Anthropic at 94%? The Market Just Called the Race a Month Early

It's the last Sunday of August, GPT-6 rumors are pinging every group chat, Google just teased something Tuesday, and Polymarket is out here telling us the September AI king is already crowned. Sure. Let's fade some markets.

Will Anthropic have the best AI model at the end of September 2026?

Position: NO at 6¢.

The market is pricing this at 94% NO — basically saying Anthropic has no chance of holding the crown a month from now. Fine, Claude isn't the consensus #1 today. But 6¢ implies near-certainty on a subjective benchmark, one month out, with OpenAI, Google, xAI, and Anthropic all shipping constantly.

Our model has true NO closer to 80%, meaning YES should be trading around 20¢, not 6. One Anthropic drop, one benchmark shuffle, one vibes-based Twitter consensus shift and this thing moves.

You're basically getting a free lotto ticket on a frontier lab launch. Edge score 14/100 — the AI likes it.

What kills it: "best model" is a popularity contest, and if nobody ships anything spicy in September, the incumbent narrative just holds and your 6¢ expires worthless. But at this price, you don't need to be right often.

Jiri Lehecka vs. Pablo Carreno Busta — Total Sets O/U 3.5

Position: NO (under 3.5 sets) at 45¢.

US Open first round. Lehecka is a legit top-25 guy playing his best tennis. PCB is a 35-year-old returning from a long injury layoff, ranked deep in the triple digits. And the market has this as a coin flip on whether it goes 4+ sets.

We've got true NO around 58%. Best-of-5 with a clear favorite versus a rusty vet? That's a straight-sets script more often than not. Carreno Busta can grind, but grinding for three tight sets against a hard-hitting 23-year-old on a hot New York court is a different sport than what his body wants to do right now.

Getting 45¢ on the more likely outcome is the definition of a soft line.

What kills it: PCB is a former semifinalist here and knows how to steal a set on serve. One tiebreak swing and this ticket is dead. Best-of-5 variance is real.

Sweden Democrats — 3rd Most Seats in 2026 Election

Position: NO at 67¢.

Market says 33% chance SD lands in 3rd. We say closer to 45%. Why the gap? SD has been the 2nd or 3rd biggest party in every recent Swedish election. Current polling has them in a dogfight for 2nd with the Moderates, which means 3rd is very much on the table — not some tail outcome.

The market is treating "3rd place" like it's a weird specific bucket. It's not. It's one of the two most likely finishes for this party.

Edge score 12/100, Grade A. Nothing sexy, just a mispriced bucket in a multi-outcome market.

What kills it: if SD's polling holds or ticks up, they lock in 2nd and this is dust. It's a bet on a specific ordering, not on the party doing well.

---

Three fades, one slate. GL out there — and if Anthropic drops a model Monday, don't say we didn't warn you.