Short answer: Sometimes, a little. But probably less than you think — and less than polling does.
High-probability PM odds become media headlines. The narrative that a candidate is a heavy favourite can amplify the perception — and potentially the reality — of that outcome. Mostly affects political markets.
Example: "90% odds" becomes the story, not the underlying analysis.Traders watching odds may change their own behavior — voter turnout decisions, donation choices, media coverage priorities. This effect is small and similar in magnitude to polling suppression effects.
Intentional large positions can shift displayed odds and create headlines. CFTC-regulated markets have position limits. UMA oracle on Polymarket introduces additional governance manipulation risk.
See: Can prediction markets be manipulated? →Honest caveat: Reflexivity in PMs is plausible but understudied. No published study has directly measured PM-specific reflexive effects in isolation from baseline media forecasting.
Kalshi's market data is publicly accessible via API without authentication. However, Kalshi's institutional-heavy user base and less prominent homepage probability display may reduce the voter-behavior pathway compared to traditional polling.
Honestly noted: odds are still visible on third-party trackers.
Polymarket odds are always public and surface constantly in media coverage. This creates higher exposure to the narrative-effect pathway than Kalshi.
Election prediction markets probably have some reflexive effect. The evidence is weak but plausible.
Weather, sports, and economic markets almost certainly don't cause their own outcomes.
Prediction markets are not uniquely dangerous compared to polling — the mechanisms are nearly identical.
If this concerns you, read the honest case for banning them — not just the defense. Why people want to ban PMs →
