-LeBron, non-AD, non-Westbrook lineups work best? How to load manage, how to construct

rotations.

· Which of Vogel's decisions were good, which bad, which need a little tweak?

· What works with LBJ in the lineup, with AD, with Russ, with all of them, with two bigs?

· How to fix or alleviate the turnover problem, how to get defensive rebounds with this roster. How to do it

while also winning the game, not one statistical category.

· Individual play this season, whom do we overrate or underrate?

Of course I won't find perfect answers to all of these, solutions that work every game vs every opponent regardless of injuries, officiating, shooting, individual mistakes etc. But the data should provide clear tendencies that can help improve the rotations. Note, I'm just looking at the 10 most used lineups for each player (5-man rotations but also 2- or 3-man combos) and how they fared so far. When in doubt, I'll also consider additional data from previous seasons.

1) 5-man lineups including the big 3:

First shocker: there's only 1 lineup that outscored opponents on average that includes all of the big 3! Should we consider benching one, is it small sample size theater or did the lineups get dragged down by other factors/players? Let's see. The best lineup:

Westbrook/Bradley/Bazemore/James/Davis +2.8 points per 100 possessions, 34:25 minutes playing time

What was good about this lineup? They got to the free throw line more than their opponents, they got a boatload of steals (+12.0) and won the turnover battle by a huuuuge margin of 24 turnovers per 100 possessions.

Why was the lineup just +2.8? Because they couldn't shoot to save their lives. Got 34 additional field goal attempts per 100 but just ONE more made field goal... Field goal percentage 17.7% lower than their opponents, effective field goal percentage 21.8% lower, 3 point percentage 18.2% lower. Could that one glaring weakness be fixed in an otherwise good lineup? Not over a bigger sample size in all likelihood. While Bradley is 38.8% on 3s this season, Bazemore has shot poorly and neither one provides the volume necessary. This lineup is likely to face a lot of packed paints with tons of outside jump shots for the big 3. This lineup in fact took 6 more 3s than opponents but have 2.4 fewer makes than opponents.

Conclusion: the lineup was solid. In order to improve on the glaring weakness and become not only better but sustainable, it needs a tweak though. There's something to work with though, a foundation to start from. Let's move on to the worst lineup including the big 3:

Westbrook/Bazemore/James/Davis/Jordan -15.7 points per 100 possessions, 47:11 minutes playing time

What was bad about this lineup? They lose big time on the boards, offensive and defensive alike, opponents get a lot more steals (-6.9), they lose big time in 3 point volume (-12.7) and makes (-3.9), they lose the turnover battle, albeit by a relatively small margin (2.9 more turnovers per 100).

What did they do right with this lineup? They blocked way more shots (+8.8) and committed fewer fouls (-3.9). Shooting percentages and free throws were about equal.

Conclusion: this is one of the lineups that doesn't work because it struggles to defend the perimeter, it struggles to defend the paint (not the rim) and when they do get stops they don't get the defensive boards. On offense, they lack spacing and 3 point volume with two bigs out there and no good shooter. The shrinked floor leads to drives into the paint and cramped passing lanes where opponents get steals. Short: they get spaced out on defense, with bigs either trying to contest outside, in no-man's-land between the paint and the arc, sliding over to provide help defense for a poor perimeter defense or busy blocking a shot, failing to secure the rebound. On offense, they fail to space out the opponent themselves and can't play to their offensive strengths. Even fouling less ain't a positive because they're too far away to actually commit fouls, opponents are not.

Note: There's a similar lineup with Bradley replacing LBJ that's -18.3 points per 100. They rebound significantly better and seem to have slightly better perimeter defense but they lose their shot-blocking edge now, their shooting efficiencies plummet and they get killed in steals and turnovers. Vogel played these two lineups a lot early and seems to have learned his lesson. This was a failed experiment. Vogel's conclusion here was to use Bazemore significantly less and play AD more at the 5 instead of the 4, two conclusions that the data back up, more on that below. The glass is half full, we can expect better lineups moving forward. The glass is half empty, we're in for different kinds of experiment and haven't figured out winning lineups yet. Yup, I believe both statements are true.

Back to this question: Should we consider benching one, is it small sample size theater or did the lineups get dragged down by other factors/players? It's not too small of a sample size in my opinion, those were the most used lineups with the big3 in general as well as the most used with them, DJ at the 5 and/or Bazemore in. So who drags it down or which player type drags it down? I'm checking onto this one next:

2) Which lineups work well with LBJ?

