How I got an edge betting on the Texas primary

Here's something most casual prediction market traders don't think about: primary elections and runoff elections are fundamentally different animals. Different electorates, different turnout patterns, different dynamics. And when the market prices them the same way, that's where the edge lives.

On Episode 4 of World at Odds, Andrew and I broke down a live opportunity in the Texas Republican Senate primary. I put real money behind it — over $7,500 deployed across multiple correlated bets — and the trade netted roughly +$3,000 in realized profit. This post walks through the thinking, the data, and every trade I made, so you can spot the same patterns next time.

The setup: Paxton vs. Cornyn on Polymarket

The Texas Republican Senate primary has been one of the biggest political prediction markets of 2025-2026. Two heavyweights: Ken Paxton, the Texas Attorney General riding a populist wave, and John Cornyn, the establishment three-term senator trying to reclaim his seat.

Before the March 3 primary, Polymarket had Paxton trading at roughly 82 cents — an 82% implied probability that he'd win the nomination outright. Cornyn was sitting at around 16-17 cents. The market was treating it as a done deal. I disagreed.

But here's the thing most traders missed: Texas requires a majority (over 50%) to win a primary outright. With multiple candidates in the race, the odds of anyone clearing 50% were lower than the market was pricing in. And if nobody clears 50%, the top two go to a runoff.

Why Texas runoff elections create betting edge

This is where the edge comes from. A runoff election in Texas is structurally different from a primary in three critical ways:

March 3 primary results: the market repricing

The primary results confirmed what the structural analysis suggested. With a crowded field, nobody cleared 50%. Paxton led the first round — but he didn't win. The race went to a May 26 runoff between Paxton and Cornyn.

And here's where the market moved dramatically:

The market completely repriced. Paxton went from 82% to roughly 45%. Cornyn went from roughly 17% to 55%. That's a massive swing — and I'd been building my position in Cornyn at 16-17 cents in the days before the primary. Over 11,600 shares, $2,266 deployed.

How to find edge in prediction market primaries

Let me break down exactly what the thought process looked like — not after the fact, but the way we were thinking about it before the primary:

Step 1: Identify the structural mispricing

The market was pricing Paxton at 82% to win the nomination, but it wasn't distinguishing between "win the primary outright" and "win the nomination eventually." With a crowded field, the probability of anyone getting over 50% was meaningfully lower than 82%. The market was conflating first-round momentum with final outcome.

Step 2: Model the runoff scenario

If the race goes to a runoff, what changes? Three things favor Cornyn in a runoff: lower turnout benefits organized campaigns, consolidation of moderate voters, and endorsements from eliminated candidates. Cornyn's establishment backing and decades of infrastructure become an advantage, not a liability.

Step 3: Price the conditional

The question isn't "will Paxton win?" It's "what is Paxton's probability of winning, conditional on a runoff?" The market was pricing Paxton at 82% overall, but if you broke it down: maybe 40% chance he wins outright in the primary, and maybe 40% chance he wins a runoff. That gives you roughly 40% + (60% × 40%) = 64% — materially lower than 82%.

My actual Polymarket trades: $7,500 across 3 legs

One thing we care about on this show is accountability. Anyone can explain a trade in hindsight. Here's what I actually did, with real numbers from my Polymarket account.

Leg 1: Cornyn to win the primary (Yes shares)

In the days before the March 3 primary, I accumulated over 11,600 Cornyn Yes shares at an average entry of about 16-17 cents. Total cost: $2,266. This was the core thesis — Cornyn was underpriced because the market was conflating "Paxton leads the primary" with "Paxton wins the nomination."

Once the runoff was confirmed, I didn't just hold and hope. I sold in tranches as the price moved up, locking in gains at each level:

The key decision was not selling everything at the first pop. When Cornyn hit 32 cents I took 2,700 shares off — that locked in a ~2x on a chunk and reduced risk. But I held 9,000+ shares through because the thesis hadn't changed: Cornyn was still underpriced for a runoff. As the market caught up, I kept scaling out. The last big tranche sold at 90 cents — a 5.6x on shares bought at 16 cents.

Leg 2: No outright winner (the runoff hedge)

This was the second expression of the same thesis. Polymarket had a separate market on whether any candidate would win the GOP primary outright (without a runoff). I bought No shares — essentially betting that the race would go to a runoff.

I accumulated about 3,660 No shares at an average of ~80 cents. Total cost: $2,948. When no candidate cleared 50% and the runoff was confirmed, these shares resolved to $1.00. I sold most at 98-99 cents, netting about +$804.

Leg 3: Jasmine Crockett — Dem nominee (the loss)

I also took a position on Jasmine Crockett to win the Democratic Senate nomination. I bought Yes shares at prices ranging from 32 cents down to 1 cent, deploying about $2,171 total. This one didn't work. The position was marked as a loss, and I recovered almost nothing — about $23 on the exits. Net loss: -$2,148.

I'm including it here because I think it's important. We track every bet on our scorecard — wins and losses. If you only show the winners, you're not being honest about the process. The Crockett trade was a separate thesis from the Cornyn runoff thesis, and it was wrong.

The full picture

Roughly +$2,900 on about $7,500 deployed — a 39% return on capital. Not bad for a two-week trade. But notice: without the Crockett loss, the Cornyn and runoff trades alone would have been +$5,222. That's the nature of the game. You size your winners bigger than your losers, you take +EV bets, and you let the math do its thing over time.

Structural edge in prediction markets: where to look next

This trade wasn't about having better political information than everyone else. I wasn't talking to insiders or running polls. The edge came from understanding the structure of the event — how Texas runoffs work, how turnout patterns shift, how consolidation changes the math.

Prediction markets are great at pricing simple binary outcomes. They're less great at pricing multi-stage events with conditional probabilities. Anytime you see a market that conflates "leading the first stage" with "winning the whole thing," there's potential edge.

Here are some other places the same pattern shows up:

Texas runoff outlook: what happens May 26

The Texas runoff is scheduled for May 26, 2026. As of this writing, Cornyn is trading at around 55-61% across platforms (Polymarket has him higher, Kalshi has the race tighter). I've already taken most of my profit off the table — sold the majority of my position between 43 and 90 cents — but I'm still watching the race closely and hold a smaller residual position.

Whether Cornyn wins or loses the runoff, the process was right: find structural mispricings, model the conditional probabilities, size up on +EV, and sell in tranches as the market catches up. Sometimes you'll lose (hi, Crockett). But if you do that consistently, you win in the long run.

Which, if you've been paying attention, is kind of the whole point of this show.