Who Will Actually Win the LLM Race in 2026?

Inspiration

Raw prediction market odds say who’s winning, but not whether that lead is earned, stable, or fair. We wanted to understand who is actually performing well in the LLM race, not just benefiting from momentum and narrative lock-in.

What it does

Turns Kalshi prediction market data into a fair competition and scenario intelligence engine, revealing fragile leaders, undervalued challengers, and how the race shifts under realistic market shocks.

How we built it

Modeled market probabilities as time-series signals, engineered momentum and volatility features, used Random Forests to attribute what drives belief, and built a fairness-adjusted ranking plus interactive scenario simulator.

Challenges we ran into

Separating real competitive progress from self-reinforcing consensus in reflexive market data, while keeping complex statistics intuitive and interpretable.

Accomplishments that we're proud of

Showing that dominance is volatile, challengers can be consistently undervalued, and fair-adjusted rankings tell a very different story than headline odds.

What we learned

Prediction markets are efficient but not fair. Market share compounds belief, while execution efficiency reveals true competitive strength.

What's next for Who Will Actually Win the LLM Race in 2026?

Extending this into a general fair-market intelligence framework for emerging technology races where uncertainty, narrative, and execution collide.

Built With

• hex