Westbrook/Reaves/LBJ/Melo/AD +58.8 in 7:13 minutes, small sample size

They had exceptional spacing in this one, volume shooting from 3, great percentages from both 2 and 3, no rebounding issues, dominated in assists and limiting turnovers. Sounds great but wanna double-check due to small sample size.

Westbrook/Reaves/Melo/LBJ/Jordan +30.0 in 9:59 minutes, small sample size

Typical small sample size theater here: +66.7% in 3 point percentage, +50% in rebounding percentage on both ends, +30 in assists, +23.5% in field goal percentage are nowhere near sustainable. So is +20 in turnovers per 100, the negative outlier. But expect this +30 to drop by at least 20 points over a larger sample size.

Westbrook/Bradley/THT/LBJ/AD +17.7 in 13:19 minutes, limited sample size

Kind of small sample size but the outliers aren't as extreme: this lineup doesn't take a lot of 3s (-16.4) but shot well(+20.8% from 3). They draw a ton of fouls and their effective field goal percentage was 8% better than that of opponents. This is the lineup people call bully-ball with a little bit of shooting in it and a lot of slashing. They got rebounding issues (-12.5%) but win the turnover battle by 11.8, both of which can be considered outliers. I love this lineup as a situational lineup. Vs an opponent you can bully, it should be awesome. Vs opponents you can't bully it could produce a lot of turnovers and see their efficiency plummet unless they shoot well that day. Not too shabby defensively, that helps their floor quite a bit. Offensive variance is the one to be wary about here.

Westbrook/Bradley/Melo/LBJ/AD +14.8 in 6:39 minutes, small sample size

Same lineup with Melo replacing THT. They shoot well from 3 (+16.7%), even though that seems to be a small outlier as well but Melo and Bradley are two very good spacing options. They rebounded very well (+16.7 total rebounding percentage) but turned it over a lot (+14.8). Hardly got to the free throw line (-12.4 attempts per 100) with fouls on the defensive end apparently the bigger issue. I expect all of these values to regress to the mean over a bigger sample size though. Melo in for THT is the go-to move if bully-ball doesn't work, paints are packed, THT can't convert from 3, no driving lanes available. This lineup should see less variance than the one above.

4-man combos including LBJ:

The four best (and only) lineups with a positive net rating all include Russell Westbrook. Which is weird because LeBron should do better without him vs bench lineups and having a shooter instead of Russ, right? Well, both is true to some degree, more on that soon. Which 4-man combos work well and why? I feel most confident in the second-best combo, this one:

LBJ/Melo/AD/Russ +14.3 points in 25:38 minutes playing time

Why this combo? Because they're solid across the board, win the turnover battle (5.4 fewer per 100), have good spacing (+7.1 attempts, +6.2 makes) on reasonably sustainable shooting percentages, there's no extreme outliers/high-variance in any category that could 'cause a huge opponent run and, obviously, it is comprised of the four best players on the team.

LBJ/Melo/Reaves/Russ +30.4 in 33:43 minutes playing time

The best lineup includes Austin Reaves and while the shooting is irrationally good and for sure an outlier, the sample size is bigger. I'm concerned though that fouling more (+5.1) and getting significantly fewer free throws (-12) while losing the turnover battle (2.7 more per 100) are no major outliers. Obviously this lineup is defensively worse and you lose a major scoring threat with AD. Furthermore you add more variance by playing a rookie. All that is why I don't believe this is the best 4 man combo in the long run. I do believe it's a good combo though, the eye-test and bigger sample size give a strong indication in that direction.

LBJ/Bradley/AD/Russ +9.6 in 65:30 minutes playing time

Bradley replacing Melo compared to the first lineup. Defense for offense while playing a little smaller. Still, rebounding is no issue here (even), they get significantly more steals (+5.6), win the turnover battle by a huge margin (12.6 fewer) which seems to be the difference why they win these minutes: they play good defense, play sound basketball, have a decent volume of 3s (even) regardless of shooting worse than opponents (-2.4% from 3), it seems to be enough to keep defenses honest and limit turnovers. The issue of this lineup is lack of efficient offense (-8.3% efg) so far but sound defense should always provide at least a solid lineup option here.

