On 27 January 2020, Émile Servan-Schreiber reshared a Hypermind comparison of polling and prediction-market prices for the Paris mayoral race. His short introduction stressed that the market was not a poll. The attached question asked who would be elected mayor of Paris in 2020.
The image’s leading candidate was Anne Hidalgo at 55.9%, followed by Benjamin Griveaux at 26.5%. Rachida Dati stood at 7.8% and Cédric Villani at 6.9%. David Belliard, Gaspard Gantzer and the other-candidate category were each displayed at 1.0%. Those are the published, rounded figures; they need not sum to exactly 100%.
Two rankings, two questions
Hypermind’s accompanying text cited first-round vote intentions of 24% for Hidalgo, 18% for Dati and 16% for Griveaux, attributed to an Odoxa–Le Figaro survey. It argued that Griveaux nevertheless remained the market’s strongest challenger to Hidalgo. The image supports that market ranking.
The exact poll comparison requires a caveat. Odoxa’s official release dated 26 January reports 23%, 20% and 16% respectively and describes two January survey waves. The provenance of Hypermind’s 24–18–16 figures remains unresolved. Both versions nevertheless place Dati ahead of Griveaux on first-round voting intentions, the ranking Hypermind contrasted with its market.
Even with perfectly matched dates, the measures answer different questions. A first-round vote-intention percentage describes support for a candidate’s list at that stage. A probability of becoming mayor evaluates an eventual outcome. Eighteen percent intended support cannot be compared directly with a 7.8% winning chance as if one were a corrected version of the other.
The eventual mayor
The City of Paris records that the newly elected Conseil de Paris reconfirmed Hidalgo as mayor on 3 July 2020, following the second-round election on 28 June. She received 96 votes from the 163 councillors. This is the institutional outcome named by the market question, rather than a citywide vote-share target.
Hidalgo’s election agrees with the January market’s leading outcome. It does not prove that the ranking of every challenger was well judged, or that the market outperformed a vote-intention poll. Each measure should be judged against its stated target and the information available when it was published.
