← Map

METHODOLOGY

Every assumption published. Every error acknowledged.

How We Build the Projections

Every assumption is published.
Every error is acknowledged.
Every projection has a confidence level.
Forward-looking analysis without honest backtesting is prediction theatre.
Validation

We tested this methodology against two past elections before applying it to 2027

Backtest
2011 → 2015
Winner called✓ Correct
State accuracy78.4% (29/37)
National error
buhari0.2 pp
jonathan0.2 pp
Party consolidations are predictable. 0.2pp national error when transitions involve known actors and stable ethnic alignments.
Backtest
2019 → 2023
Winner called✓ Correct
State accuracy56.8% (21/37)
National error
tinubu10.2 pp ⚠
obi9.4 pp ⚠
atiku0.7 pp
kwankwaso1.4 pp ⚠
Methodology underestimates new political forces emerging from near-zero (Obi). Overestimates vote transfer when machine votes are involved (Atiku). Fixed in 2027 model.

Our 2027 projections have a confidence ceiling of 78.4% state-level accuracy and a floor of 56.8%. 2027 resembles 2011→2015 more than 2019→2023 — consolidation of existing forces rather than new movement emerging. Confidence sits closer to ceiling than floor.

Model Updates

Two findings that changed the model

Finding 1 — PDP machinery is stickier than models assume

The 2019→2023 backtest showed that PDP's institutional base in Bauchi (426,607 votes), Kaduna (554,360), Katsina (489,045), and other northern states delivered votes beyond what trendlines predicted — even as Atiku's national numbers collapsed to a career low.

This confirmed what the political evidence already suggested: Atiku's 2023 PDP votes were the machine's votes, not his personal votes. When he left PDP those votes stayed with PDP.

Action taken: Revised Atiku's personal transfer rate to ADC downward. Northeast: 10%→8% · Northwest: 8%→6% · Adamawa unchanged — personal loyalty confirmed.
Finding 2 — Anti-incumbent sentiment is stronger than models typically capture

The 2011→2015 backtest showed anti-incumbent sentiment worth approximately 15 percentage points against Jonathan — higher than standard models assume. Our current APC erosion factor of 9% is deliberately conservative.

If 2027 economic conditions resemble 2015 — sustained fuel price crisis, currency pressure, unemployment — the actual erosion could reach 14.000000000000002%.

Action taken: Replaced the fixed 9% erosion rate with a reviewable range of 814.000000000000002%. Current setting: 9% (April 2026) · Next review: July 2026
The Polling Unit Dataset

A second verification layer — 170,000 sheets vs INEC

In 2023 Mark Essien and Nigerian volunteers transcribed 170,000 polling unit result sheets into spreadsheets in 5 days. The data was used as evidence in Peter Obi's presidential tribunal petition.

We obtained all 37 state files from the public Google Drive and ran a systematic comparison against INEC's official collation results. Coverage: 100% Results_Found across all states, though 4 states (Sokoto, Taraba, Yobe, Zamfara) show data quality issues suggesting most sheets were unreadable.

What we found: 7,154,064 votes appear in INEC's official results that are not in the polling unit sheets — a 44.1% national inflation. In 33 of 37 states this inflation affected all parties roughly proportionally and winners were consistent. In Rivers State, the pattern was different.

Rivers PU winner
LP (54.9%)
Rivers INEC winner
APC (46.6%)
LP votes removed
−25,336
APC votes added
+135,515

We use this dataset as a verification layer — not a replacement for official results. Where the two datasets are consistent, they reinforce each other. Where they diverge, we show both and let readers judge. The tribunal dismissed Obi's petition. Our role is to show the data.

Full gap analysis → /polling-units
Data Sources

Where the numbers come from

INEC National Collation Centre
Official 2023 presidential results
Accessed via Dataphyte PowerBI dashboard · All 37 states · Cross-checked against INEC PDF reports
geoBoundaries / GRID3
Nigeria state and LGA boundary files
Coverage: 37 states, 774 LGAs · License: CC BY 4.0
Citizen Report Vol II (TechCabal/Zikoko 2026)
Youth voter sentiment and turnout intentions
Sample: 10,681 young Nigerians, all 36 states + FCT · Used for: turnout scenario assumptions
Vote Blocs

Three vote blocs. Three different models.

Bloc 1 — LP / Obi → ADC

Peter Obi moved from LP to ADC. His 2023 votes were a mix of genuine personal preference and ethnic solidarity. The ethnic solidarity component (Southeast) transfers at high rates regardless of party vehicle. The genuine preference component (Middle Belt, FCT, Plateau) transfers at medium rates. The protest vote component (Lagos, Ogun) transfers at medium-low rates — those voters chose Obi specifically; whether they follow him to ADC depends on how strongly they engage.

ZoneRateReasoning
Southeast88–90%Ethnic solidarity follows the man
South-South65–68%Preference + anti-APC, not pure ethnic
Middle Belt75–78%Christian preference — follows Obi
Southwest70–72%Urban protest vote — medium transfer
Northeast60–62%Scattered LP votes — moderate follow
Northwest60–62%Same
Bloc 2 — NNPP / Kwankwaso → ADC

Kwankwaso's votes are personal movement votes. NNPP was built around him specifically. When he moves, his voters move — particularly in Kano where 997,279 people voted for his movement personally, not for any party ideology. Outside Kano his reach is real but limited.

ZoneRateReasoning
Kano88–90%Personal movement — follows him
NW other55–58%Some network, not as personal
All other26–30%Scattered, limited personal pull
Bloc 3 — PDP / Atiku → ADC

This is the most misunderstood vote bloc in the 2027 analysis. Atiku's 6.98M votes in 2023 were primarily institutional PDP votes delivered by state machinery that did not follow him to ADC. The backtests confirm this. We apply the lowest transfer rates of any bloc — 6–8% in the Northwest and Northeast — with Adamawa as the sole exception where personal loyalty is genuine.