3) The combos that don't work with LBJ:

ALL combos when the big man is not AD at the 5. Only Dwight, only DJ, AD plus DJ next to LBJ, they all lose by 9 to 28 points per 100. All of them lose big time on the boards, big time in assists, all of them lose the turnover battle, all of them lose from 3 and in field goals made. Way too many negatives, I'd be confident to throw these into the bin. In consequence, this also means that the second big is one of the issues dragging down the rating of the starting lineup, simply because it doesn't work with LeBron.

4) 3-man combos with LBJ:

LBJ/AD/Melo +10.7 in 59:16 minutes

No surprises here, a lineup that should dominate second units and beat starters, they do. Strengths: great efficiency (+18.1% from 3, +8.6% from the field, +13.6% effective field goal percentage), playmaking (+6.8 in assists), blocks (+5.4). Weaknesses: fouling too much (+4.6 fouls, -10.3 in free throw attempts).

LBJ/Bradley/Westbrook +6.6 in 74:13 minutes

The surprise here is decent 3 point shooting (about even in volume and makes, only slightly worse in percentage). Strengths: getting steals (+5.0), winning the turnover battle (10.1 fewer per 100). Weaknesses: effective field goal percentage (-6.9%) is a slight negative, other than that no weaknesses.

LBJ/Melo/Westbrook +6.1 in 68:47 minutes

There's the second core to build around when AD sits. Russ in for AD offers very similar, albeit slightly worse results to the first lineup: efficiency is also a major strength (+10.5% from 3, +7.5 from the field, +8.6 efg), rebounding is about the same, assists to turnovers are slightly worse, blocks are close to a neutral instead of a strength. Fouls and free throw discrepancy remain a weakness (+5.2 fouls, -13.2 free throws). AD being the better player and being way better defensively doesn't make the difference surprising. What's surprising here is that there's no major shift in playing style/any statistical category apart from maybe blocks.

These are the only positive 3-man combos including LBJ. How about the big 3 together?

LBJ/AD/Westbrook -3.7 in 140:13 minutes

Strengths: blocks (+4.0), winning the turnover battle (6.4 fewer). Weaknesses: rebounding (-6.8%), efficiency everywhere but from the free throw line (-5.7% efg total). Looking at individuals, the rebounding value points to issues within the defensive scheme/the 5-man lineups in my opinion. Which confirms itself when we look at LBJ's worst 3-man combos:

Blocks remain a strength throughout but the rebounding issue gets exacerbated with Bazemore in, the assist rate drops with either Bazemore or DJ in, the turnover rate explodes with DJ in while they also lose big time in 3 pointers made and field goals (even when they shoot significantly better than opponents) and the defensive rebounding issue and assist discrepancy are at their worst when both Bazemore and DJ play next to LeBron.

Final check: Bazemore's 5-man lineups: his four best lineups all include Malik Monk and either Rajon Rondo or Russell Westbrook. So that's how you play him effectively, not next to LBJ. DJ's 5-man lineups: his four best lineups are all small sample size. Looking at 3- and 4-man combos shows that the combination of DJ/Monk and Bazemore is by far the best for him (3-man combos) and adding either AD to it (+26.4 points, a little outlier with 3 point shooting but no major weakness) or adding Russ to it (+12.8 points, quite balanced without big outliers or weaknesses) or - even better - both to it (+16.7 points in a solid 23:20 minutes, no outliers or weaknesses so far, good efficiency overall, solid in assists and turnovers, strong in shot blocking) could be an option but stay far away from playing Bazemore or DJ or both with LeBron.

5) Finding a forward combo for LBJ, 2-man combos:

First of all, I don't put much stock in this data as Dwight Howard and Talen-Horton Tucker currently rank as the worst partners for LBJ. Dwight has positive lineups with LBJ in 19/20 (when adding two-way wings like Green and KCP or KCP and Kuzma or Caruso and Kuzma plus either AD or Rondo), THT has positive lineups with LBJ last season. But when the two-man combo tells me the same that the 3-,4- or 5-man lineups were showing, along with the eye-test, then it should be helpful. So let's go to the good combos:

LBJ and Reaves +29.3 in 65:17 minutes

Unsustainably hot shooting but overall that sound that this should remain a winning combo regardless of shooting variance. Major strength: assists. The eye-test confirms that, Reaves makes quick decisions and the ball moves with him. This is certainly a good fit next to LeBron.

