The Prince Returned, and Our Numbers Called It

IJ Ventures
17 hours ago
12 min read

Bangladesh just held the biggest democratic exercise of 2026. We built a forecast from scratch. Here's how it held up.

The New York Editorial · Quantitative Analysis Desk Published February 18, 2026

DHAKA — On February 12, 2026, Bangladesh voted. And it wasn't close.

Read our detailed analytical report here: https://thenyeditorial.github.io/2026-Bangladesh-Post-Election-Analysis/

The Bangladesh Nationalist Party, led by Tarique Rahman, took 209 of 297 declared seats in the 13th Jatiya Sangsad, handing the party a two-thirds supermajority in the country's first genuinely competitive election since 2008. Rahman, who spent 17 years in self-imposed exile in London before returning in December, will become Prime Minister. He has never held government office.

This was the first general election since the July 2024 uprising, when Gen Z protesters toppled Sheikh Hasina after fifteen years in power. Her Awami League, winner of four straight elections, was banned from the ballot entirely. That absence reshaped the race from its foundations. What would have been a three-way fight became a straight contest between BNP and the Jamaat-e-Islami-led 11-party alliance. Voters chose the familiar over the untested. They chose stability.

Five days before polls opened, we published a data-driven forecast. We got the big calls right.

We projected BNP at roughly 185 seats, with a range of 155 to 215. The actual: 209. Inside the range. We projected the Jamaat alliance at roughly 80 seats, with a range of 55 to 110. The actual: 77. Three seats off our central number. We estimated turnout at about 60%. The Election Commission reported 59.88%. We gave BNP a 70% chance of winning an outright majority. They won a supermajority.

Both major alliances, turnout, the winner, the direction of travel. All called. What follows is the full story of what the data got right, where the margins moved, and the deeper currents that carried Bangladesh to this result.

The Official Results

Of 299 seats contested (one was suspended, two more withheld by court order in Chattogram), the Election Commission declared results for 297.

BNP won 209. That's 70.4% of declared seats. Jamaat-e-Islami took 69, its best result in any general election ever. The National Citizen Party, the youth outfit born directly from the July uprising, won 6 of 30 seats it ran in. Eight independents won, including Rumeen Farhana, who'd been expelled from BNP. The last 5 seats went to smaller parties: Khelafat Majlish took 2, with Islami Andolon Bangladesh, Gono Odhikar Parishad, and Gonoshonghoti Andolon picking up 1 each.

Roll it up to the alliance level: BNP and allies, 212 seats. Jamaat's 11-party bloc including NCP, 77.

Turnout hit 59.88%, a sharp jump from the boycott-depressed 41.8% in 2024. Over 50 parties fielded candidates, a national record, and more than 2,000 individuals were on the ballot.

There was also a referendum. Voters were handed a pink ballot on the July National Charter, a package of constitutional reforms: PM term limits, a bicameral parliament, stronger judicial independence, a permanent caretaker government system for future elections. It passed. 62.74% of valid votes said "Yes."

Our Forecast vs. Reality

We built the forecast on seven national surveys covering over 80,000 respondents, then calibrated it against 30 years of FPTP seat-conversion data from every Bangladeshi election since 1991. Here is how each call landed.

BNP seats. We said roughly 185, range 155 to 215. Actual: 209. That's 24 above our central number, but well inside the published range. We set the upper bound at 215 specifically because we modeled a scenario where undecided voters broke hard for BNP and the rebel candidate effect fizzled. That's exactly what played out. Grade: A.

Jamaat alliance seats. We said roughly 80, range 55 to 110. Actual: 77. Three seats below center. Less than a 4% deviation. This was our tightest call. Grade: A+.

Turnout. We said about 60%. Official number: 59.88%. Off by 0.12 points. Grade: A+.

BNP majority. We gave it a 70% probability. BNP cleared not just the 151-seat majority line but the 200-seat supermajority threshold. Right call, bigger magnitude than we expected. Grade: A.

BNP supermajority. We put this at 25% probability. It happened. The 209-seat result sits almost dead center in our supermajority scenario of 200 to 220. Not our most likely outcome, but one we explicitly mapped. Grade: A−.

FPTP magnification. We flagged this repeatedly before the vote, pointing out that in 2001 a 0.9-point gap in vote share had produced a 131-seat gap in parliament. Sure enough, BNP's estimated 45% of the popular vote converted into 70.4% of seats. A multiplier of about 1.56x. The system did what the system does. Grade: A.

