Eight points. That is the margin by which Amherst County, Virginia, rejected historical horse racing machines in 2021, and it is also the single most useful number in the whole Churchill Downs vs Sands debate currently circulating on the sell side. Not a resort valuation, not a Macau visitation figure. A county referendum that failed by eight points.
The call, in one line: county ballots beat statewide ambition
Jefferies Equity Research analyst David Katz told clients in a note Wednesday that Churchill Downs has reason to be upbeat about the midterm elections while Las Vegas Sands does not. The reasoning is almost embarrassingly concrete. Churchill Downs has two local Virginia referenda, in Amherst and Pulaski counties, that Katz expected to break its way. Sands has a Texas casino ambition that runs through Lieutenant Governor Dan Patrick, whose reelection Katz put at a 71 percent likelihood.
One company is waiting on voters in two counties. The other is waiting on a single statewide officeholder who has spent years blocking the thing it wants. That asymmetry is the entire thesis, and it is worth understanding properly, because most of the commentary around election-driven gaming stock moves gets the mechanism backwards.
The first misconception: that national elections are what move gaming stocks
They mostly don’t. What moves regional gaming earnings is municipal and county-level permission, and this cycle is a clean illustration. Churchill Downs has been absorbing competitive pressure in Virginia since the Cordish Cos. casino opened in Petersburg, and its route to defending that market runs through HHR terminals, approved county by county.
Amherst said no once. What changed, in Katz’s framing, is not the electorate’s appetite for gambling but the money: Virginia’s HB1742 increases the distribution percentage localities receive from historical horse racing wagering revenue, taken out of the amount the horse racing licensee retains. That is the qualifying term behind the headline. It does not hand the county the whole pot, as some of the commentary implies. It enlarges the county’s cut, and a referendum that failed on a smaller share is a materially different proposition when the share is bigger. Pulaski looked like the softer target, with satellite parimutuel wagering on the ballot and local officials already aligned with Churchill Downs.
Neither outcome reprices the company. Both are incremental defense of a core earnings engine in a state where a competitor has already landed. That is the honest size of the catalyst, and it is still larger than anything Sands gets from this election.
The second misconception: that a friendly incumbent winning easily is bullish
This is where probability and consequence get conflated, and New York is the case study. Katz saw Governor Kathy Hochul cruising to reelection against prediction-market supporter Bruce Blakeman at a 94 percent likelihood. Ninety-four percent sounds like a result. It buys almost nothing.
Hochul has backed downstate casino development, but she has not endorsed iGaming and has kept a restrictive posture toward prediction markets. Katz’s own words were that her favored reelection reads as status quo rather than a catalyst for online gaming expansion. A high-probability outcome with zero policy delta is not a trade. It is noise wearing the costume of information.
Compare Ohio, where the note gave pro-online-gaming Democrat Dr. Amy Acton an 83 percent chance of winning the governorship, with DraftKings and Rush Street Interactive named as likely beneficiaries. Lower probability, far higher payoff if it lands, because the policy actually changes. The lesson for anyone reading these notes: multiply the probability by the change in the rules, not by the margin of victory.
Reading the numbers against what they actually deliver
The note stacks a lot of percentages. They are not equivalent, and lining them up alongside their consequence makes the difference obvious.
| Race or measure | Reported probability | Who it touches | Policy change if it lands |
|---|---|---|---|
| Dan Patrick reelection (Texas Lt. Gov.) | 71% | Las Vegas Sands | None — blocks Texas casino hopes |
| Kathy Hochul reelection (New York) | 94% | Downstate casino and iGaming hopefuls | Status quo, per Katz |
| Amy Acton governorship (Ohio) | 83% | DraftKings, Rush Street Interactive | Opens the door to online gaming expansion |
| Alan Wilson governorship (South Carolina) | 94% | Online sports betting operators | Pro-OSB governor in a non-OSB state |
| Nebraska online sports betting measure (per Kalshi) | 68.5% | Sportsbook operators | New legal OSB market |
| Law banning sports prediction markets enacted in 2026 (per Polymarket) | ~25% | Kalshi, Polymarket, sportsbooks | Sports prediction markets banned |
Note the shape of it. The two biggest numbers on the board, both 94 percent, are the two that change the least in the near term. The Texas number is the only one where a high probability of an outcome is a high probability of nothing happening, which is precisely why it reads as negative for Sands rather than neutral.
The third misconception: that a prediction market percentage is a forecast
It is a price. A contract trading at 68.5 cents implies a 68.5 percent chance in the same way decimal odds of 1.46 do, and like any betting market it carries a built-in margin between the buy and sell side. Thin volume on a county-level or single-chamber question widens that gap further. Treating Kalshi’s 68.5 percent on Nebraska OSB or Polymarket’s roughly 25 percent on a 2026 prediction-market ban as settled probability, rather than as what traders are willing to pay with a spread attached, is a category error that keeps showing up in election-season research.
There is a neat recursion here that I think goes underappreciated. The Cruz bill, expected to be introduced after Election Day, would mandate state and/or tribal approval before event contracts on sporting events and other gambling propositions could be offered. Polymarket’s own market on a law banning sports prediction markets being enacted in 2026 sits at about 25 percent for Yes. Katz flagged that the proposal could materially alter market access for Kalshi, Polymarket and traditional sportsbook operators, with separate CFTC event-contract rulemaking capable of shaping the permissible product set, plus proposals touching sensitive-event contracts, gambling-loss deductions and identity verification.
So the instrument people are using to handicap gaming policy is itself a gaming policy question. Twenty-five percent is the market’s own price on legislation that would end sports prediction markets next year. Price it accordingly.
The downside case nobody puts in the slide deck
Election results cut both ways, and Chicago is the sharpest example in the note. Mayor Brandon Johnson was given virtually no chance of reelection, which potentially renders moot a proposed budget raising the city’s tax on online sports betting from 10.25 percent to 20 percent. That budget sits with the City Council, and Katz’s analysis indicated the increase would crimp DraftKings’ earnings while not significantly affecting Rush Street Interactive.
Roughly doubling a municipal tax on gross gaming revenue in a major metro is not a rounding error for any operator with meaningful Illinois handle. It is also the kind of item that gets buried under expansion headlines, because new markets are a better story than an old market getting more expensive. If you are only tracking where gambling becomes legal, you are reading half the ledger.
What I’d take from this, and what I wouldn’t
I would not own a casino stock because of an election. Churchill Downs vs Sands, framed as a midterm trade, is really a statement about where optionality sits: Churchill Downs has many small, independent, locally decided shots on goal, while Sands has concentrated a large one in a state where the gatekeeper is favored to stay put. Diversified optionality is structurally worth more than a single binary, and that would be true in a year with no ballots on it at all.
What the note genuinely establishes is narrower than the headline suggests. It says Churchill Downs faces a near-term calendar of events that can only help, and Sands faces one that can only disappoint. It does not say anything about Macau, Singapore, racing margins or capital allocation, which is where the actual value in either company lives.
What stays open is the federal question. If the Cruz bill or CFTC rulemaking reshapes event contracts, the prediction markets currently being used to price gaming policy may stop being usable for that purpose, and the sell side loses a data source it has grown fond of fast. Twenty-five percent says probably not next year. Twenty-five percent is not nothing.
One housekeeping point, since prediction markets keep appearing in this analysis as though they were neutral infrastructure: they are wagering products, with a margin and the same negative expected value over time as any other. Use them as a signal if you like. If you are putting money through them, set deposit and loss limits first, and treat the exposure as what it is.
