Skip to main content
ABC News
From The Warriors To The Knicks, How We’re Predicting The 2018-19 NBA

The NBA is back in action this week, which means it’s time to fire up our NBA predictions again and see what the league’s future might hold. You may be surprised to learn that the 2018-19 Golden State Warriors and Houston Rockets look good on paper, while the New York Knicks and Sacramento Kings … don’t.

OK, so that’s not really surprising at all. But in the vast space between the favorites and bottom-feeders, there are still some interesting storylines and players to watch. And we’ll be bringing you a new way to follow those developments all season long after a tweak to the way we make predictions this year. Scroll down to the bottom if you want the full details, but in a nutshell, our interactive page now contains two projection options: pure Elo and CARMELO.1 The former is well-known to FiveThirtyEight readers — it’s the simple version that appears on our franchise history pages — but the latter is more complex, using team depth charts and constantly updating player ratings to track just how much talent is on each team (after accounting for trades, injuries and disgruntled teammates).

Let’s tip things off with a look at our preseason CARMELO projections for each conference, starting with the dominant West:

Projections for the 2018-19 Western Conference

Efficiency rankings and playoff probabilities for the 2018-19 Western Conference, according to FiveThirtyEight’s CARMELO model

Warriors 1751 63-19 1 3 >99 61% 49%
Rockets 1662 56-26 2 8 98 15 9
Jazz 1636 55-27 14 1 98 8 4
Thunder 1625 53-29 4 7 95 7 4
Nuggets 1587 50-32 3 25 91 3 1
Lakers 1582 48-34 8 14 84 3 1
Timberwolves 1559 45-37 5 21 74 1 <1
Pelicans 1555 45-37 11 9 73 1 <1
Trail Blazers 1501 39-43 15 18 34 <1 <1
Spurs 1472 37-45 17 19 25 <1 <1
Grizzlies 1436 33-49 28 10 12 <1 <1
Clippers 1434 33-49 20 24 12 <1 <1
Mavericks 1389 28-54 29 20 3 <1 <1
Suns 1361 26-56 22 30 1 <1 <1
Kings 1330 23-59 24 29 <1 <1 <1

As of Oct. 14, 2018

The favorites: Obviously, the Warriors are big favorites to win the West for a fifth straight season, with a 61 percent chance of returning to the NBA Finals yet again. Sure, the Rockets pushed Golden State to the brink during last year’s playoffs, and they added one-time superstar Carmelo Anthony over the summer, but the Warriors still have a sizable talent edge — even after accounting for the injury time likely to be missed by prized free-agent pickup DeMarcus Cousins. Houston does headline the next tier of Western challengers, though — a group that also includes the surprising Jazz and the steady Thunder (who could be better after ditching Anthony2 and trading for Dennis Schroder). But who are we kidding? It’s the Warriors’ title to win. Again.

Moving up? After improving to 46 wins last year from 40 the previous season, the Nuggets could still be on the rise in 2019 — our model calls for 50 wins in Denver. Likewise, the Lakers should be the league’s biggest gainer this year after signing some guy from Ohio over the summer. LeBron James’s new team is a strange mishmash of spare veteran parts that may not fully mesh together, but it still has an 84 percent chance of getting the franchise back to the playoffs for the first time since 2013. And despite losing Cousins, the Pelicans could end up vying for 50 wins again after snagging Julius Randle from the Lakers (particularly if Anthony Davis keeps playing like an MVP).

Moving down? The West’s middle class has a few contenders to be the conference’s least-improved team: First, the Blazers appear to be headed for a regression after playing over their heads last season, flaming out in the playoffs and doing little to improve over the summer. Also, the Spurs will test Gregg Popovich’s legendary coaching acumen with a threadbare roster that was made even worse by recent injuries such as guard Dejounte Murray’s preseason ACL tear. Every year, Popovich seems to defy the doomsayers, but there are limits. And the Clippers could take a step back in a full season spent without both DeAndre Jordan and Blake Griffin. Plus we have to provisionally place the Timberwolves in this category, given the drama involving Jimmy Butler. We still project them to win 45 games after incorporating some probability3 that Butler won’t take the floor for them, but that number would drop to 40 wins if Butler does actually leave.

