Q&A: I have Grave Concerns About Marco Belinelli

Editor’s Note: Liberty Ballers’ Randy Cohen, AKA blindloyalty76, has introduced a 3 part series in which he takes a deep dive analysis into how Marco Belinelli will affect the Sixers down the final stretch of the season. You can read Part 1 here, and Part 2 here. Before posting the final installment, Randy would like to address some of the responses to his previous pieces in the form of the following Q&A.

Q&A: Why Do I Think Marco Belinelli Is More Likely to Hurt The Sixers Than Help Them?

The first two parts of my three-part series on Belli drew a large number of thoughtful comments. I thought I’d respond to some of them here, in Q&A format.

Q: Given that we have an open roster spot and we need someone to play wing, why not Belli? What’s the problem?

A: As I said at the top of Part 1, I think this is a good argument and I’m not outraged by our addition of Belli. But, since you asked, here are the potential costs I see, none of which is enormous in expectation but which, added up, do in fact make Belli a significant error in my opinion.

1) He will take minutes away from Justin Anderson, who I think is meaningfully better than Belli assuming he returns fully healthy from his ankle sprain.

2) He will probably take some minutes from TJ, who I think is a lot better than Belli.

3) He will take some minutes from TLC, who I think is about as bad as Belli but who I think is more likely to improve between now and the playoffs than Belli is.

4) In the regular season, I’m fine if Marco takes some minutes from good players like Cov and JJ; let them get their rest and if it costs us the 4 seed, so be it. But because our management mistakenly (IMHO) thinks Belli is good, I worry that during the playoffs the team may miscalculate the cost-benefit analysis of giving Belli minutes over our good players.

5) He might, indeed probably will, force us to drop either a a promising young player like TLC or Richaun or a competent player like Booker should we pick up Ilyasova. And if another player gets bought out, say a wing who is’superior to Belli, we probably won’t pursue that guy because Belli is established on the squad.

6) His likely availability may have prevented us from making a deal for a good player. I was particularly partial to the Bayless-and-a-pick for Tyreke deal, after discussions with Tyreke hoping to get a handshake deal on an extension if we don’t land LBJ or PG. I thought, and still think, that with Tyreke this team is a serious threat to win the East this year, plus I like Tyreke as a long-run piece, and I thought we should have gone for it.. I can’t help but suspect the team didn’t do it in part because they thought (wrongly, IMHO!) “Belli gets you most of what Reke does, and without sacrificing a pick.” Of course that’s pure speculation, they may not even have known Belli would be bought out. But it’s on the list of possible costs.

7) I have not done the work to back this up, but I strongly suspect that there are better players than Belli available out there, in the G League or whatever.

8) Marco may, weirdly, take minutes away from far better players who play very different positions. In particular I suspect he will take minutes that would otherwise have gone to Ersan Ilyasova. I’ll say more about this in Part 3. I actually think this is the biggest cost; so as not to keep you in suspense I’ll just say that despite the fact that it forces Dario into a tough defensive assignment for him, I think that a grouping of, for example

Amir Ersan Dario JJ TJ is far superior to the Marco alternative:

Amir Dario Marco JJ TJ

I think it’s as good offensively and far better defensively. Again, I’ll go into more detail about this sort of tradeoff in Part 3.

Q: Given that a series of coaches who are all good and at least one of whom is great; Pop, Budenholzer, BB, Clifford; have decided Belli is good enough to play 20+ minutes a night for them, shouldn’t we consider the possibility that the analysis in the ever-expanding Belinelli series is totally wrong and Belli is a quality NBA backup?

A: Absolutely! The way I see it is, if all you knew about Belli was that that set of coaches played him a lot, you’d say it’s 95+% likely that he is at least pretty good. If you then looked at his counting stats, say as summarized by BPM or similar, you’d lower that probability, but only a tiny bit; probably, one should conclude, his floor-spreading and BBIQ are being underestimated by combining simple stuff like points, steals, and assists. Michael Lewis called Shane Battier the “No Stats All-Star” because he did so many little things that don’t show in the box score; Marco could easily be such a player, the way Kyle Anderson arguably is for Pop this year.

But after seeing the on-off stats over five years, I would argue that we have to drop our estimate of the likelihood Marco is good by quite a lot. Does that mean it’s 50% likely he’s a decent backup (or better)? 20% likely? I’m closer to the latter than the former, but your mileage may vary. For sure, though, it’s possible we’re missing something important that the coaches are seeing. That said, the reverse is also possible! When Earl Weaver managed the Orioles in the 1970s, he said walks were important and sacrifice bunts, not so much. Virtually all managers, including many smart ones, disagreed. Now we know he was right and they were wrong. We are in an era of rapid change in our understanding of basketball. My guess is that in five years everyone will agree that Marco Belinelli should have spent his 30s playing in Europe because he wasn’t good enough to deserve NBA minutes, and my guess is that Daryl Morey knows it right now. But maybe I’ll be wrong, we’ll see!