Rebel candidates. We counted 92 BNP rebels running in 79 constituencies and modeled them costing the party 15 to 25 seats. Only 8 independents won nationally. Rahman's personal authority, combined with BNP's ground-level discipline, shut the rebels down almost everywhere. We spotted the right risk. We overcooked the size of it. Grade: B.

NCP seats. We estimated about 15. They won 6. This was our biggest miss. The NCP's decision to hitch itself to Jamaat drove away high-profile founders like Tasnim Jara and Tasnuva Jabeen, and it confused the party's whole identity. We didn't fully account for how badly that alliance would damage them. Grade: B−.

Overall: A−. Both major alliances landed inside our ranges. Turnout was dead on. The winner, the majority, the structural dynamics, the FPTP effect. All correct. The one real miss was overrating the NCP. On the questions that matter most in election forecasting, the model did its job.

What the Polls Told Us

The most important single survey in our stack was the Eminence Associates for Social Development (EASD) projection, which estimated about 208 BNP seats. The final count was 209. EASD undershot Jamaat at about 46, though. That gap hints at something we see globally with Islamist parties: their supporters tend to understate their preference in surveys, but show up and vote. Jamaat's ground operation outperformed its polling numbers, and it has done this before.

The Communication & Research Foundation, working with Bangladesh Election and Public Opinion Studies, gave us our single most useful structural input. Their finding: nearly half of former Awami League voters had moved to BNP, with a meaningful chunk going to Jamaat. When the party that held 40% of the electorate for fifteen years vanishes from the ballot, those voters don't just stay home. They redistribute. That redistribution decided the election. It wasn't an ideological swing. It was a structural one, forced by AL's absence.

The Bangladesh Youth Leadership Center ran a youth survey in late 2025 that revealed something we should have weighted more heavily. Young voters did gravitate toward Jamaat early on, especially on campuses. Jamaat swept all four major university student elections in 2025. But as the actual election approached, the trend reversed. Journalist David Bergman tracked two competing pulls: BNP's tough posture toward AL at the local level pushed some former AL voters toward Jamaat as the softer alternative, while Jamaat's protest branding pulled in idealistic students. In the end, pragmatism won. The same generation that burned tires in July 2024 walked into polling stations in February 2026 and voted for the party they thought could actually run the country.

One data point worth sitting with: EASD nailed BNP at 208 but put Jamaat at just 46. Our model placed the alliance at 80 and caught the actual 77 almost exactly. The reason is that we applied a historical correction for Islamist-party survey underperformance. That single adjustment turned out to be one of the most valuable things in the entire model.

Seven Forces Behind the Numbers

1. The Stability Mandate

Two years of upheaval. An uprising. A regime collapse. An interim government. An extortion crisis that swept the country. Rising prices. None of that creates an appetite for more experimentation. Voters wanted someone who could turn the lights back on and keep them on. BNP, whatever its flaws, looked like the party capable of doing that. As The Researchers wrote in their post-election analysis, the generation that ignited upheaval ultimately voted to institutionalize stability. The kids who started the revolution didn't vote for revolution. They voted for jobs.

2. Awami League's Absence and Where Its Voters Went

This was the single biggest variable in the entire election, and it had nothing to do with campaigning. When the party that won the last four elections is banned, roughly 40% of the historical electorate needs somewhere to go. The Communication & Research Foundation found that about half went to BNP. A smaller but significant share went to Jamaat. The rest sat it out or scattered. BNP absorbed the center. Jamaat absorbed the protest vote. That's not ideology at work. That's math.

3. The Homecoming

Tarique Rahman landed in Bangladesh in December 2025 after 17 years abroad. The effect was immediate. His rallies drew millions. BNP's organizational machinery, which had been dormant under Hasina's repression, snapped back to life the moment it had a leader in the country again. He won Bogura-6 with 216,284 votes, more than double his closest rival. CNN called it a remarkable turnaround for a man who only returned weeks before the vote. The son of Ziaur Rahman, he is now set to become Prime Minister at 60, without ever having held government office.

