Four thousand, five hundred and five votes separated Victor Wembanyama from Anthony Edwards in the race for the Western Conference’s final 2026 NBA All-Star starting spot. Both had arguments rooted in exceptional basketball. Wembanyama could wreck an opponent’s offense without touching the ball, while Edwards could manufacture points against defenses built to stop him.
The NBA’s voting system couldn’t separate them.
On January 19, both players finished with identical weighted scores of 5.75. Wembanyama had received 1,965,462 fan votes. Edwards collected 1,960,957.
That narrow margin sent the San Antonio center into the starting lineup.
No box-score calculation could determine which player deserved those extra votes. Even a sophisticated impact model needed information about the voters themselves to forecast the result.
The outcome exposed a problem facing anyone developing statistical models for NBA All-Star selections. Basketball value, public recognition, and coaches’ judgment don’t always produce the same rankings.
Analysts could measure individual performance through Estimated Plus-Minus, Box Plus/Minus, and forward-looking projection systems such as DARKO. Those tools offered detailed assessments of what players contributed on the court.
Yet the final roster depended on people making decisions, not computers sorting players by impact.
The NBA’s voting formula created its own statistical contest
The NBA laid out its selection rules months before the January announcement.
Under the 2026 format, fans controlled 50 percent of the starting vote. Current NBA players and a media panel accounted for 25 percent each.
NBA Communications confirmed those weights in its January 19 release detailing the final voting results.
The league ranked candidates separately within each voting group. Officials then combined those positions into a weighted score, counting fan rankings twice and player and media rankings once each.
They divided the total by four.
Five players with the lowest combined scores became starters in each conference.
Unlike previous seasons, the NBA no longer required specific numbers of guards and frontcourt players in the starting lineup.
That change simplified one part of the prediction process. However, it couldn’t resolve the uncertainty surrounding individual voters.
Wembanyama and Edwards divided the electorate
The Western Conference results showed how dramatically opinions differed.
Players ranked Edwards fourth and Wembanyama seventh. Meanwhile, the media placed Wembanyama fourth and Edwards fifth.
Fans favored Wembanyama by one position, giving him the edge that would eventually decide the starting spot.
Those rankings produced identical weighted scores of 5.75.
Under the NBA’s published rules, the player with more fan votes won the tiebreaker. Wembanyama finished 4,505 votes ahead.
A statistical forecast could have identified both candidates as likely All-Stars. It might even have assigned each a realistic probability of starting.
However, evaluating their basketball credentials wasn’t enough. The model needed to anticipate a tiny difference in public support between two enormously popular players.
Edwards eventually joined the team as a coaches’ reserve.
His selection confirmed that the debate wasn’t about whether he belonged among the conference’s best players. The question concerned which recognition he would receive.
NBA All-Star prediction models needed to distinguish between forecasting basketball excellence and forecasting the ballot.
That problem extended to players whose contributions couldn’t be reduced to a scoring average. Some dominated the ball, others controlled the paint, and a few influenced winning in ways conventional statistics struggled to capture.
What advanced metrics knew before the ballots closed
Basketball analytics could explain much of the talent hierarchy entering the 2026 selection process.
Nikola Jokić and Shai Gilgeous-Alexander offered relatively straightforward cases.
Jokić orchestrated Denver’s offense through scoring, passing, screening, and rebounding. His influence extended to possessions in which he never attempted a shot.
Gilgeous-Alexander relentlessly attacked opposing defenses while organizing Oklahoma City’s offense.
Both brought elite statistical production, established reputations, and major responsibilities on competitive teams.
The harder decisions began farther down the ballot.
Impact ratings couldn’t settle every comparison
Dunks and Threes’ Estimated Plus-Minus, or EPM, estimates a player’s contribution in points per 100 possessions. It combines statistical information with adjusted plus-minus calculations that account for teammates and opponents.
That helps analysts compare players with different responsibilities.
A guard might create offense through difficult pull-up jumpers. Meanwhile, a defensive center could change an opponent’s entire approach to attacking the basket.
Basketball-Reference’s Box Plus/Minus, developed by Daniel Myers, offers another perspective. It uses box-score statistics, player position, and team performance to estimate impact.
However, both approaches have limitations.
A defender might force three opponents to pass out of the paint without recording a block. Another player could spend an entire possession directing teammates into the correct positions.
Those contributions matter even when traditional statistics overlook them.
The models also answer a question different from the one on an All-Star ballot.
They estimate basketball value. They don’t measure how popular a player has become or predict which candidates coaches will reward.
DARKO added a test of staying power
DARKO approaches player evaluation through projections rather than simply ranking current production.