LBJ and Melo +13.3 in 128:14 minutes

The guy who carries the most gravity as an outside shooter, furthermore a friend for many years who wants to win a ring. Even if he does screw up on defense, it's likely not due to lack of effort. He'll take any criticism from LeBron, especially in the big games. Melo has been far exceeding expectations so far and the data 100% backs up the positive impact he's made. Almost all of Melo's 5-man lineups are positives or big positives, especially when the team plays big at small forward. Strengths: the Melo/LBJ combo wins because those lineups are efficient from 2 and from 3. They lose some at the free throw line but as long as there's some resemblance of shooting and spacing, these lineups should remain clear positives.

Ariza and Nunn as forward partners remain to be seen, THT at minimum seems to work with the big 3 plus Avery Bradley in. Bazemore at the 3 next to LeBron doesn't maximise the potential or is flat-out bad. AD at the 4, those numbers are still a little skewed for this season.

6) Finding rotations when LeBron plays:

a) Big 3 available, means AD at the 5 should be automatic (it ain't yet with Vogel):

Bradley should be the fourth guy, with either THT or Melo being the fifth. Once Reaves is back, you can give that big 4 plus Reaves lineup some more run. That's three playable, very positive lineups to start. Possible downsides: effort on the defensive end is a necessity , including and in particular for the big 3. And the team can't lose Avery Bradley right now, would be great to find another viable lineup during the season that includes Monk or Ellington or Nunn instead of Bradley or Reaves.

b) Big 3 playing with two bigs:

Can't play Melo with this lineup unless LeBron wants to chase around the Donovan Mitchells of this world. So the next best option is once again Avery Bradley. So far they've been a -5.9 per 100 possessions with Bradley and DeAndre Jordan in. BUT: small sample size and the difference came exclusively at the free throw line and from turnovers. If Vogel doesn't want to start AD at the 5, this may be worth trying some additional minutes. Vogel better remains a short leash in case of emergency though. Reaves and Jordan will be the next alternative that COULD work to some degree. It's safe to say though that AD at the 5 will always be the much better lineup over a bigger sample size. I'm expecting lineups like big 3 + DJ + Ellington or Monk to get absolutely torched on the perimeter as they can't defend the point of attack at all and such a lineup would force LeBron to defend out on the perimeter instead of roaming in between as a free safety of sorts. Going off the data so far, this only makes sense should it become clear that AD really can't play the 5 in the regular season. Before going to two bigs lineups with AD though, I'd consider playing just Dwight or just DJ. There's a strong indication so far that AD should never play the 4 when LeBron's on the floor with him.

c) Non-AD minutes with Russ and LeBron together:

Reaves and Melo has been the best addition to Russ and LeBron, regardless whether DJ or Dwight is the center. Both lineups are small sample size so far. With DJ, they are +30 but there's so many outliers that it can't hold up. With Dwight, they've been +4 but been atrocious on the boards (outlier as well, still not looking good). Instead of looking for the winning combos to Russ/LeBron plus a big, I'd look for a better lineup that has only one of Russ or LeBron on the floor and gives each more shooting.

d) Both AD and Russ sit:

No 5-man lineups without both to be found among the top10 most used. Same for 4-man combos with LeBron, they all include either Russ or AD. 3-man combos: LBJ/Bazemore/DJ is the only one (-2.0 per 100) so let's try to find one from last season where you had AD but not Russ. Nope, Gasol, Harrell, Schroder, very different lineups. But noteworthy: LeBron's best lineups are the ones with good 3-point shooting and solid defense. Duh, narrative confirmed XD. Two years back? There you go, Dwight, LeBron, Rondo. Plus Kuzma and either KCP or Caruso. The one with Caruso took fewer 3s but shot a much better percentage, was a much better lineup (+36.6 to +15.5) and was far superior in effective field goal percentage (+22.7 to +4.4). It's a bit of a reach. Nonetheless I take this one as a base to find a potential bench lineup with LBJ that works better than the ones under 6b. Many have commented that Austin Reaves is the closest thing to Caruso. And instead of trying to find an equivalent to KCP or Kuzma, I'm just going with the known commodity here, plug in Melo next to Reaves. Rondo/Reaves/LBJ/Melo/Dwight would be a lineup I'd try. You don't need too much shooting volume but accuracy and Melo provides plenty of volume already. With Reaves out, I could imagine any other guard working as well. Monk as he works really well next to Rondo, Ellington to have not one but two snipers out there, THT because you already have volume shooting and could add some defense and even giving Bazemore another chance as he's probably the closest thing to KCP. When Nunn and Ariza come back, there's more options. Lots of experiments but I believe it's worth investing the time because a) it would provide additional rest for AD and b) it prevents a two-big lineup with AD, all of which have been negatives so far.