Lottery watch: With Mike Conley back in the lineup and a few new faces on board, the Grizzlies could be one of the most improved teams in the league — but that still probably won’t be enough to pull them into the playoff picture. (We give them a 12 percent postseason probability.) The rest of the West’s bottom-feeders are even further out of contention: The Mavericks, Suns and Kings each have less than a 4 percent chance of making the playoffs. That might change slightly if Dallas rookie Luka Doncic outplays CARMELO’s bearish projection, but most likely all three teams will be vying for the first pick in the draft, with the Kings having an edge in the race to the bottom.

Projections for the 2018-19 Eastern Conference

Efficiency rankings and playoff probabilities for the 2018-19 Eastern Conference, according to FiveThirtyEight’s CARMELO model

Raptors 1655 56-26 7 4 >99 34% 12%
Celtics 1633 54-28 6 5 99 26 9
76ers 1617 53-29 12 2 98 19 5
Wizards 1569 48-34 10 11 92 9 2
Bucks 1555 47-35 9 16 88 6 <1
Pacers 1520 43-39 16 13 77 2 <1
Heat 1503 41-41 23 6 65 2 <1
Pistons 1489 40-42 19 12 58 <1 <1
Hornets 1474 38-44 13 23 46 <1 <1
Nets 1455 36-46 18 22 34 <1 <1
Magic 1422 32-50 25 15 19 <1 <1
Cavaliers 1396 30-52 21 27 11 <1 <1
Hawks 1373 28-54 30 17 6 <1 <1
Bulls 1367 27-55 26 26 5 <1 <1
Knicks 1326 23-59 27 28 1 <1 <1

As of Oct. 14, 2018

The favorites: A season after battling for the East’s top record — then promptly being demolished by the Cavs in the playoffs — the Raptors are back atop the conference’s projected pecking order. Who knows whether Kawhi Leonard will stay in Toronto beyond 2019, but for now, he makes the Raps (who, by the way, won’t have LeBron around in the conference playoffs to torment them anymore) favorites to advance to their first-ever NBA Final. Of course, the Celtics will have something to say about that with Gordon Hayward and Kyrie Irving returning from injury, and they’ve had a tendency to outperform these kinds of projections under coach Brad Stevens anyway. And the Sixers are hoping to put a chaotic offseason behind them, building on last year’s breakout with more stellar play from Ben Simmons and Joel Embiid. Our model doesn’t expect much from former No. 1 pick Markelle Fultz, so anything he provides could give Philly extra upside.

Moving up? The Bucks are a perennial pick to make “The Leap” behind MVP candidate Giannis Antetokounmpo, though they haven’t quite put it all together yet. Our model still has faith; it calls for an improved record in Milwaukee — and an outside chance at an NBA Finals berth. Similarly, the Wizards always tantalize with potential before finding ways to disappoint, but better health from John Wall and, yes, the upgrade from Marcin Gortat to Dwight Howard (who still brings defensive value) has our model optimistic about Washington’s chances. Finally, the Hornets are looking to bounce back from two straight nightmarish 36-win campaigns, and they have the talent to at least challenge for a .500 record and a playoff berth in 2019.

Moving down? No team will fall harder in 2019 than the Cavaliers, who may belong in the lottery conversation below after losing LeBron James again via free agency. Despite what Tristan Thompson thinks, don’t expect a fifth-straight Finals appearance for the Cavs, or anything close to it. The Pacers and Heat, two feel-good surprise teams from a season ago, could also see a drop-off: Miami mostly stood pat with a roster that was scarcely better than league average, while Indiana’s acquisition of FiveThirtyEight favorite Tyreke Evans might not be enough to offset the regression due for a team that probably overachieved last year. Lastly, the Pistons aren’t projected to get any better, and that alone is bad news for a team that missed the playoffs last year despite going all-in on a midseason trade for Griffin. We give Detroit a bit above a coin-flip’s shot at the postseason but little chance to do any damage once there.

