Meta tests a prediction-market app with its own employees

Meta is testing a prediction-market app with its employees, according to The Wall Street Journal. Before anyone else gets a look, the company’s own staff are the first ones placing bets on the future.

Prediction markets have become one of the more talked-about corners of consumer tech, and Meta appears to want a seat at the table without the baggage of real-money wagering.

What Meta is reportedly building

Prior reporting described the project as an app called Arena, with CEO Mark Zuckerberg leading the effort.

The core concept is familiar. Users forecast how events will turn out, and the market reflects what the crowd collectively thinks will happen.

Arena is reportedly designed around virtual currency rather than real money, a play-money model that separates it from platforms like Polymarket and Kalshi, where users put real stakes on the line.

Meta also reportedly plans to lean on its Llama AI model. Llama would generate event questions for users to forecast and serve up personalized market recommendations.

The app is reportedly aimed at users aged 18–34. According to the research findings, Meta is anticipating at least 100 million monthly active participants.

The prototype testing among employees is reportedly a step ahead of a potential public rollout on iOS and Android. Meta has not confirmed any public launch dates.

Meta has been here before

This is not Meta’s first attempt at crowd forecasting. Arena follows Forecast, an earlier app the company launched around 2020-2021.

Forecast was discontinued in 2022, reportedly because of high operational costs.

The research findings also indicate Meta is exploring partnerships with Polymarket and Kalshi to strengthen Arena’s operations. Nothing has been announced, and an exploration is not a deal.

What this means for prediction markets

The play-money design is the strategic center of the whole plan. Real-money prediction markets carry regulatory questions and a psychological barrier: many people are curious about forecasting but have no interest in risking cash on it.

The research findings frame this as a way to broaden the user base for forecasting while giving Meta valuable data on user preferences and engagement. That data could be used for future monetization.

Forecast already failed once on cost grounds, which means Meta has firsthand evidence that running a forecasting platform at scale is expensive.

The AI layer adds another variable. Having Llama generate questions could keep the app stocked with fresh markets around the clock, something human-curated platforms struggle to match.

For now, the most concrete fact is the internal test itself. The things worth watching are straightforward. Does Meta confirm a public release date for iOS and Android? Do any partnerships with Polymarket or Kalshi materialize? And can the company keep costs in check this time around?