Analysis of the results

1. Seats: the median was right for the PQ

Our headline figure was the number of ridings where each party was ahead: 65 for the PQ, above the majority line (64). That figure gives the leading party all of its close races. The median of our 5,000 simulations, which accounts for the odds in each riding, gave the PQ 59 seats, and a minority government in 66% of cases. The PQ won 59.

• 90% of simulations
• middle half
• simulation median
• leading (published figure)
• seats won

Total seat gap (sum of the gaps by party): 12 with the leading count, 12 with the median scenario (PQ 59 · PLQ 38 · PCQ 17 · QS 7 · CAQ 6), 16 with rounded means. Both simulation-based methods give the CAQ a few seats (6 and 8); it won none, which happened in only 3.9% of our simulations.

2. The vote: CAQ overestimated, QS underestimated

Our poll average was 1.4 points from the result on average. The two biggest gaps: the CAQ (−2.8 points) and Québec solidaire (+2.1 points). The PQ and the PLQ are within one point.

• final projection
• result

3. The polls: Léger and Pallas closest

Each firm’s last poll in our projection, compared with the result (average gap over the five main parties, in points). Léger was the closest (0.90 points). The polls that put the CAQ near 20% pulled our average up for that party.

Shares of the five main parties after removing undecided voters, as used in our average. Result: valid vote, Élections Québec.

4. Riding by riding: favourites won more often than expected

Added up over the 127 ridings, our probabilities implied about 19 misses. There were 8. Favourites at 90% or more all won, and even in races under 60%, the favourite won 13 times out of 14. In this election, favourites won more often than our probabilities implied. One election is not enough to conclude that the model was too cautious, but it is a reason to re-examine its uncertainty.

• favourites who won
• rate expected from our probabilities

Average gap per riding, five main parties: 3.4 points.

5. Where the model missed

Seven of the eight missed ridings were projected for the PQ. Three patterns explain them:

• Québec solidaire in its strongholds: its incumbents did 12.7 points better than our projection on average (7 ridings), and the gap was just as large elsewhere the party is well established, such as Rosemont (+11.1) and Gouin (+21.2). Our model underestimated this local strength: Taschereau and Rosemont stayed with QS.
• The Quebec City region: the PQ did 3.9 points worse than projected, to the benefit of the PCQ and QS. Three of the region’s eleven ridings were missed (Jean-Talon, Louis-Hébert, Taschereau).
• The CAQ’s collapse outside the big cities: −3.9 points on average. Its votes went in every direction, mostly helping the PLQ and the PCQ.

Average gap by region (actual minus projected vote, in points)

The 8 missed ridings

Regions: Statistics Canada census metropolitan areas, based on where most of each riding’s voters live.

6. For the next election

The model will be reviewed before the next election. Three areas:

• A better-suited headline figure: also show the most likely number of seats from our simulations, alongside the ridings leading.
• Better account for parties’ local strength, such as Québec solidaire’s in its strongholds.
• Re-examine our uncertainty margins in light of this election.

Sources: Élections Québec results (open data, October 6, 2026, 2:23 a.m., 100% of polling stations); 127QC final projection of October 4, 2026 (unchanged); pollsters’ reports. See the scorecard · Method