Lottery watch: The bottom of the East is a disaster zone. A couple of bad teams from 2018 (the Nets and Magic) are slowly moving more toward respectability — if not playoff contention — but the Hawks, Bulls and Knicks figure to be among the league’s worst yet again. New York in particular has the worst CARMELO rating in the NBA, after accounting for the fact that Kristaps Porzingis might play only minimally this season (if at all). After potentially missing the playoffs for a sixth straight season, the best the Knicks can hope for might be a return to the days of frozen envelopes in the draft lottery.


For those seeking more details on our new model, welcome! We haven’t changed much as it pertains to these preseason forecasts — they’re still based on our CARMELO projections for each player, combined with depth charts for every team, and we still factor in playoff experience in our simulations for the postseason. The real change will come once the season begins, at which time we will begin updating player talent ratings on the fly during the campaign.

Those ratings will be based on 2018-19 Box Plus/Minus and Real Plus-Minus numbers as they are released by and ESPN, respectively. Starting with a given weight assigned to our preseason CARMELO offensive and defensive plus-minus ratings,4 we mix in current-season performance in proportion with a player’s 2018-19 playing time to generate new ratings that balance between our preseason projections and a player’s in-season performance.

Those numbers will feed into our depth charts to fuel each team’s CARMELO rating, a best guess at the amount of talent on the roster at a given moment. At first, these will use preseason playing-time estimates except in the cases of major injuries and trades; we’re still tinkering with how to handle smaller transactions and changes in expected minutes during the season. (Stay tuned for updates about subsequent versions of this new model!) Then we use those ratings within an Elo-like framework to simulate out the rest of the season and playoffs.5

All of this means that we are retiring our old friend the “CARM-Elo” rating, which was always an uneasy compromise between basic Elo and a more complicated depth chart-based system. You might be wondering why we’re making this change; the rationale lies with how our in-season projections have handled dominant superteams like the Warriors, who have effectively rendered the regular season meaningless.6 While our preseason forecasts usually perform quite well in various prediction contests,7 our in-season updating process was flawed for a number of reasons: We accounted for injuries and trades only indirectly (through game results), and those game results could also be highly misleading, particularly down the stretch as powerhouse teams rested players — or simply coasted to reserve energy for the playoffs. Our new CARMELO system, then, will update in an almost opposite way: It will pick up on team performance indirectly (as players’ individual numbers rise or fall) but will be grounded in who is on each team — and how good our algorithm thinks they are.

As for our pure Elo forecast, it’s based on a two-track Elo system, in which each team simultaneously maintains a faster-moving “regular-season Elo” and a slower-moving “playoff Elo.” A team’s playoff Elo tends to reduce the effect of late-season wins and losses, and we’ve found it to be a bit more predictive of playoff success historically.

You can find both systems on our 2018-19 NBA predictions page, with CARMELO as the default. (But you can toggle between the two.) Because it addresses many of the old method’s shortcomings, we think our new CARMELO system will improve predictions and give you a better glimpse into what’s going on in the NBA. If you’ve made it this far, thanks for reading, and we look forward to spending another basketball season with you.


  1. Better known as the Career-Arc Regression Model Estimator with Local Optimization.

  2. A classic case of addition by subtraction.

  3. Specifically, the admittedly nebulous guess of a 50-50 chance.

  4. Based on how much of a sample size went into the ratings — so rookies get less weight on their prior ratings.

  5. Like we do in our other models, we run these simulations “hot,” meaning the ratings change with each simulated game going forward to reflect the variance in how a team’s season could potentially play out.

  6. Or, at least, even more meaningless than usual. It’s not as though the NBA regular season was ever the most consequential affair.

  7. Except for that one year when we didn’t use Real Plus-Minus in CARMELO — but we don’t talk about that. ;)

Neil Paine was the acting sports editor at FiveThirtyEight.