e) Russ sits, AD plays:

Melo, Monk, Bradley have been positives next to either AD or LeBron. Looking at the starters and the non-AD lineups, Melo and Bradley could already be in for heavy minutes. If they're playable, that's a good option for sure, e.g. Rondo/Bradley/THT/LBJ/AD is a lineup that's +17.7 with Russ at the point, might be worthy trying with Rondo. Monk at point but as a secondary playmaker could work as well: Monk/THT/Melo/LBJ/AD are -7.8 per 100 so far but in just 4:13 and they've turned it over like crazy (+12.2) and fouled a lot. Rebounded well and dominated in shooting efficiency so if they can get the turnovers and fouls down, who knows... Nunn at the point in either lineup could become an interesting option.

7) LeBron sits:

And here's one of the crucial questions. Fortunately, there's a little bit of data now that LeBron has missed some games. Westbrook/Ellington/THT/Melo/AD has been an exceptional lineup (+43.6 per 100) in limited minutes (18:05) so far. Low volume from 3 (-20.5 attempts) but great efficiency (+18.2% from 3, +18.4 from the field, +21.5 efg), they've been great on the defensive glass (+23.5%), even in turnovers and absolutely crushed opponents at the free throw line (+38.5 in attempts, 10.3 fewer fouls than opponents). Looks to me as if the gravity of Melo and Ellington absolutely has opponents scrambling on defense with lethal players in Russ, AD and THT using the space to penetrate and draw fouls. I'd be extremely curious about this lineup's shot chart. Of course, everybody being locked in defensively is a necessity when you play both Melo and Ellington but I love the potential of this lineup. I'd even consider starting them. Or why not throw off opponents and come out in the 3rd with this one, especially after Melo and Wayne have hit some shots in the first half?

Wanna play AD at the 4? Here's an option: Westbrook/Monk/Bazemore/AD/Jordan is +16.7 so far in 23:20 minutes. They in turn get crushed at the free throw line (-21.5) but look otherwise sound, with a huge edge in 2 pointers taken and made. Not sure this one holds up but it's worth looking into it as a bench lineup. Of course you can run the same lineup with AD at the 5 and Melo for DJ. That one is +8.6 so far, almost exclusively because they've shot horrendously bad from 3 and from the field. Nonetheless it easily looks like the superior lineup once you exclude shooting percentages.

The fourth quarter lineup from the Detroit game: Westbrook/Bradley/THT/Melo/AD. A potential starting lineup without LeBron. +13.4 in 48:37 minutes, looking extremely sound with smaller to decent advantages in a lot of categories. Don't think this one has a ton of noise attached to it.

8) Rotations to consider, summary:

Starters without LeBron: the Knicks are next, right? I like THT vs RJ Barrett and Melo vs Randle in this one. Noel is one of the best matchups for AD. Kemba starting would be a boon. He's fallen off to 28% from 3 in November and Fournier struggles even more. What scares me about the Knicks is their bench with Rose, Toppin, Quickley and Burks. So I'm paying much more attention to the bench minutes in this one. Can't be slow vs their bench, Melo starting is an easy call. Only question to me is Bradley or Ellington starting. I'd go with Westbrook/Ellington/THT/Melo/AD, for several reasons: 1) The added spacing should help to get the Knicks rotating. They defend the paint well but Kemba ain't no match for Russ if driving lanes open up and AD should be able to benefit. 2) I want the quicker Bradley vs their backup backcourt on defense. 3) The Knicks are physical but they foul quite a bit. That lineup has been awesome in drawing fouls so far. Of course this could backfire, especially if they shoot poorly from 3 or don't move the ball well but this one looks most promising to me.

Bench rotations are gonna be tricky with so many guys out at the 3/4. Probably need to give Bazemore some run at the 3. As I'm struggling with it myself, I think Vogel hasn't been in such an envious position and Pelinka definitely needs to add another 3/4, even more so after it's clear that AD should play the 5 for the vast majority of his minutes.

Sorry about the different font sizes. Had to copy and paste once and didn't find a way to edit it.

What's your take on these? Has Melo been the key factor for you? Is Bradley a good fit? Would you like to see AD play the 4? What starting lineup do you like? Which one do you like for non-LeBron minutes? Has Vogel done a good job, factoring in how injuries limit his options or do you have a suggestion what he could do better?