ZoneRateReasoning
Adamawa40–60%Home state — personal loyalty real
Northeast8%Revised down — backtest confirmed PDP machine stays with PDP
Northwest6%Same — Katsina, Jigawa stay PDP
Middle Belt5%Minimal personal following
South3%Near zero
Scenarios

Four scenarios. Four different assumptions about what happens next.

A
Obi-Kwankwaso + Atiku campaigns
MEDIUM
What has to be true

Atiku formally withdraws from presidential ambition. He campaigns actively for the ticket, particularly in the Northeast and his Adamawa network. His endorsement moves some former PDP voters to ADC.

Why this confidence level

LOW but possible.

What would change this: Public announcement from Atiku confirming VP-level role or non-ticket senior position.
B
Obi-Kwankwaso + Atiku neutral
← DEFAULT
MEDIUM-HIGH
What has to be true

Atiku is in ADC but not on the ticket and not actively campaigning. His presence is nominal. Former PDP voters in the North largely stay with whoever PDP runs or stay home.

Why this confidence level

Most consistent with Atiku's documented political behaviour and current signals.

What would change this: Atiku publicly campaigning → move to Scenario A. Atiku running as candidate → move to Scenario C.
C
Atiku leads ADC ticket
LOW-MEDIUM
What has to be true

ADC primary or negotiation produces Atiku as presidential candidate. Obi accepts VP or campaigns nominally. The Obidient movement does not transfer enthusiasm to an Atiku-led ticket.

Why this confidence level

Would require Obi to accept subordination to a candidate his own movement rejected in 2023 when given the explicit choice.

What would change this: Internal ADC power dynamics, financial pressure, party structure decisions.
D
Fragmented — 2023 repeats
MEDIUM
What has to be true

ADC negotiations collapse. One or more principals leave. Multiple opposition candidates split the anti-APC vote again. Tinubu wins with 35–38% in a three or four-way race.

Why this confidence level

The incentive structure that produced 2023's split still exists. Nigerian political history offers more examples of failed opposition unity than successful ones.

What would change this: ADC formally confirming Obi-Kwankwaso ticket with Atiku publicly endorsing → move to A or B.
Machinery Model

Why we reduce opposition projections in APC governor states

APC controls 30 of 36 state governments. But not all APC governors are equal. Some won their seats on APC and have full, loyal machinery. Others won on PDP, LP, or NNPP and personally defected — their administrative machinery switched, their grassroots did not. We apply different discounts to each.

ORIGINAL APC GOVERNORHigh reliability machineryDiscount: 4–7%
Higher discount in tight states (7%); lower where opposition margin absorbs it (4%).
States: Lagos, Borno, Kwara, Gombe, Yobe, Jigawa, Katsina, Kebbi, Kaduna, Kogi, Niger, Nasarawa, Ekiti, Ogun, Ondo, Cross River, Benue, Ebonyi, Imo
DEFECTED GOVERNORLow reliability machinery — grassroots stayedDiscount: 1–3%
Governor switched personally. Ward-level structures built under PDP/LP remain opposition-aligned.
States: Delta, Akwa Ibom, Edo, Bayelsa, Enugu, Taraba, Osun (→ Accord)
SPECIAL CASESVery low reliability — structural hostilityDiscount: 1–2%
Kano: APC won via tribunal, not election. Plateau: LP governor who defected — his voters are Obi voters.
States: Kano (1%), Plateau (2%)

We are not making allegations about specific individuals or institutions. We are applying historical base rates from documented patterns in Nigerian result management to forward-looking analysis. Every discount is published here.

Map Legend

What HIGH, MEDIUM, LOW, and TOSS-UP actually mean on our maps

HIGH CONFIDENCE

Opposition margin >20% in 2023. Even accounting for machinery discount and transfer rate uncertainty, the outcome is unlikely to change.

Anambra — opposition 98.4%. No model uncertainty changes this.
MEDIUM CONFIDENCE

Margin 10–20%. Outcome is directionally clear but sensitive to transfer rate assumptions. Could shift one category under different scenario selections.

LOW CONFIDENCE

Margin 5–10%. Machinery discount and transfer rate uncertainty both matter here. One or two percentage points in either assumption changes the call.

TOSS-UP

Margin under 5%. Ground game, turnout, and result management decide this state. Our model cannot call it. Watch these states most closely.

Honest Limits

What we don't know

Whether ADC resolves its ticket before the registration deadline

What PDP does — their candidate choice changes the northern math more than any other variable

What the economy looks like in early 2027 — our APC erosion range is 8–14.000000000000002% depending on this

Whether Obi's ADC energy matches his 2023 LP energy — the vehicle changed

Whether INEC's administration improves from 2023 — our model applies historical base rates for result management

Whether a new political force emerges the way Obi did in 2023 — if it does, our accuracy drops toward 56.8%

We update assumptions as these questions resolve. Every change is logged in the version history below.
Changelog

Every change to this model, logged

DateChangeReason
April 2026Atiku NE rate 10%→8%, NW rate 8%→6%Backtest confirmed PDP machinery stickiness
April 2026APC erosion: fixed 9% → range 8–14%, current 9%Backtest anti-incumbent finding
April 2026Lagos confidence override → LOW regardless of margin2019→2023 backtest miss
April 2026Governor defection flags — 9 states reclassifiedFortune's April 19 data
April 2026Initial model published
Share this methodology
Last updated: April 2026 · Maintained by Fortune Ishaku, Applift Technologies
Data source: INEC via Dataphyte