4. The NCP's Alliance Mistake

The National Citizen Party was born from the uprising. Its founders were the student leaders who put their bodies on the line in July 2024. Then, ahead of the election, the party joined Jamaat's 11-party alliance. For a secular, youth-driven movement to align with Bangladesh's main Islamist party was jarring. IUB lecturer Khandakar Tahmid Rejwan said NCP leader Nahid Islam had effectively sold his political autonomy. Founding members Tasnim Jara and Tasnuva Jabeen resigned in protest. Jara ran as an independent in Dhaka-9, pulled in over 44,000 votes, and still lost to the NCP's own candidate. The youth vote splintered. NCP won 6 of 30 seats. Twenty percent. Street power, it turns out, does not automatically become ballot-box power.

5. The Brutality of First-Past-the-Post

This is the structural story of the election, and it deserves more attention than it usually gets. Under FPTP, you don't need a majority of votes. You just need more votes than the next person in each constituency. BNP's support was spread efficiently across all 291 seats it contested. Jamaat's support was deep in certain pockets (Pabna, Kurigram, parts of Gazipur) but thin elsewhere. The result: BNP's estimated 45% of the popular vote translated into 70.4% of seats. A multiplier of 1.56x. Jamaat's roughly 25% of votes got them 23.2% of seats. And everyone else, combined? About 30% of votes, 6.4% of seats. We flagged this dynamic before the election, citing the 2001 precedent where a 0.9-point vote gap between BNP and AL produced a 131-seat gap. The system did it again.

6. AI Disinformation at Industrial Scale

Between August 2025 and February 2026, the US Center for the Study of Organized Hate identified over 700,000 AI-generated social media posts targeting the Bangladesh election. The Election Commission flagged about 86,000 pieces of disinformation, nearly 36,000 of them violent in nature. Deepfakes targeted both Tarique Rahman (133 instances) and Jamaat leader Shafiqur Rahman (54 instances). Fabricated quotes, manipulated video, synthetic audio. This was, by most measures, the first major "Gen Z election" to face AI manipulation at this scale. Whether it actually changed outcomes is hard to say. The election went off peacefully, which suggests most voters discounted the noise. But the scale of it is worth recording.

7. The Referendum

The pink ballot. Alongside the parliamentary vote, citizens decided whether to approve the July National Charter, the constitutional reform package born out of the 2024 uprising's demands. PM term limits. A bicameral parliament. Stronger courts. A permanent caretaker government for election periods. It passed with 62.74%. But not without controversy: some constituencies returned turnout figures that didn't add up. Rajshahi-4 reportedly logged 244.3% turnout, with votes dramatically exceeding registered voters. BNP had initially resisted the charter but flipped to "Yes" on January 30, a late pivot that Jamaat was quick to call opportunistic. The new parliament now has 180 working days to implement the reforms.

Where the Margins Moved

Both major alliance results landed within our ranges. But our central BNP number of 185 undershot the actual 209 by 24 seats, and that gap is worth understanding.

We gave too much weight to three things that all pointed toward a tighter race. First, the rebel candidates. We modeled 92 BNP rebels in 79 seats costing the party 15 to 25 seats. In practice, BNP's discipline held. Rahman's authority kept the rebels from doing serious damage. Second, Jamaat's campus sweep. They won all four major university student elections in 2025, and we treated that as a signal of broader youth momentum. It wasn't. University students don't represent the wider 18-to-37 cohort, which cares more about getting a job than about ideological alignment. Third, the undecided voters. Seventeen percent were undecided in late January polls. We split them 55/30/15 in favor of BNP/Jamaat/others. They actually broke much harder for BNP, possibly 70% or more.

That last point matters. The turnout numbers tell the story. At 2:00 PM, the Election Commission was reporting 47.91%. By close of polls, the final number was 59.88%. That late surge, roughly 12 points in two and a half hours, was overwhelmingly made up of voters choosing the perceived winner. Bandwagon effects are well-documented in FPTP elections. Bangladesh 2026 was a textbook case. Our range of 155 to 215 was built to capture exactly this kind of scenario. The upper bound existed for a reason.

The biggest clean miss was the NCP. We said about 15 seats. They got 6. Their alliance with Jamaat, which cost them key founders and scrambled their identity, was a late development we didn't fully price in. As Professor Nazmul Islam Nazrul of Dhaka University told Al Jazeera, the NCP's result was "not insignificant given its limited experience in electoral politics." Fair enough. But it was well short of what the revolutionary moment seemed to promise.

Turnout: 59.88% vs. Our ~60%

This was the number. A miss of 0.12 points, and arguably the most important prediction in the entire forecast.