Its system weighs recent performances against a player’s established history, using statistical techniques including a Kalman filter.
That distinction becomes valuable when a younger player suddenly produces career-best numbers.
A season-to-date ranking immediately rewards the improvement. DARKO may respond more cautiously until additional games strengthen the evidence.
However, All-Star coaches don’t need proof that a breakout will last another three seasons. They can reward what a player has already accomplished.
The numbers could identify valuable players. Predicting the roster meant understanding which evidence the voters would trust.
Portland and Atlanta brought that tension into focus. Their emerging stars weren’t merely putting up better statistics. They were changing how their teams played.
Breakout stars forced analysts to reconsider their assumptions
Deni Avdija and Jalen Johnson entered the 2025-26 season without established All-Star résumés.
By February, both had earned their first invitations.
Their production challenged expectations shaped by previous seasons, while Detroit’s Jalen Duren made a compelling case through a completely different style of basketball.
Portland gave Avdija the ball and watched his numbers climb
The Trail Blazers handed Deni Avdija substantially more offensive responsibility.
He attacked defenders off the dribble, created opportunities for teammates, and carried a greater share of the Trail Blazers’ scoring.
The difference appeared immediately in his numbers.
According to NBA.com’s February 2026 breakdown of first-time All-Stars, Avdija averaged 25.5 points, 7.2 rebounds, and 6.7 assists through games played on February 1.
His scoring average had climbed 8.6 points from the previous season. Meanwhile, his assists increased by 2.8 per game.
Only Jokić, Luka Dončić, and Avdija averaged at least 25 points, seven rebounds, and six assists at that stage.
That wasn’t merely a player taking a few additional shots.
Portland had given Avdija more control over its offense. He initiated possessions, attacked mismatches, and created scoring chances when teammates couldn’t find space.
The tactical shift changed how defenses had to guard him.
An opponent couldn’t simply prepare for his scoring. It also needed to account for his passing and growing command of the ball.
His development created a difficult forecasting question.
How heavily should a model weigh previous seasons when a player’s responsibilities change so dramatically?
A system emphasizing current production would recognize Avdija’s All-Star case quickly. However, a longer-term projection might need additional evidence before treating the improvement as permanent.
Western Conference coaches didn’t need another year.
They selected Avdija as a reserve, making him the first Israeli player to earn an NBA All-Star invitation.
His emergence showed how quickly a breakout season could outrun an established statistical reputation.
Jalen Johnson made the box score impossible to dismiss
Atlanta’s Jalen Johnson presented an even broader statistical profile.
At the reserve announcement, NBA.com credited him with averages of 23.1 points, 10.5 rebounds, and 8.0 assists.
Those three numbers told the story.
Johnson could score in transition, finish above the rim, collect rebounds in traffic, and initiate possessions as a passer.
His eight assists per game reflected an unusual level of playmaking responsibility for a forward.
The combination helped him stand out even when Atlanta’s team results complicated the discussion.
A traditional scoring ranking might have undervalued him compared with a higher-volume shooter. Broader statistical models could capture more of his contribution through passing and rebounding.
Every defensive rebound offered another chance to initiate a possession. His passing also created shots without forcing him to attempt every basket himself.
Johnson earned his first selection as an Eastern Conference reserve.
His breakthrough demonstrated why analysts couldn’t judge individual production solely through a team’s record or a player’s established reputation.
Jalen Duren challenged scoring-heavy rankings
Jalen Duren brought an entirely different argument.
The Detroit center averaged 18.0 points and 10.7 rebounds when NBA.com reviewed the first-time selections.
He also collected 3.8 offensive rebounds per game, placing him among the league’s leading players in that category.
Those extra possessions mattered.
A missed shot didn’t always end Detroit’s attack when Duren established position beneath the rim. His offensive rebounding kept possessions alive, while his interior finishing provided an efficient scoring option.
Defensively, he contributed to a Pistons unit that ranked among the NBA’s strongest.
Some of his most valuable work never appeared in a conventional box score.
A center can change an opponent’s decision before the shot occurs. Drivers may hesitate, reverse direction, or pass when they encounter a defender protecting the paint.
Impact models attempt to measure that influence. However, separating one player’s contribution from the surrounding defense remains difficult.
Coaches who prepared for Detroit saw those problems firsthand.
Duren joined teammate Cade Cunningham at All-Star Weekend, giving the Pistons two selections.
For the 2026 NBA All-Star predictions, Avdija, Johnson, and Duren demonstrated three paths to recognition: offensive creation, all-around production, and interior value.
Their performances demanded different forms of evaluation. The established veterans presented another challenge entirely.
LeBron James showed why a résumé still carries weight
At 41, LeBron James entered the selection period with an unmatched history of All-Star recognition.
