⭐ Validated cases
The method against the real result: already-held elections, with cross-referenced market × poll divergence and an open dataset.
The most honest case here: the two markets disagreed. The winner market (electoral college, ~US$3.7bn, the largest election market in history) gave Trump ~56% against a poll near-tie, and was right; the popular-vote market gave Harris ~74% and was wrong (Trump won the popular vote too). AFOS shows the market’s hit and its miss side by side.
The market's months-long favorite (López Aliaga) placed 3rd and missed the runoff: sustained divergence, not noise. In the June 7 runoff the market gave Fujimori ~68% while polls saw a technical tie, and she won by ~0.27pp: the JNE proclaimed Keiko Fujimori (50.135% × 49.865%) on July 3. Right on direction, overstated the margin. Total market volume: ~US$107M.
Jara led the first-round vote, but the market priced Kast at ~66% to win, and Kast won the runoff 58×42. The gap was the signal. Total market volume: ~US$49M.
De la Espriella won the first round (May 31, 43.7%), with the market already pricing him favorite (43.5%): near-zero divergence. In the June 21 runoff he won by ~0.96pp (49.66% × 48.70%), though the market gave near-certain victory (88.5%): right winner, overstated margin. Total market volume: ~US$37M.
The AfD was 2nd in votes (~21%) but the market gave it only ~3% to win the most seats. Vote share is not winning. CDU/CSU won, as the market (~97%) called. Total market volume: ~US$106M.
The market swung ~85% Conservative (Jan) → ~80% Liberal (Apr) with the vote near-tied, and the Liberals won (169×144). The swing was the signal. Total market volume: ~US$12M.
Starmer's Labour won 411 of 650 seats. The market gave it about 99% to win the most seats while polls measured about 40% of the vote: first-past-the-post turned 33.7% of votes into 63% of seats. Reform was third in votes yet took only 5 seats. Total market volume: ~US$1.76M.
Sheinbaum won the 2024 presidency with about 59.8%, the largest vote count in Mexican history. From January the market already gave her about 90% to win, while polls measured her vote share around 50%. The market called the outcome early and the result outran the polls. Total market volume: ~US$2.08M.
Lee Jae-myung won the 2025 snap election, called after the martial-law crisis. The market gave Lee about 95% to win and even nailed the margin (the 8-to-11pp band; the actual margin was 8.27pp), while polls measured his vote share around 49%. One of the largest election markets ever outside the US (~US$290M).
By SINGLE party, Le Pen’s RN was the largest group (143 seats), and the deepest market (~US$917k) priced that at ~99% and was right. What flipped was the GOVERNMENT: the left-wing NFP coalition held the most seats by coalition (182) via the front républicain. The near-RN-majority hype (230-270 seats) lived in polls and thin markets and did not survive at volume.
The largest election in history (~980M eligible voters). The prediction market (~US$835k) and the polls agreed with each other and overestimated together: right on the winner (Modi’s NDA), very wrong on the size, projecting 350 to 400 of 543 seats against ~293. The BJP lost the single-party majority it governed with (240). The vote was a near-tie (43.8% × 41.48%): the opposition unified, and Modi’s stable vote became a seat defeat.
📊 Live: market odds now
Ongoing elections with active Polymarket markets. Odds only, without the polls and divergence layer of the cases above.
Loading map data...
Global Election Calendar
Elections with Polymarket Data
Data: Polymarket | Live volumes and odds
The Global module of AFOS Analytics transforms the world election calendar into real-time intelligence.
The interactive map allows you to visualize, country by country:
- • ongoing elections
- • upcoming elections
- • completed cycles
Creating a global thermometer of political risk and economic opportunity.
Integrated Data Layer
By interacting with the map, you access consolidated data combining:
- • prediction markets (real-money bets)
- • official election calendar
- • political trend analysis
How it works
Each country on the map represents an active or upcoming election event.
Colors indicate the time horizon of elections:
- • light tones (interactive) → short-term elections (up to 12 months)
- • medium tones (interactive) → medium-term elections (1 to 3 years)
- • dark tones (non-interactive) → scheduled elections
- • date not yet defined
Interaction indicates availability of detailed data.
Clicking on a country
The user accesses organized information such as:
- • probability of election scenarios
- • level of market interest
- • related events and elections
Transforming the map into an interactive geopolitical decision interface, not just visualization.
Dynamic Global Calendar
The system tracks elections continuously, on a global scale.
Every year, dozens of countries hold national elections, generating:
- • currency impact
- • market movements
- • geopolitical risk repricing
The result is a continuous flow of events with real economic impact.
Strategic Purpose
The Global module is an infrastructure for systemic reading of the global political environment.
It allows you to:
- • anticipate market movements
- • identify political risk by region
- • monitor election cycles in real time
- • make decisions based on data
- • not narrative
Executive Translation
Politics generates volatility.
Volatility generates asset repricing.
Global organizes this complexity into an interface that allows you to: