Below you’ll find the seat-by-seat projections produced by the model that is powering the CBC’s Poll Tracker.
Seats are classified by the following categories:
Safe: The party has a very high likelihood of winning the seat. Only an enormous error in the polls, the projection model or a very powerful local dynamic would result in another party winning the riding.
Likely: The party has a high likelihood of winning the seat, though there is an outside chance (less than 5%) that another party could win.
Leaning: The party is projected to be leaning and has a good chance of winning the seat, with odds ranging from roughly 68% to 95%.
Toss: Two or more parties are in contention for the seat, with each contending party considered to have a chance of winning of less than 68%.
Possible: The party is not favoured to win the seat, but has an outside chance of winning. This is the flipside of the “Leaning” rating.
Unlikely: The party is not in serious contention for the seat, but there is an outside chance that, with enough polling and modelling error and/or local dynamics at play, the party could pull off an upset. This is the flipside of the “Likely” rating.
No Rating (-): The party is not in contention to win the seat. This is the flipside of the “Safe” rating.
A full explanation of how the seat projection model works can be found at the bottom of the Poll Tracker. In brief, these projections are based on a regional proportional swing model, adjusting the results from the last election in each riding according to how the polls have moved since then. Other factors, such as incumbency, star candidates and leaders on the ballot are taken into account. The classification of each seat is then based on the likelihood of the model making an error based on its own past errors and the errors of the polls in previous campaigns.
Bookmark this page, as the projections will be updated here whenever the Poll Tracker is updated.