Prévisions
Où en est l'appui des partis dans les sondages aujourd'hui, et où il se dirige. Tous les sondages nationaux, réunis en une moyenne corrigée du penchant de chaque sondeur, puis prolongés par un modèle fondé sur les fondamentaux économiques.
Prévisions nationales
Moyenne des sondages et prévision
Les lignes pleines montrent la moyenne des sondages, corrigée des effets de maison, jusqu’à la ligne d’aujourd’hui. Les lignes pointillées montrent la prévision. Les zones ombrées indiquent la fourchette à 95 %.
| Parti | Sondages aujourd’hui | Dans 3 mois | Fourchette à 95 % |
|---|---|---|---|
| LPCLiberal | 47.1% | 45.5% | 38.5–52.8% |
| CPCConservative | 30.7% | 30.9% | 27.4–34.7% |
| NDPNDP | 11.3% | 11.7% | 6.4–16.9% |
| BQBloc Québécois | 6.2% | 6.2% | 4.6–7.8% |
| GPCGreen | 2.1% | 2.4% | 1.0–4.6% |
| PPCPeople's | 1.1% | 1.3% | 1.0–2.7% |
Moyenne des sondages au 3 oct. 2026. La prévision part de la moyenne du 3 oct. 2026 et est recalculée chaque semaine.
Probabilités de victoire selon le dernier instantané des prévisions · mis à jour quotidiennement
Instantané : Aug 17, 2025
Sondages récents(2,175 au total)
| Firme | Fin de terrainⓘ | LPC | CPC | NDP | BQ | GPC | PPC |
|---|---|---|---|---|---|---|---|
| Liaison Strategies | 3 oct. 2026 | 46.0% | 28.0% | 15.0% | 6.0% | 2.0% | 2.0% |
| Nanos Research | 2 oct. 2026 | 49.1% | 28.2% | 10.9% | 6.3% | 3.4% | 1.6% |
| Liaison Strategies | 26 sept. 2026 | 46.0% | 29.0% | 15.0% | 6.0% | 2.0% | 2.0% |
| Nanos Research | 25 sept. 2026 | 49.9% | 27.9% | 11.0% | 5.6% | 3.5% | 1.4% |
| Abacus Data | 23 sept. 2026 | 46.0% | 34.0% | 9.0% | 7.0% | 2.0% | 1.0% |
| Liaison Strategies | 19 sept. 2026 | 45.0% | 31.0% | 15.0% | 6.0% | 2.0% | 1.0% |
| Nanos Research | 18 sept. 2026 | 48.1% | 31.0% | 11.2% | 4.8% | 3.3% | 1.0% |
| Liaison Strategies | 12 sept. 2026 | 44.0% | 32.0% | 14.0% | 6.0% | 2.0% | 2.0% |
| Nanos Research | 11 sept. 2026 | 48.3% | 30.9% | 10.9% | 4.9% | 3.2% | 1.4% |
| Abacus Data | 9 sept. 2026 | 47.0% | 33.0% | 8.0% | 7.0% | 3.0% | 1.0% |
| Angus Reid | 7 sept. 2026 | 46.0% | 32.0% | 12.0% | 5.0% | 2.0% | — |
| Léger | 7 sept. 2026 | 49.0% | 33.0% | 6.0% | 7.0% | 3.0% | — |
| Liaison Strategies | 5 sept. 2026 | 45.0% | 31.0% | 13.0% | 6.0% | 2.0% | 2.0% |
| Nanos Research | 4 sept. 2026 | 47.3% | 31.3% | 10.8% | 5.0% | 3.4% | 1.6% |
| Liaison Strategies | 29 août 2026 | 46.0% | 29.0% | 13.0% | 6.0% | 3.0% | 2.0% |
| Nanos Research | 28 août 2026 | 43.7% | 32.6% | 12.0% | 6.0% | 3.9% | 1.4% |
| Ipsos | 27 août 2026 | 44.0% | 33.0% | 10.0% | 8.0% | 2.0% | 2.0% |
| Abacus Data | 26 août 2026 | 45.0% | 35.0% | 9.0% | 7.0% | 2.0% | 1.0% |
| Liaison Strategies | 22 août 2026 | 43.0% | 32.0% | 13.0% | 6.0% | 3.0% | 2.0% |
| Nanos Research | 21 août 2026 | 44.1% | 30.0% | 12.9% | 6.6% | 4.0% | 2.0% |
| Liaison Strategies | 15 août 2026 | 43.0% | 32.0% | 13.0% | 6.0% | 3.0% | 2.0% |
| Nanos Research | 14 août 2026 | 45.7% | 28.5% | 13.7% | 6.2% | 4.0% | 1.5% |
| Abacus Data | 12 août 2026 | 42.0% | 35.0% | 10.0% | 8.0% | 4.0% | 2.0% |
| Liaison Strategies | 8 août 2026 | 44.0% | 31.0% | 13.0% | 6.0% | 3.0% | 2.0% |
| Nanos Research | 7 août 2026 | 45.6% | 29.1% | 12.9% | 6.8% | 4.0% | 1.3% |
| Ekos Research | 4 août 2026 | 44.7% | 27.1% | 17.0% | 4.4% | 3.5% | 2.2% |
| Léger | 3 août 2026 | 46.0% | 34.0% | 8.0% | 6.0% | 4.0% | — |
| Liaison Strategies | 1 août 2026 | 43.0% | 34.0% | 12.0% | 6.0% | 2.0% | 2.0% |
| Nanos Research | 31 juill. 2026 | 47.1% | 27.9% | 12.7% | 6.1% | 3.8% | 1.9% |
| Abacus Data | 29 juill. 2026 | 44.0% | 38.0% | 9.0% | 6.0% | 3.0% | 1.0% |
Comment fonctionnent les prévisions
A plain-English summary of the Trajectory Model and how it produces the projections above.
L’idée de base
Polls are noisy snapshots of public opinion. The Trajectory Model treats each party's true support as a hidden variable that drifts gradually over time, pulled toward a “fundamental” level set by the economy. This is a Bayesian state-space model (think of it as a Kalman filter with economic covariates and full posterior uncertainty).
Fondamentaux économiques
The model learns from 20+ years of monthly data how GDP growth, inflation (CPI), unemployment, and time in office historically relate to governing-party support. These coefficients anchor a long-run equilibrium: if the economy improves, the model expects support to drift upward toward that level over time.
Retour à la moyenne (AR(1))
Support doesn't jump to the fundamental level overnight. A mean-reversion parameter κ controls how quickly it converges, typically over several months. Parties far from their fundamental tend to drift back; those close to it stay put. This dampens extreme poll swings in the forecast.
Signal de valence
News sentiment around the governing party — measured daily from Canadian media and smoothed with a 28-day EWMA — enters the Bayesian trajectory model directly as a fitted covariate (γ_valence ≈ 0.05). Favourable coverage nudges the fundamental support level up; negative coverage nudges it down. The effect is modest by design: economic fundamentals dominate over the medium term, and the sentiment coefficient is calibrated from data, not assumed.
Prévision combinée
The trajectory model's 3-month projection is blended with the current poll aggregate. At short horizons the polls dominate; at longer horizons the economic model carries more weight. This “blended” estimate is what drives the seat projection and riding win probabilities on the map.
Intervalles de prédiction
The shaded bands on the chart are 95% ranges: for the poll average, the spread of polls in the window; for the forecast, the Bayesian posterior. The p10–p90 intervals in the accuracy tracker are further calibrated using split conformal prediction: interval widths are fit on a held-out historical period and evaluated prospectively, so coverage statistics are out-of-sample.
Implemented in PyMC · trained on Statistics Canada + Bank of Canada economic series · poll aggregate from Polling Canada
Literature: Lewis-Beck (1988), Erikson & Wlezien (2012), Daoust & Dassonneville (2018), Vovk et al. (2005)
Comment la moyenne des sondages est calculée
Fenêtre glissante. The aggregate combines polls published within a 21-day rolling window. Polls conducted closer to today receive more weight than older ones, following an exponential decay curve with a 14-day half-life. This balances responsiveness to new data against the noise inherent in any single poll.
Correction des effets de maison. Each polling firm has a characteristic tendency — a "house effect" — to over- or under-estimate each party's support relative to the final certified result. We estimate these biases from the firm's track record across past elections and subtract them before combining results, so that the aggregate reflects the field and not any single firm's methodology.
Lire l’intervalle de confiance à 95 %. The shaded band around each solid line reflects the spread of polls within the window after house effect correction. A wide interval means pollsters are producing divergent readings; a narrow interval means they are converging. It is not a margin of error for any single poll — it describes how much the aggregate could plausibly shift if the underlying polls moved to the edges of their individual ranges.
Fréquence de mise à jour. The aggregate refreshes automatically as new polls are ingested, typically within hours of publication. The date under the table above is the date the most recent calculation was run.
Effets de maison des sondeurs
Per-pollster bias corrections applied before the aggregate is computed. Positive = firm historically over-predicts this party; the aggregator subtracts this value. Estimated via James-Stein shrinkage from GE44+GE45 final-week polls vs. certified results.
| Sondeur | LPC | CPC | NDP | BQ | GPC | PPC |
|---|---|---|---|---|---|---|
| Abacus Data | -0.5 | -0.9 | +1.0 | -0.2 | +0.1 | +0.3 |
| Angus Reid | -0.4 | -0.6 | +0.3 | — | +0.1 | — |
| Ekos Research | -0.2 | -1.2 | +0.5 | -0.4 | +0.4 | +0.7 |
| Forum Research | -0.6 | -0.5 | -0.0 | -0.2 | +0.2 | +0.8 |
| Ipsos | -0.6 | -0.8 | +0.8 | -0.1 | +0.2 | -0.1 |
| Liaison Strategies | -0.2 | -0.4 | +0.2 | -0.1 | +0.0 | +0.1 |
| Léger | -0.3 | -0.5 | +0.4 | -0.1 | +0.1 | +0.3 |
| Mainstreet Research | -0.4 | -0.5 | +0.2 | -0.3 | +0.1 | +0.7 |
| Nanos Research | -0.4 | -0.9 | +0.6 | -0.2 | +0.5 | +0.3 |
Values in pp (percentage points). "—" = insufficient calibration data for this combination.