In 2024, turnout under Hasina was 41.8%, depressed by the BNP and Jamaat boycott. This time, nearly 60% of 127.7 million registered voters showed up. That's roughly 18 million more people casting ballots than in 2024. The Election Commission called it one of the most peaceful and credible elections in decades. The EU observation mission, with over 200 observers on the ground, reported strong enthusiasm. Jamaat's alliance filed irregularity complaints at certain polling centers, but the overall picture was one of a country that took the vote seriously.

Higher turnout helped BNP. That was our thesis going in, and the data confirmed it. The extra voters weren't Jamaat supporters who'd been sitting on the fence. They were ordinary Bangladeshis who wanted a functioning government and saw BNP as the safest bet.

What Comes Next

Tarique Rahman will be Prime Minister. His two-thirds majority gives him the constitutional muscle to push through or reshape the July Charter reforms. But supermajorities are a double-edged thing. BNP owns every outcome now. There is no coalition partner to absorb the blame when things go wrong.

The problems waiting on his desk are serious. Inflation has battered household budgets. The garment sector, which makes Bangladesh the world's second-largest clothing exporter, has stalled. Youth unemployment is acute: according to the Bureau of Statistics, 87% of the unemployed are educated and 21% are graduates. Corruption sits deep in the system (Transparency International ranks Bangladesh 152nd out of 182). Extortion networks grew rapidly after the uprising. And then there's India. Hasina is still there, in self-imposed exile. The relationship between Dhaka and New Delhi is frayed, and China has been happy to fill the gap.

The referendum passed, so the new parliament must act as a constituent assembly and deliver constitutional reforms within 180 working days. That's the clock.

For the opposition, Jamaat's 69 seats are a historic high but probably also a ceiling under current conditions. The party accepted the results on February 15, pledging to serve as a "vigilant, principled, peaceful opposition." The NCP's 6 seats give the revolutionary generation a small but real voice in parliament. Spokesman Asif Mahmud said they'd rebuild and focus on local elections due in a year.

Muhammad Yunus, the Nobel laureate who led the interim government, offered the last word before handing over. "We have handed the keys of the house back to the people," he said. "Our task was to repair the locks; now it is up the elected representatives to build the home."

-----------------------------------------------

Methodology: Seven national surveys, 80,000+ aggregated respondents: EASD seat projection, CRF/BEPOS voter migration analysis, BYLC youth survey, SHUJAN proportional representation poll, and three party-commissioned surveys. Historical calibration against 30 years of FPTP seat-conversion data (1991 to 2024). Structural adjustments for AL voter redistribution, rebel candidate effects (92 rebels, 79 constituencies), undecided voter allocation (55/30/15 BNP/Jamaat/other), turnout elasticity (+1 pp ≈ +2–3 BNP seats), Gen Z campus extrapolation correction, and geographic weighting for rural/urban splits. Confidence intervals: ±30 seats for BNP (155–215), ±27.5 for Jamaat alliance (55–110).

Sources: Bangladesh Election Commission unofficial results gazette, Feb. 14, 2026, via BSS · Dhaka Tribune constituency tracker · Al Jazeera, Feb. 13–15, 2026 · CNN, Feb. 12–13, 2026 · The Daily Star, Feb. 12–13, 2026 · bdnews24.com, Feb. 13, 2026 · Wikipedia: 2026 Bangladeshi general election; 2026 Bangladeshi constitutional referendum · DVB/Reuters, Feb. 13, 2026 · The Researchers, Feb. 13, 2026 · Zee News, Feb. 13, 2026 · CBC News, Feb. 11, 2026 · The Diplomat, Feb. 12, 2026 · India TV, Feb. 12–13, 2026 · MediaBangladesh.net, Feb. 13, 2026 · Sunday Guardian Live, Feb. 13, 2026 · EASD seat projection, via Dhaka Tribune, Jan. 2026 · CRF/BEPOS voter migration study, Jan. 2026 · BYLC youth survey, Oct.–Nov. 2025 · David Bergman, AL voter migration analysis · US Center for the Study of Organized Hate, Aug. 2025–Feb. 2026 · Transparency International Bangladesh, 2025 · EU Election Observation Mission, Feb. 12, 2026 · Rumor Scanner Bangladesh, 2025–2026.

All polling data, election statistics, historical tables, and probability assessments underlying both the original forecast and this post-election analysis were compiled and analyzed by Muhammad Rakibul Islam (Rakib). Raw dataset available under MIT License.