OluSeyi Gideon Olaniyi

"Can a transparent, expert-weighted model paired with Monte Carlo simulation reliably forecast Nigerian presidential elections? " Every election season, the arguments sound the same. People argue from feeling, from loyalty, and sooner or later someone says the result is already decided. I wanted something better to argue from, so I built a model that shows its working. I ran 50,000 Monte Carlo simulations of the race between Bola Tinubu (APC), Atiku Abubakar (ADC) and Peter Obi (NDC). Here’s what came out: - Tinubu wins 79 of every 100 simulated elections - Obi wins about 11 - Atiku wins about 10 - 1 in 4 runs goes to a run-off The real story is two forces pulling in opposite directions. Being the sitting president is worth about 12 points to Tinubu. The pain of prices, hardship and insecurity costs him about 11.5. They almost cancel each other out. The run-off rule matters more than most people think. To win outright, a candidate needs the most votes and at least 25% in 24 of the 37 units (36 states plus the FCT). When nobody manages that, the picture changes completely. In the simulated run-offs, Tinubu and Obi each win 37% and Atiku 26%. If the model is right about what moves the result, each camp has one lever that matters most: - Tinubu: cost of living. A credited reform dividend lifts his chance from 79% to 91% - Obi: turnout. At around 42% turnout, his chance climbs from 11% to 34% - Atiku: win the North decisively and force a run-off, which lifts him to 27% Nearly eight in ten is a strong favourite, not a certainty. A forecast isn’t a prophecy. It’s a set of assumptions one can inspect and challenge, and nothing here is an endorsement of any candidate or party. This is a living model, and I’ll keep updating it as new data comes in between now and (Jan 16 2027) polling day. Once INEC declares the official results, I’ll run a full post-election evaluation, treating the actual outcome as true out-of-sample ground truth. I’ll measure calibration with the Brier score, error on vote share at national and state level with mean absolute error, and how accurately the model called each state against the 25% threshold. If it holds up, it becomes a baseline for future elections. If it doesn’t, the errors will show exactly which assumptions need rethinking. Either way, the evaluation will be published openly. Check comment section to read the reports. #Nigeria2027 #DataScience #ElectionForecasting #MonteCarlo #Nigeria #Elections