Resolved global election case study

Polymarket Peru Presidential Election: 2026 Market Postmortem

Keiko Fujimori won Peru’s 2026 presidential election after a very close runoff, and the Polymarket winner event is resolved. Peru’s National Jury of Elections, the JNE, proclaimed Fujimori president-elect on July 3 after reporting 9,223,396 votes for her ticket and 9,173,755 for Roberto Sánchez’s ticket.

The narrow 49,641-vote difference makes this market a valuable lesson: a winner contract can move to near certainty while certification, challenges and margin contracts still require precise official data.

Official result and timeline

ItemVerified fact
General election first roundApril 12, 2026
Presidential runoffJune 7, 2026
Proclaimed winnerKeiko Fujimori
Fujimori ticket votes9,223,396
Roberto Sánchez ticket votes9,173,755
Vote difference49,641
JNE proclamationJuly 3, 2026
Polymarket eventResolved

Polymarket’s event page can show the winner at 100% now. That is the historical end state, not the probability available before the vote or during the count.

Why this election required three analytical phases

First round: a large candidate field

Peru’s first round included many candidates. The market had to price both who would advance and who could ultimately win. Candidate fragmentation makes a simple national polling rank insufficient.

Runoff: a new two-person choice

Once Fujimori and Sánchez advanced, endorsements, anti-candidate voting and regional transfer patterns became central. First-round vote shares did not transfer mechanically.

With the runoff extremely close, remaining ballots, challenged records and official adjudication mattered. Market trading during this phase was nowcasting the count and institutional process rather than forecasting voter behavior.

Combining all three phases in one undifferentiated chart hides the changing information problem.

Calculate the margin from official votes

The difference was:

9,223,396 - 9,173,755 = 49,641 votes

The two-ticket valid-vote total was:

9,223,396 + 9,173,755 = 18,397,151 votes

That implies an approximate winner share of 50.135% and a gap near 0.27 percentage points. Use the JNE’s official figures and the exact denominator specified by a margin market. Rounded media figures can move a result between very narrow 0.1-point brackets.

Polymarket also listed runoff margin events, including fine 0.1% ranges. Such contracts require far more numerical precision than the winner market.

Winner certainty is not settlement certainty

Near the end of a close count, three probabilities may differ:

  • probability the current leader has the most votes after counting;
  • probability official challenges alter the totals enough to change the leader;
  • probability the market resolves under its exact written rule by a particular time.

A trader could be confident Fujimori would be proclaimed and still face delay, spread and rule interpretation. At 99¢, maximum gross upside is only 1¢ per share, while a reversal loses almost the full purchase price.

This asymmetry is why “almost certain” is not the same as “free money.”

ONPE and JNE have different roles

Peru’s electoral system includes institutions with distinct responsibilities. The Oficina Nacional de Procesos Electorales, ONPE, organizes the voting process and publishes count information. The Jurado Nacional de Elecciones, JNE, handles electoral justice functions and proclaims results.

For research:

  • use ONPE for process and count data;
  • use JNE for formal proclamations and relevant adjudication;
  • follow the market’s designated resolution source and evidence hierarchy;
  • do not treat a campaign statement as official action.

An article should name the institution instead of vaguely saying “Peru confirmed the result.”

How to audit the historical Polymarket odds

Start with the event and token identifiers, not a screenshot. Pull historical price data and mark institutional milestones.

Recommended checkpoints:

  1. one month before the first round;
  2. final pre-first-round trading hour;
  3. when the runoff candidates became clear;
  4. one week before the runoff;
  5. final pre-runoff trading hour;
  6. count milestones at fixed reporting percentages;
  7. major challenge or adjudication announcements;
  8. JNE proclamation.

For every checkpoint, store bid, ask, midpoint, last trade, depth and UTC time. A last trade can be stale. A midpoint can be non-executable. An order book can be thin.

Separate forecast evaluation before voting from count prediction after voting. They test different skills.

Multi-candidate market math

Before the runoff, candidate winner tokens represented mutually exclusive ultimate outcomes. Their displayed percentages might not total exactly 100 because quotes can be asynchronous and spreads differ.

