Seat Projection — 46th Federal Election
343 seats · Majority at 172 · 10,000 Monte Carlo simulations
Poll Tracker
| Riding ↕ | Prov. | Rating ↕ | Projected | LPC% | CPC% | NDP% | BQ% |
|---|
🔬 Scenario Modeler
How Our Model Works
RidingWatch aggregates every available national and regional poll using a multi-stage statistical pipeline. Here's exactly what happens from raw polling data to the seat projections you see on this page.
① Poll Collection
We collect every publicly released federal voting-intention poll from all major Canadian firms: Abacus Data, Ipsos, Léger, Research Co., Nanos, Angus Reid, Mainstreet, Ekos, and Campaign Research. Each poll is logged with its field dates, sample size, and methodology (online panel, IVR phone, live telephone, or hybrid).
② House Effect Correction
Every polling firm has measurable systematic biases — a "house effect" — that cause it to consistently over- or understate a given party. We calculate each firm's historical lean relative to the eventual election result, then apply an additive correction to each new poll before it enters the aggregate. No single pollster can pull the average.
③ Time-Decay Weighting
Older polls contain less information about current voter intent. We apply exponential time-decay with a half-life of 21 days, meaning a poll conducted three weeks ago carries roughly half the weight of one conducted yesterday. Polls older than 90 days are excluded entirely.
④ Sample-Size Weighting
Larger samples have smaller margins of error and carry proportionally more weight. A 3,000-sample poll receives greater influence than a 500-sample poll. We combine time-decay and sample-size weights multiplicatively before computing the final average.
⑤ Regional Swing Model
National vote-share changes are distributed to individual ridings using a uniform swing model calibrated to the regional breakdown (Atlantic, Québec, Ontario, Prairies, BC). Québec is modelled separately to account for the Bloc Québécois's regionally concentrated vote. The 45th federal election (April 28, 2025) serves as the baseline for all swing calculations into the 46th election.
⑥ Monte Carlo Simulation
We run 10,000 full-election simulations per update cycle. In each simulation, we add random noise drawn from a t-distribution (accounting for sampling error and correlated regional error) to every riding's projected vote shares, then translate those into seats. The seat ranges and win probabilities you see are the direct output of these simulations.
Riding Ratings
Safe (>15pp margin), Likely (8–15pp), Lean (3–8pp), and Tossup (<3pp) categories are assigned based on the projected margin between first and second place in each riding's Monte Carlo distribution. A riding is labelled "Tossup" when either major candidate wins at least 30% of simulations.
Limitations
All models are simplifications. Uniform swing underestimates local candidate effects and incumbency advantage. Our regional breakdown does not (yet) incorporate sub-provincial polling. Seat projections widen in uncertainty during periods of rapid vote-share movement. Treat all projections as probability estimates, not predictions.
Data Sources & Update Frequency
Poll data is sourced from public press releases and media coverage. The aggregate is updated within 24 hours of any new poll's release. Historical riding results are drawn from Elections Canada official returns. All methodology code is available for review on request.
About RidingWatch
Our Mission
RidingWatch exists to give Canadians — whether campaign professionals, journalists, academics, or engaged voters — the most transparent and methodologically rigorous view of the federal electoral landscape available anywhere. We believe election forecasting should be open, explainable, and honest about uncertainty.
Non-Partisan Commitment
RidingWatch is strictly non-partisan. We have no affiliation with any political party, campaign, PAC, or media outlet. Our model does not accept direction from any campaign. If you believe our methodology contains errors or bias, we encourage you to reach out — scrutiny makes models better.
How to Use This Tool
Start with the national polling averages and seat projection at the top of the page. Drill into any region using the accordion breakdown, or search for your riding directly in the table. Click any riding to see its full projection, historical vote trend, and local context. Use the Scenario Modeler to explore how a shift in national support would redistribute seats across the chamber.
A Note on Projections
No model can perfectly predict an election. Projections are estimates of current probability — not guarantees. Seat totals shown are the median outcome across 10,000 simulations; the true result will likely differ. Local factors, candidate quality, and ground-game effects are not fully captured. Use these numbers as one input among many, not as a final verdict.