Dean Logan runs the Los Angeles County election office. His jurisdiction is the largest in the United States, serving over 5 million registered voters. In June, after the LA mayoral primary, something new happened that his office wasn't prepared for.
Early election returns differed from what prediction markets had priced. Within hours, conspiracy theories about fraud began circulating. His office spent 48 hours fielding calls from voters convinced the election was being stolen.
The fraud never happened. The prediction markets had just been wrong. But the public couldn't tell the difference — and election officials across the country now have a problem nobody prepared them for.
The trust problem nobody saw coming
Logan is not alone. The Partnership for Large Election Jurisdictions ran a webinar in August to discuss the issue. Election officials from LA, Pennsylvania, Georgia, Arizona, and others compared notes.
The pattern is consistent. Prediction markets price an outcome at 70%. The actual result goes the other way. Twitter accounts frame the mismatch as evidence of fraud. Election workers answer accusations for days.
Jim Allen, elections director of Delaware County, Pennsylvania, now bans his workers from using prediction markets entirely. He compared it to referees betting on a basketball game they are officiating. One poll worker objected during training: "What if we just want to make a minor bet on turnout?" Allen's answer: "No, this is all bad."
Why 2026 is different from 2020
In 2020, prediction markets existed but barely traded. PredictIt had $850 position caps. Polymarket was tiny. Election officials could ignore them.
In 2026, Kalshi and Polymarket are federally regulated financial exchanges with $750 million+ in trading on midterm markets alone. Media outlets cite odds as leading indicators. Campaign strategists watch them daily. They are no longer ignorable.
What changed: scale, credibility, and media coverage. CNN, Bloomberg, and Fox Business now quote Kalshi odds alongside polling averages. When those numbers diverge from reality, the question "why?" becomes public.
The three new attack vectors
1. Market-manipulated disinformation. A bad actor can place a $10 million trade on a thin market to move the implied probability dramatically. If the market now shows 85% confidence in an outcome and the actual election goes the other way, the attacker now has credible-looking "evidence" that something is wrong. Even if the trade loses the attacker money, the disinformation value may be worth it.
2. Insider trading by election officials. Election officials have access to early returns before the public. In 2024, Kalshi fined three congressional candidates for trading on their own races. The Trump teleprompter operator case ($90,000 in trades on words Trump would say) shows how far the exposure extends. Election workers with early access to ballot counts are a theoretical attack surface.
3. Narrative weaponization. Candidates can cite prediction market odds to question election integrity. "I was 70% on Kalshi — how did I lose?" is now a line we will hear in post-election disputes. The implied legitimacy of market prices gives bad-faith arguments a veneer of statistical rigor.
What Kalshi and Polymarket are doing (and not doing)
Both platforms declined to comment to Votebeat for its August 10 story.
Kalshi has publicly argued that market odds are better than polls because real money is at stake. True — but that argument assumes the odds reflect genuine information. When a whale manipulates a thin market, the odds reflect the whale, not the information.
Position limits help. Kalshi caps retail trades at various thresholds depending on the market. Polymarket has no position limits on its global exchange. The CFTC has signaled it is reviewing whether position limits are strict enough.
Neither platform has published a protocol for handling disputes between election officials and market pricing. In practice, when the two disagree, voters are left to guess who to trust.
What should actually happen
The responsible playbook, if any platform wants to implement it:
- Position limits on election markets that prevent any single trader from moving implied probabilities by more than 2-3 points
- Blackout periods for trading between poll close and official certification
- Clear disclaimers on election market pages that market prices are not predictions of outcomes
- Formal partnerships with election integrity organizations to coordinate response when prices and results diverge
- Audit trails for large positions on election contracts, available to election officials for post-event review
None of these are hard. Most would require minimal engineering. All would reduce the disinformation surface area. None has been publicly committed to by Kalshi or Polymarket as of October 2026.
What the crowd is getting wrong
Most retail traders think prediction markets are "always right" because they beat polls on headline events. They're not always right. They were wrong on Brexit at 85% Remain. They were wrong on Trump 2016 at 75% Clinton. They were slightly off on Trump 2024 but close enough that the narrative covered it.
Markets are a probabilistic tool, not a prediction oracle. When the market says 85%, there is a 15% chance it's wrong. That is not evidence of fraud when the 15% outcome happens. That is math.
Teaching voters this distinction is now part of the election integrity portfolio. It should have started three years ago.
What Juno lets you do
Juno's design assumes election markets will face regulatory scrutiny and public criticism. Our approach: strict position limits, mandatory disclaimers, transparent order books, and a resolution protocol that requires official state certification before any payout.
The alternative — Wild West election markets with whale manipulation and no safeguards — is the risk that will eventually force federal intervention on all prediction markets. Clean markets protect the entire category.
The reframe
Prediction markets won the 2024 narrative battle against polls. In 2026, they face a different test: whether they can be trusted at scale by voters who don't understand probability. Dean Logan is right to worry. The industry has to answer before November 3, or the next election cycle will look very different.