Tail riskM1
US military action against Canada
A tail risk. A low-probability risk tracked with warning indicators, not a forecast. A separate yes/no question: it is not part of any sum. All tail risks.
US forces carry out a kinetic or occupying military action against Canadian territory, forces or infrastructure by 2027-12-31. Excludes cyber operations, border enforcement, and routine joint exercises.
- Current odds
- 0.05% (range 0.05%–2%), as of . An estimate, not a measurement.
Articles that cite this record
How the odds have moved
| As of | Odds | Reason recorded |
|---|---|---|
| 6 October 2026 | 0.05% 0.05%–2% | From our model: we looked at 22 cases since 1983 where the US used force against another country, and 14 tense cases where it did not, and noted which of six groups of warning signs came first: a military build-up, war logistics or medical preparation, evacuations or diplomatic withdrawals, official threats or justifications for force, legal or coalition steps toward force, and a cut in defence or intelligence cooperation. None of the six is present for Canada today. With no signs, the model falls back on a floor: in 1,241 years of US relations with its treaty allies, the US never once attacked one. Counting that as less than one attack in 1,241 years gives about 0.05% by the end of 2027. The range is 0.05% to 2.1%, depending on how two of today's facts are read: if the Pentagon's pause of the Permanent Joint Board on Defence counts as a cut in cooperation, about 1%; if the tariff emergency counts as a legal step toward force, about 2.1%. The odds would rise quickly if several signs appeared. The weak points: the floor rests on zero events, so it is a lower bound, not an observed rate; and reading today's ambiguous signs differently lifts it to about 2%. |
| 6 October 2026 | 0.05% 0.05%–2% | From our model: we looked at 22 cases since 1983 where the US used force against another country, and 14 tense cases where it did not, and noted which of six groups of warning signs came first: a military build-up, war logistics or medical preparation, evacuations or diplomatic withdrawals, official threats or justifications for force, legal or coalition steps toward force, and a cut in defence or intelligence cooperation. None of the six is present for Canada today. With no signs, the model falls back on a floor: in 1,241 years of US relations with its treaty allies, the US never once attacked one. Counting that as less than one attack in 1,241 years gives about 0.05% by the end of 2027. The range is 0.05% to 2.1%, depending on how two of today's facts are read: if the Pentagon's pause of the Permanent Joint Board on Defence counts as a cut in cooperation, about 1%; if the tariff emergency counts as a legal step toward force, about 2.1%. The odds would rise quickly if several signs appeared. The weak points: the floor rests on zero events, so it is a lower bound, not an observed rate; and reading today's ambiguous signs differently lifts it to about 2%. |
| 2 October 2026 | 0.05% 0.05%–2% | From our model: we looked at 22 cases since 1983 where the US used force against another country, and 14 tense cases where it did not, and noted which of six groups of warning signs came first: a military build-up, war logistics or medical preparation, evacuations or diplomatic withdrawals, official threats or justifications for force, legal or coalition steps toward force, and a cut in defence or intelligence cooperation. None of the six is present for Canada today. With no signs, the model falls back on a floor: in 1,241 years of US relations with its treaty allies, the US never once attacked one. Counting that as less than one attack in 1,241 years gives about 0.05% by the end of 2027. The range is 0.05% to 2.1%, depending on how two of today's facts are read: if the Pentagon's pause of the Permanent Joint Board on Defence counts as a cut in cooperation, about 1%; if the tariff emergency counts as a legal step toward force, about 2.1%. The odds would rise quickly if several signs appeared. The weak points: the floor rests on zero events, so it is a lower bound, not an observed rate; and reading today's ambiguous signs differently lifts it to about 2%. |
Other readings on this scale: M2. All of them on the odds page.