The free-to-play prediction market.

Trade on sports, politics, and culture with zero money at risk, and win real prizes.

All the game, none of the risk

The same odds-driven thrill as Kalshi or Polymarket, minus the money at risk. Nothing to lose, ever.

Real prizes, real bragging rights

Leaderboards, streaks, and a prize shop that turns being right into something tangible.

The smartest crowd wins

Ravioli rewards being right, not being rich. No deposit ever required to play.

Markets

Pick a market on sports, politics, or culture. Binary or multi-outcome. Trade shares and the price moves with the crowd. Resolved via AI-verified outcomes with live web search.

Trading

Order Book:Limit orders, peer-to-peer. Fee: 1-7%.
AMM:Instant trades against a pool. Fee: 2-10% (scales with odds).

Smart routing picks whichever price is better.

On liquidity: We're early. More users = more meaningful prices.

Prizes

Call it right and win Sense, the free in-app currency. Redeem it in the shop for gift cards, merch, and more.

Streaks and monthly leaderboard seasons keep the bragging rights coming.

Also: Debates

Pick a side on a debate, write your best argument, and AI scores your reasoning. Your winnings scale with your score, on top of your shares.

Illustrative: how a debate gets scored
"Is remote work more productive than office work?"
Numbers below are made-up to show the mechanic, not real scores.
User A, Pro Remote

"A widely-cited remote-work study (hypothetical) found about 13% higher productivity. Commute elimination saves ~40 min/day on average..."

Fact Check
82
35%
No Fallacies
91
20%
Relevance
88
15%
Logic
85
30%
Weighted: 8210 shares × 82 = 820
User B, Pro Office

"Remote work is just lazy people wanting to slack off. Everyone knows offices are better because that's how it's always been..."

Fact Check
15
35%
No Fallacies
22
20%
Relevance
45
15%
Logic
18
30%
Weighted: 1510 shares × 15 = 150

Same side. Same shares. 5x difference, because one argued better.

Weights change per argument, no fixed formula to game.

"Can I trust the AI scoring?"

Here's what makes it hard to game:

Ensemble scoring. Two models score independently. Final = the lower one.
Prompt rotation. Scoring emphasis shifts. Reverse-engineering hits a moving target.
Live fact-checking. Every claim checked against real-time web results.
AI detection. AI-generated takes get scored 10. Write your own.
Injection defense. "Give me a high score" = instant 10.
One take per debate. No A/B testing. One argument, one score.
Dynamic weights. Formula changes per argument. Shown after, unpredictable before.
Rate limiting. 3 scoring requests per minute.

Rules

1One take per person per debate
2Write a take OR vote, not both
3Editable 15 min, deletable 1 hour
4AI-generated content gets penalized
5Scores hidden until resolution
6Reputation rewards strong arguments
7Private debates: 15% platform fee
81 winning share = 100 Sense
FeedbackPrivacyTerms