However, injuries had interrupted his season.
James missed the Lakers’ first 14 games while dealing with sciatica. When coaches chose the reserves on February 1, he had appeared in 30 games.
According to the Lakers’ official announcement, he averaged 21.9 points, 5.8 rebounds, and 6.6 assists.
Those figures remained impressive, although several younger Western Conference candidates had contributed more total minutes.
An analyst emphasizing per-game production could make a strong argument for James.
A model rewarding total season contributions would have to account for his missed time.
Coaches valued what the box score didn’t fully explain
James still controlled possessions through passing, pace, and decision-making.
He recognized defensive rotations before they developed, organized teammates, and generated scoring opportunities late in games.
The Lakers’ official release noted that he scored 109 fourth-quarter points in January, tying Kevin Durant for the league lead that month.
That production strengthened his contemporary All-Star case without requiring voters to rely entirely on his career achievements.
Yet his reputation remained impossible to separate completely from the discussion.
James had spent more than two decades demonstrating elite decision-making against NBA defenses.
Statistical projections could incorporate his historical performance when estimating future impact. Coaches also brought years of experience preparing against him.
Neither approach automatically guaranteed another invitation.
Western Conference coaches selected James as a reserve, extending his record to 22 consecutive All-Star selections.
The starting vote told a different story.
For the first time since his rookie season, James wasn’t an All-Star starter.
Fans, current players, and media members had placed five Western Conference candidates ahead of him. Coaches subsequently included him among their seven reserves.
The different results exposed another problem for forecasters: the starting ballot and reserve selections rewarded players through separate electorates.
Stephen Curry revealed another voting divide
Golden State’s Stephen Curry provided a useful comparison.
He finished third in the Western Conference fan rankings and second among players. The media panel placed him sixth.
Those differences didn’t prevent Curry from earning a starting position.
His enormous following and the respect of fellow players helped establish his weighted ranking.
However, a forecasting system using only current-season impact measures wouldn’t automatically reproduce those three sets of preferences.
Popularity didn’t erase Curry’s basketball credentials.
His shooting continued to stretch defenses and create space for Golden State’s offense.
Still, the results showed why an All-Star ballot couldn’t function as a straightforward statistical ranking.
The voters weighed performance, recognition, availability, and established trust differently.
Then Kawhi Leonard exposed a complication that had little to do with the voters at all.
Kawhi Leonard was the selection no reserve ranking could explain
When NBA head coaches finalized their reserve choices on February 1, they left Kawhi Leonard off the original 24-player All-Star roster.
His omission immediately raised questions.
The Clippers forward had been scoring at an elite level while contributing defensively.
According to NBA Communications, Leonard averaged 27.6 points, 6.1 rebounds, and 3.6 assists through 36 games when the league announced his eventual addition.
He also led the NBA with 2.06 steals per game.
His shooting percentages reached 49.7 percent from the field, 39.1 percent from three, and 91.3 percent at the foul line.
Those numbers supported a formidable All-Star case.
Leonard had also scored 55 points against Detroit on December 28, demonstrating the kind of individual brilliance that routinely earns recognition.
However, coaches chose seven other Western Conference reserves.
Two days later, Commissioner Adam Silver added Leonard to the All-Star player pool.
The explanation came from the competition’s rules, not a previously unpublished statistical ranking.
The commissioner wasn’t filling an ordinary reserve spot
The distinction matters because Silver didn’t overturn a coaches’ selection.
He also didn’t replace one of the original seven Western Conference reserves.
Instead, the commissioner used separate authority established under the NBA’s new All-Star format.
The 2026 competition featured two American teams and one international team, with each squad requiring at least eight players.
In its November 2025 format announcement, the NBA established minimum representation of 16 U.S. players and eight international players.
The league gave Silver authority to add participants if the initial selections failed to meet those requirements.
After the coaches completed the original 24-player roster, the American pool remained one player short.
Silver chose Leonard on February 3, bringing the U.S. group to the required 16.
That decision made Leonard the 25th selected All-Star, before injuries produced additional changes.
He occupied a commissioner-selected addition, not a standard reserve position. The NBA didn’t publish a ranked commissioner ballot that analysts could treat as an eighth reserve selection.
Leonard’s individual credentials remained strong, but the roster requirement created the opportunity for his inclusion.
A model ranking the NBA’s best players could identify Leonard’s value. Predicting his addition required understanding the commissioner’s authority and the event’s format.
For forecasters, this wasn’t another disagreement over player performance.
It was a different selection mechanism altogether.
Injuries produced another version of the All-Star roster
Even Silver’s addition didn’t settle the final lineup.
