This Week's Contract
Will Thomas Massie be the Republican nominee for KY-04? (Polymarket, 70c) is a congressional primary contract resolving May 19, 2026. At $288,592 Vol with roughly four weeks to resolution, it sits squarely in the 1w-1mo horizon window. Massie, a libertarian-leaning incumbent who survived a Trump-era primary challenge in 2022, is the heavy favorite in the district he has held since 2012.
The Le calibration for this contract applies a slope of b = 1.11 in the (politics, 1w-1mo) cell — the weakest underconfidence reading in the political domain. The recalibrated probability lands at 71%, just 2pp above the market's 70c price. That's an unusually tight gap for a political contract, and the reason tells us something important about primary markets versus general elections. But to understand why 2pp is small, you first need to understand why political markets typically produce gaps of 10–15pp.
Political Market Psychology: Why Politics Is the Most Mispriced Domain
Le (2026) analyzed 292 million trades across 327,000 binary contracts and found that politics is the single most underconfident domain at nearly every time horizon. On Kalshi, the political 1w-1mo slope is 1.83 — meaning prices are compressed dramatically toward 50% relative to their true probabilities. A 70c political contract one week out doesn't represent 70% confidence. After logit-space recalibration, it corresponds to roughly 83% true probability. The politics domain intercept in Le's decomposition is +0.15, the highest of any category studied, and it persists across both short and long horizons (1mo+ slope: 1.73).
Why does politics produce such persistent underconfidence? Le (2026) decomposes calibration variance into four components, but doesn't identify psychological mechanisms — for that, we turn to Snowberg & Wolfers (2010). Their analysis establishes that the favorite-longshot bias in prediction markets is driven by probability misperceptions, not risk-love. Retail participants systematically overweight small probabilities (their preferred candidate's comeback) and underweight large ones (the frontrunner's actual dominance). This is a cognitive bias rooted in prospect theory-style probability weighting, not rational behavior. It will persist as long as retail participants constitute a meaningful share of political market volume — which, given the emotional salience of electoral outcomes, is structural rather than cyclical.
The second mechanism is what researchers call the bilateral cancellation hypothesis. In political markets, partisans on opposing sides trade aggressively out of conviction rather than probability assessment. A Democrat buys "No" on a Republican nominee out of motivated reasoning; a Republican buys "Yes" for the same reason. Both sides push prices toward 50% from their respective directions. The result is a systematic compression of prices toward the midpoint even when the underlying probability distribution is highly skewed. This dynamic is most pronounced in high-salience national races, which explains why Le's slopes are steepest for general elections and party nomination contracts with broad partisan interest.
A third layer comes from Reichenbach & Walther (2025), who document a systematic "Yes" bias on Polymarket specifically. Traders overtrade the "Yes" option across contract types, creating a persistent edge for "No" positions. In a primary contract framed as "Will X be the nominee?", this means the incumbent favorite's "Yes" price is likely inflated not just by calibration bias, but by platform-level directional skew.
The 2024 presidential election is the cleanest illustration of all three effects operating simultaneously. When Polymarket had Trump at 62c in the final weeks before November 2024, that was not a 62% probability statement. Applying Le's political slope to that price produces a calibrated probability closer to 75%. The market's compression toward 50% — driven by Democratic-leaning traders buying Harris "Yes," by motivated Republican optimism, and by Polymarket's "Yes" bias — masked a substantially more skewed true probability.
Back to the Massie contract: the 2pp calibration edge is small precisely because KY-04 is a low-salience primary with minimal partisan cross-fire. There is no Democratic base placing motivated "No" bets, no national media narrative compressing the price. The bilateral cancellation effect is nearly absent. The Le slope of 1.11 here is closer to a mildly inefficient market than the 1.83 seen in high-stakes national political contracts. For traders, the takeaway is not that this contract is mispriced — it largely isn't. The takeaway is that the same framework that finds a 2pp edge here finds a 13pp edge on a Senate race with comparable surface-level odds, because the psychology of political trading scales with partisan salience, not just with probability.
Methodology
Calibration analysis applies Le (2026) logistic recalibration to contracts with >$50K volume and >7 days to resolution. The Le slope transforms market prices in logit space: p* = p^b / (p^b + (1-p)^b), where b is the slope for a given (platform, category, horizon) cell, estimated from 292 million historical trades. Slopes >1 indicate underconfidence (prices compressed toward 50%); slopes <1 indicate overconfidence.