Methodology
No real polling. No pollster has published a probability sample of the 2027 race: no NOI Polls, no Afrobarometer question on it, and Wikipedia’s article on the election has no polling section. The published figures come from online polls that anyone could answer. This page averages them anyway, so that the model has numbers to run on, but they can say little about how Nigerians will vote. In 2023 the online polls overstated Peter Obi badly: several put him first, and he came third with 25%. Treat every figure here as a scenario, not a forecast.
The poll average. Polls are kept by hand from the press, because Wikipedia has no table to read. The latest poll from each company counts, weighted by recency: the weight halves every few days before the newest poll, under the site-wide rule (). Polls left with under 1% of the weight are dropped. Each poll is rescaled to the votes for a candidate. Tinubu, Obi and Atiku are averaged straight; everyone else shares what is left. A candidate tested in under a quarter of the weight goes into Others: Sandy Onor, the PDP’s candidate, isn’t named in any published figures yet, so his vote is inside Others. Polls run by a campaign are listed greyed and left out: the Tinubu Must Go movement’s September poll is one.
The states. Polls are national only. Each state starts from its own 2023 presidential vote, handed on to the 2027 candidates by the shares in the table below (Kwankwaso is Obi’s running mate, so most of his 2023 vote goes to Obi). Makinde, Onor and Sowore get a home-state lift: Makinde ×12 in Oyo and ×3 elsewhere in the South West, Onor ×8 in Cross River, Sowore ×2 in Ondo. Then each candidate is swung by one amount in every state until the national total, weighted by each state’s 2023 vote, equals the poll average: in log-odds by default, or in points (Uniform) or by a ratio (Proportional) on the switch. So the national figures come from the polls alone; the table and the home lifts only shape the regional pattern, and the map is an estimate, not a poll.
The 25% rule. Section 134 of the Constitution: the most votes, and not less than a quarter of the votes in each of at least two-thirds of all the states and the Federal Capital Territory. The tribunal and the Supreme Court read that in 2023 as 24 of 37, with the FCT counted like any state rather than as a separate requirement, and this page does the same. If nobody clears it, the second election is between the leader and “one among the remaining candidates who has a majority of votes in the highest number of states”, read here as the candidate other than the leader who is ahead in the most states (ties broken by votes).
Atiku off the ballot. On 5 November 2026 the Federal High Court in Abuja voided the ADC leadership under David Mark and the decisions it took, which Atiku’s opponents say include his nomination; the ADC says it doesn’t, and has appealed and asked for a stay. INEC says it will follow the final ruling. The switch hands Atiku’s vote on, state by state, by zone (table below), on the assumption that the ADC fields no presidential candidate. The Senate follows the switch too, as the ruling may reach the ADC’s other nominations.
The runoff. No poll has tested a head-to-head, so the runoff hands every knocked-out candidate’s voters on: “Second choices” uses a hand-set table of where each candidate’s voters lean (below), with 15% staying home; “Change vote” sends a fifth more of the opposition’s voters against the incumbent; “Incumbent holds” the reverse, as governors and the federal machine turn out for Tinubu. The table can be typed into. The 2CP % on the runoff map is set against the same two camps’ two-candidate share in 2023 (Obi + Kwankwaso for Obi).
The Senate. Every district’s 2023 vote, as declared at its collation centre (Civic Hive’s results, from the declaration sheets), is handed on the same way as the presidential vote: APC to the APC, the PDP’s vote by Atiku’s row (most of it to the ADC), Labour’s and the NNPP’s by Obi’s and Kwankwaso’s rows. Small parties (APGA, SDP, YPP and the rest) keep 60% of their 2023 vote as their own, unswung, and hand the rest on by the “others” row. Then every party is swung by the same amounts as its presidential candidate. So the Senate moves with the presidential race: there is no polling of it, and the model knows nothing of the 2027 Senate candidates. The ± compares with the 2023 seats by party name. The ADC and NDC were not contenders in 2023, so they gain from zero; the PDP’s loss is its 2023 seats going to the parties its voters went to.
Senate now. Since 2023 most opposition senators have crossed to the APC: after four by-election winners were sworn in in June 2026 the press counted the APC on 88, the ADC on 9, the PDP on 5, the NDC on 4 and APGA, Labour and Accord on 1 each, with no seat vacant (Vanguard, 24 June; ThisDay, 25 June). The National Assembly’s own list is out of date, no reliable tally after June was found, and the Senate disputed Enyinnaya Abaribe’s move to Labour. The “Senate now · Nov 2026” view on the Senate tab draws it. Those moves aren’t in the projection, which runs on the 2023 vote, not on who sits where. The governors who moved with them (in Delta, Akwa Ibom, Enugu and elsewhere) are why the APC may do better in the South than this model shows.
The compass. Each party “answers” the 130 statements of the compass quiz, graded on its record and platform; the same grades decide left and right in the seat charts, by the site-wide rule (an invisible 45° line through Nigeria’s average party). Every major party backs the 2014 law against same-sex marriage, so the social axis is crowded at the conservative end. The PDP’s grades are low confidence after the party’s split, and so are the AAC’s.
Sources. 2023 results: Wikipedia, “2023 Nigerian presidential election” (by state; Kwara’s and Yobe’s rows don’t add up, so their valid votes are the sum of the candidates), and Civic Hive’s 2023 senatorial results; 2023 Senate winners from Wikipedia, “2023 Nigerian Senate elections”. Candidates: INEC’s final list, via Wikipedia, “2027 Nigerian general election”. Polls: the publications linked in the poll table.