The NBA named its starters on January 19. Coaches completed their reserve selections on February 1, and Leonard joined the player pool two days later.
Then injuries forced more changes.
Shai Gilgeous-Alexander, Stephen Curry, and Giannis Antetokounmpo couldn’t participate.
The league brought in Alperen Şengün, Brandon Ingram, and De’Aaron Fox as replacements.
Organizers also moved Norman Powell to Team World while adjusting the groups.
Those changes created a problem for anyone reviewing the accuracy of a preseason or midseason forecast.
Which version of the roster should count?
Selection accuracy and participation accuracy are not the same
A model predicting the original starters should face the January 19 results.
Meanwhile, a reserve forecast must be compared with the coaches’ February 1 choices.
The final participating teams present another question.
Predicting Fox’s eventual appearance requires information about injuries and replacement decisions, not simply his statistical standing against the original candidates.
Likewise, Leonard’s commissioner selection belongs in a separate category from the coaches’ reserve ballot.
Combining those outcomes can distort a model’s apparent accuracy.
Analysts shouldn’t reward a forecast for identifying an eventual injury replacement if its original assignment was to anticipate coaching selections.
Nor should they penalize a model for predicting an injured starter who legitimately earned his place before withdrawing.
The 2026 NBA All-Star roster predictions involved three distinct outcomes: the original selections, the expanded player pool, and the final participants.
Keeping those results separate provides a fairer test of any statistical forecast.
It also protects against a familiar problem in sports analytics: changing the definition of success after learning the result.
The next generation of models must predict voters as carefully as players
The 2026 selections exposed the strengths and limitations of basketball forecasting.
Wembanyama and Edwards showed how a tiny difference in fan support could settle an otherwise tied starting race.
Avdija and Johnson demonstrated how quickly emerging stars could challenge established statistical expectations. Meanwhile, Duren’s selection highlighted the value of defense and rebounding beyond scoring totals.
Leonard presented an entirely different complication.
His elite production supported a strong All-Star argument, but an administrative roster requirement ultimately brought him into the event.
James and Curry added another dimension through established reputations, availability, and different voting preferences.
Future forecasting systems will need to account for those differences without pretending that one formula can explain every decision.
The process should begin with a fixed evaluation date.
Analysts can use EPM, BPM, DARKO, and traditional statistics to estimate current performance and expected impact. Minutes played, team success, and availability provide additional evidence.
However, the next stage must address the people casting the ballots.
Fan forecasts need information about popularity and voting behavior. Player rankings reflect another group’s preferences, while media panels may emphasize different aspects of performance.
Coaches bring their own evaluations to the reserve selections.
The final model must also account for competition rules, nationality requirements, and any additional selection authority held by the commissioner.
None of those steps guarantees a perfect result.
They do make the assumptions visible, allowing analysts to distinguish a strong basketball argument from an actual probability of selection.
A responsible forecast should also acknowledge uncertainty.
Identifying 24 deserving players isn’t the same as predicting the 24 names the NBA will announce. A model needs to preserve its original predictions rather than adjust them after the fact.
The next generation of NBA All-Star prediction models won’t succeed by finding one perfect impact statistic. It will succeed by separating basketball value from the probability of receiving an invitation.
The difference becomes impossible to ignore when a starting position comes down to a few thousand votes.
Wembanyama and Edwards collected nearly four million fan votes between them. Their statistical cases supported a legitimate basketball debate, but neither scoring averages nor advanced impact ratings determined the final starting spot.
The NBA’s rules sent the decision back to the fans.
Four thousand, five hundred and five votes decided the starting spot. The box score never had the final say.
READ MORE: NBA 3-Point Contest Shooting Mechanics From Bird to Lillard
FAQs
Why did Victor Wembanyama start instead of Anthony Edwards in the 2026 All-Star Game?
Both finished with a 5.75 weighted voting score. Wembanyama won the tiebreaker because he received 4,505 more fan votes.
How did the NBA choose its 2026 All-Star starters?
Fans controlled 50 percent of the vote. Current NBA players and a media panel each contributed 25 percent.
Can advanced statistics accurately predict NBA All-Star selections?
Metrics such as EPM, BPM and DARKO help evaluate performance. However, they cannot predict every fan preference, coaching decision or roster adjustment.
Why did Kawhi Leonard make the 2026 NBA All-Star roster?
Commissioner Adam Silver added Leonard after the original 24 selections to satisfy the new format’s requirement for at least 16 U.S. players.
Which breakout players earned their first NBA All-Star selections in 2026?
Deni Avdija, Jalen Johnson and Jalen Duren all earned their first selections. Their contrasting playing styles challenged conventional player rankings.
