Florida Governor: Forecasters and Markets Disagree

Three sources of information on the Florida governor's race are telling three different stories, and the gap between them is wide enough to raise questions about what each is actually measuring. Decision Desk rates the contest between Republican Byron Donalds and Democrat David Jolly a toss-up, a label that typically implies a race within a few points and genuine uncertainty about the outcome. Kalshi traders see it differently: contracts on the Donalds line price his win probability at 77%, a figure that would normally accompany a mid-single-digit polling lead, not a coin flip. A Stetson University poll completed September 21 splits the difference in a way that favors the market read, showing Donalds ahead of Jolly 51% to 39% among 830 likely voters — a 12-point spread that looks nothing like a toss-up.

The divergence is worth sitting with rather than waving away. Analytical ratings like Decision Desk's are built to be conservative, weighting historical volatility, candidate quality, and structural factors that a single poll cannot capture; they are slower to move and can lag behind a race that has genuinely broken toward one side. Prediction markets, by contrast, aggregate real-time sentiment from traders willing to put money on an outcome, which makes them responsive but also vulnerable to overcorrection, thin liquidity, and the herd behavior that comes from everyone watching the same polling. The Stetson survey sits in between as a single data point — informative, but with the usual caveats about house effects, likely-voter modeling, and the fact that one poll is not a trend.

What makes this case notable is the size of the gap rather than its mere existence. A modest disagreement between a rating and a market is unremarkable; a 12-point poll margin and a 77% market probability sitting underneath a toss-up label is not. One explanation is that Decision Desk's rating predates the Stetson poll or has not yet incorporated it, in which case the label is simply stale. Another is that the Stetson sample or methodology is an outlier that the market has overweighted, in which case both the market and the poll may be overstating Donalds' position. Without additional polling from other firms, there is no way to adjudicate between those explanations, and readers should treat the current numbers as unsettled rather than converging.

The practical stakes of resolving this are straightforward. A toss-up rating tells campaigns, donors, and national committees to keep resources flowing into the race; a market pricing in a 77% favorite tells the same audiences the contest may already be decided, which can shape fundraising, ad spending, and turnout operations well before a single additional vote is cast. If the rating and the market remain this far apart heading into the fall, Florida becomes a useful test case for which signal — analyst judgment, crowd-priced betting, or traditional polling — proves more reliable once more data arrives.

Watch tomorrow for reaction to the Missouri ruling from state election officials and whether additional pollsters weigh in on Michigan and Georgia to help resolve this week's divergent numbers.