To estimate an executable outcome distribution:

  1. collect simultaneous best bids and asks;
  2. exclude stale or disabled outcomes only according to documented rules;
  3. report the sum of bids and sum of asks;
  4. avoid renormalizing without disclosing the method;
  5. inspect depth, not just top-of-book prices.

Renormalizing midpoints to 100% can make a neat chart but hide trading cost. Publish both raw and normalized values if normalization is necessary.

The close-count evidence ladder

During a narrow election, information quality becomes decisive.

Strong evidence

  • official count releases;
  • JNE decisions and proclamations;
  • ONPE process updates;
  • published legal filings and rulings.

Useful but provisional

  • reputable decision-desk analysis;
  • documented geographic breakdowns;
  • statements by named election officials.

Weak evidence

  • candidate victory declarations;
  • unsourced fraud allegations;
  • cropped tally screenshots;
  • anonymous social posts;
  • market comments asserting inside information.

A claim of irregularity is not proof that enough votes will change to reverse a result. Quantify the number of ballots affected and the legal remedy sought.

What the market page’s volume does not prove

The event recorded very large cumulative volume. High volume can improve price discovery, but it does not by itself establish:

  • accurate prices at every date;
  • independent participants;
  • deep liquidity at the displayed quote;
  • absence of manipulation;
  • correct interpretation of legal challenges;
  • guaranteed settlement speed.

Measure spread, depth and response to verified information. Volume alone is an activity statistic.

Geographic analysis without overclaiming

Peru’s regional voting patterns can help explain the count sequence, especially when late-reporting areas favor different candidates. But analysts must use actual outstanding ballots, not stereotypes.

A proper remaining-vote model records:

  • ballots counted and pending by jurisdiction;
  • each candidate’s observed share in comparable counted ballots;
  • overseas votes and their reporting schedule;
  • challenged or observed tally sheets;
  • uncertainty intervals around the projection.

Do not assume all uncounted ballots resemble the most favorable reported district. That produces overconfident market interpretations.

Common errors

Calling a resolved page “live election odds”

The winner is known. Historical odds require timestamped retrieval.

Using the April date as the final winner date

April 12 was the first round. The decisive runoff occurred June 7.

Treating ONPE count and JNE proclamation as the same event

They are related but institutionally different milestones.

Rounding a 0.27-point margin too aggressively

Fine-bracket contracts require raw official totals.

Believing fraud claims without quantification

Evaluate evidence, affected ballots and authoritative rulings.

A publication template for narrow elections

Every update should show:

Official data timestamp:
Counted share or reporting status:
Candidate raw votes:
Current difference:
Outstanding/challenged units:
Institution responsible:
Market bid/ask timestamp:
Market rule and resolution source:
Status: forecast / nowcast / proclaimed / resolved

This prevents a page from silently moving from prediction to result while keeping the same headline.

Frequently asked questions

Who won Peru’s 2026 presidential election?

Keiko Fujimori was proclaimed president-elect by the JNE after the runoff against Roberto Sánchez.

How close was the result?

The JNE figures cited here differ by 49,641 votes, roughly 0.27 percentage points of the two-ticket valid-vote total.

Why does Polymarket show Fujimori at 100%?

The winner market is resolved. The 100% display is not a current forecast.

What is the difference between ONPE and JNE?

ONPE administers the election process and count operations; JNE performs electoral-justice functions and formally proclaims results.

Can I evaluate the market using its final price?

No. Use prices recorded before voting or before count information was available, depending on the forecast question.

Sources and verification

Risk notice: Narrow election markets can move violently on partial counts and legal claims. Use primary records, quantify uncertainty and never treat a near-$1 contract as risk-free.

Update log

  • November 17, 2026: Added JNE proclamation totals, margin calculations and separate first-round, runoff and count phases.
Polymarket Central Global Elections Desk

Reviewed by Peru JNE and ONPE Results Review. Updated November 17, 2026. Risk level: Medium.