Q: You use RPM as the core statistic to judge Belli, but RPM also says Fred VanVleet is better than Kyrie Irving, so obviously it’s nonsense. Why should we take such numbers seriously?

A: First, RPM does not say that Fred is better than Kyrie. What it does say is this: based on a noisy measure of their play so far this year, Fred has added more value per minute than Kyrie up to this point. Thus the following are all possible:

a) Fred is a better player than Kyrie.

b) Kyrie is better, but Fred has been better for one fraction of one season.

c) Kyrie has been a lot better overall this season, but Fred has indeed been better per minute.

d) Kyrie has played better per minute, but Fred, despite lacking Kyrie’s quality of play, has helped his team more because he got lucky. Like, say you have a better shooting stroke than me, you “shoot better,” but on one particular day I have a bunch of shots fall -- I legitimately added more value than you, but it’s still fair to say you shot more skillfully, I just got the rolls.

e) Kyrie has outplayed Fred so far per minute but Fred has benefitted from noise in the data so this imperfectly-measured stat has Fred higher. This is different from d), in d) Fred actually added more value, but just because he was lucky. In e) Kyrie added more value, but because the measurement is flawed it wrongly appears he didn’t.

Now, I agree with the consensus that a) is false. I also think b) is false. But I haven’t watched enough Fred Vanvleet to say with confidence that we have to write off c) and d), and I bet you haven’t either! I mean, we’re talking about a stout defender who gets around 11 assists plus rebounds per 36 and shoots 41% from 3 on volume! He’s a lot less good than Kyrie at driving the lane, but is that really the only skill that matters in basketball? If you pass better and rebound better and shoot threes just as well, per minute, isn’t it possible you’re not that far behind offensively even if you don’t have equal handles? And if you’re superior defensively, per minute, might that not make up the gap? It doesn’t sound crazy to me! Maybe FVV isn’t a legit 41% 3-point shooter, RPM doesn’t measure that. What it does say is, Toronto has killed it when he’s been on the floor, in part because he’s hit 41% of his threes. Should I conclude that’s inconceivable because he not only has a funny-looking Dutch name, he doesn’t have the common courtesy to spell it the same way Captain Beefheart does? If we’d had this exchange two months ago, instead of VanVleet people would have argued RPM was idiotic because it rated another funny-name guy high, Spencer Dinwiddie. But now I think everyone is coming to realize that Spencer Dinwiddie is just damn good -- RPM was right and the funny-name police were wrong! Maybe FVV is another case of that.

But suppose that’s all wrong and the truth is e), FVV is massively overrated by RPM, it’s a small sample and the noise overwhelmed the signal. OK, fine! What does that have to do with Marco Belinelli? Did Marco get a bad-luck draw this year, and last year, and for every single year since the stat has been kept? Give me a break! He just does not help the teams he plays for. At least not on the court.

Q: Isn’t RPM dependent on the rest of the team? And so doesn’t Belli just have a bad RPM because the Hawks were terrible?

A: Yes, it depends on the other players, and no, the lousy Hawks are not the reason Belli’s numbers are bad. The easy way to see the latter is to note he had an awful RPM with the title-winning Spurs and with the mediocre Hornets. His on-off numbers are terrible wherever he goes. But the more informative thing to understand is that the dependency goes exactly the opposite direction. What RPM does is this: say your team scores 10 points and gives up 9 during 6 minutes while you are playing. As we all know, you are +1. But here’s the thing: if you were +1 while playing with Steph and Klay and KD and Dray, and against the woeful Kings, that’s not especially impressive; indeed chances are you dragged them down. Whereas if you were playing with a bunch of jokers from the 2017-18 Hawks against the Warriors’ starters, your play was probably excellent given that you managed to flog your teammates to +1. Of course over any short span like that, there’s tons of luck. What RPM does is to take each possession, see how your team did in points scored and allowed, and then adjust for the teammates and opponents, using the statistical tool of regression analysis, to estimate how much you did. That’s why we don’t typically see the same guy have great RPM on good teams and bad RPM on bad teams -- the team quality is adjusted for. Of course the method is not perfect, but that’s how it’s supposed to work, and in my opinion, it’s pretty good!

Q: Isn’t Marco basically a less-good version of JJ?

A: It would be more accurate to say that Marco is an about-equal version of Bayless. But, sure, we can say he’s on the spectrum between JJ and Bayless. Look, if you loved Bayless’ play, and think it’s a big cost that the coach has benched him, then it makes sense to be happy to see Belli; JB without the negative feelings associated with his contract and with us having had to watch him play. But if you thought Bayless was terrible all year and you were less unhappy to at least see TLC out there, then it’s not clear to me why you’d think Belli is going to help.

Q: But Belli is good at so many things; driving, shooting the three, passing, free throws... how can he not be a good player?

A: Look, it’s just really hard to get around the fact that this sport is half defense. I think one reason folks fail to see that is that with superstars, you can make a case it isn’t true. Since the offensive team gets to decide who has the ball and where they take it, it’s possible for the very best offensive players to add 4-6 points per 100 possessions on offense, whereas it’s not clear anyone can do that much defensively; almost surely no non-center can and perhaps not even the best rim protectors are that good. So it’s natural to think offense dominates. But at the low end of the scale, it’s just the opposite -- the opponent can pick on the weakest link. So even if Belli were very good offensively, he could give it all back and more on D. But then there’s this: he’s not that good offensively! Basically he’s good-not-great at perhaps two-thirds of the things you do on offense, and poor at the remainder. He’s a good deep shooter (37-38%) but not elite. He’s OK at driving and passing, but nothing special. The free throws are legit, I gotta give him that, but he doesn’t get to the line much. And then the rest: no offensive rebounding, no screen setting, not a real volume shooter (only 15 points per 36 for his career). So, yeah, there’s no inconsistency here. I return again to bayless. Bayless shoots 37% from 3, gets more assists and rebounds than Marco, is a better defender. But no one thinks it’s just obvious that Jerryd bayless has to be a solid NBA player because he checks those boxes. He isn’t, and neither is Marco.

Q: But if he’s good offensively, and our bench is weak offensively and strong defensively, isn’t he a net benefit?

A: I cannot unequivocally say this view is wrong, but logically it seems wrong to me and I’ve never seen a shred of evidence it’s correct. Let’s say you have an offensive squad that in the average 100-possession span ties teams 120-120, and a defensive group that on average ties teams 80-80. Can we have a winning team by just switching a defensive guy, say Andre Roberson, to the offensive unit in return for say Lou Williams? I don’t see why that would obviously help. It might help, it might hurt, it might leave things the same. We are seeing a test of this thinking right now; people thought Houston wouldn’t get full value from Chris Paul because they already have a lot of ball handling in James Harden. As of now CP3 leads the league in RPM, Harden is second, and the team has played the best of anyone in the NBA. I understand it seems like if you have Covington then it matters less if your second wing is a good defender, but as I say I know of know evidence for this and I doubt it’s true at all, let alone true enough to overwhelm the fact that Belli is a -3 player.

Q: Even if he’s not great, can’t he win us a couple games down the stretch?

A: Anything is possible! But the rule of thumb is that a team gets at extra win for every 25-30 net points they score/prevent. Maybe I’ll do a full post on this as it probably sounds high to some. Remember, half the time you would have won regardless, so those points are wasted! Anyway, we have about 25 games left; if Marco plays half of each game that’s around 12.5 full games or 1250 possessions. To get us 50-60 extra points -- two extra wins -- he’d need to be 4 or 5 points better than the guys he’s replacing. I.e. the equivalent of replacing TJ with Damian Lillard as our guard off the bench. Now, since we were throwing -2 and -3 guys out there, all it would take would be a +1 or +2 player to get that level of improvement. But Marco is not a +2, or a +1, or a 0, or even a -1 or -2. So, no, he won’t win us a couple of games, on net, unless we get lucky. I don’t deny that could happen, and I hope it does!

Q: But wasn’t Marco terrific Saturday night against Orlando?

A: Well, he was +15, so, great stuff on that measure. But here’s the thing: if you think Belli was good Saturday because his +/- was good, then I don’t see how you can’t agree that his last 400 games of overall-awful +/- is compelling evidence he stinks. And if you don’t believe in +/-, then, first of all, why the heck not? But, second, why do you think he was good Saturday? He played 27 minutes and got no steals, no blocks, and only 2 each of assists and rebounds. All that looks lousy. He shot 33% from three, which is not very good, and took only three of them, so it can’t be excused as high volume. Overall he used 13 possessions with shots (12 FGA plus he was sent to the line once), and scored 15 points, that is approximately average scoring efficiency. And then he had those four (4) turnovers! So, bad defensive stats, bad passing and rebounding and ball handling stats, average shooting efficiency... it sure doesn’t look like a good game! Again, unless you like plus-minus, as I do. But if we’re doing plus-minus analysis, Marco is in trouble on all the other nights!

Q: OK, maybe he’s pretty bad, but don’t we need someone to be a shooter and initiator off the bench?

A: This is the topic of Part 3, which will be posted soon! The short version is, I think we tend to overrate the importance of fit and underrate the importance of quality. But read Part 3 when it posts and see what you think!