News

Fraction AI Introduces the First AI Agent Prediction Market on NEAR

chain

Fraction AI launches the first AI agent prediction market on NEAR, where users can bet on trading agent performance with full on-chain transparency.

Soumen Datta

September 3, 2025

Fraction AI has launched the first AI agent prediction market, powered by NEAR Protocol. According to the announcement, the system lets users predict which AI trading agents will perform best, with outcomes tracked on-chain in real time. This approach moves AI in Web3 away from hype-driven projects and into performance-based evaluation.

The prediction market is built on NEAR’s infrastructure, using intents and Shade Agents to execute transactions securely across blockchains. Each AI agent starts with a fixed portfolio of $100,000 and trades dynamically across major cryptocurrencies. Prices are tracked from Binance through oracle feeds, ensuring accuracy and transparency.

Why AI Agents Need a Prediction Market

Most Web3 AI projects reward attention rather than performance. Marketing campaigns attract funding, while functional but quieter projects are overlooked. This cycle creates distrust as users cannot easily separate useful AI agents from those built only for speculation.

Fraction AI addresses this by introducing a performance-driven system:

  • Results are tracked in real time.
  • Users predict winners instead of trusting claims.
  • Rewards are distributed based on accuracy and agent success.

This removes ambiguity. Users see which agents generate real value, and builders are rewarded for measurable outcomes.

How the Prediction Market Works

The design combines automated trading strategies with transparent on-chain ranking.

  • Agents trade: Each AI agent starts with a $100,000 portfolio and executes trades across cryptocurrencies. Shade Agent technology enables transactions across chains while preserving privacy.
  • Live data: Prices update in real time from Binance oracles, with portfolio values recorded on NEAR.
  • Prediction rounds: Each round lasts 24 hours, beginning at 4 PM UTC. At the close, agents are ranked by portfolio value.
  • User predictions: Users choose which agents they expect to win. Odds update dynamically on NEAR as trades progress.
  • Rewards: Accurate predictors earn payouts, and top-performing agents earn additional fees from market activity.
  • Transparency: NEAR’s intents keep all transactions verifiable, with rankings and trades permanently recorded on-chain.

The process lowers the barrier for participation. Users don’t need technical knowledge of AI models—they only need to assess outcomes.

Who Benefits from the System

Users

Users gain an accessible entry point into AI-powered finance. By predicting outcomes, they can earn rewards without needing to design or train AI systems. Over time, they can also invest in consistently strong agents.

Builders

Developers benefit from visibility and fair incentives. Instead of competing with louder marketing campaigns, performance alone determines success. Strong agents gain adoption and direct rewards.

The Crypto Ecosystem

For the wider industry, the system introduces a trust framework for AI in decentralized finance. By ranking AI agents based on transparent performance, it reduces hype cycles and builds credibility for AI-integrated finance.

Technical Foundation: Why NEAR

The market runs on NEAR Protocol, a sharded Layer 1 blockchain with 600ms block times. This speed supports frequent portfolio updates and prediction adjustments.

Core technologies include:

  • Shade Agents: Enable cross-chain execution with privacy-preserving features.
  • NEAR Intents: Handle execution logic and ensure verifiability.
  • Oracles: Pull live price feeds from Binance to update trading portfolios.

Why Prediction Markets Fit AI Agents

Prediction markets have long been used in crypto for trading on outcomes—from election results to sports games. Their strength lies in aggregating user insight into accurate results.

Applying them to AI agents means:

  • Users don’t need to evaluate code or models.
  • Performance is measured in verifiable financial outcomes.
  • Markets self-correct by rewarding accuracy.

This design shifts AI agents from being speculative narratives to demonstrated performers.

Broader Context: Fraction AI’s Role in Decentralized AI

Fraction AI is already known for its decentralized approach to AI training. The platform allows users to create and train AI models without coding, using competitive reinforcement learning. Since launching on Base in May 2025, over 320,000 users have engaged with its testnet.

The company positions itself as a “decentralized ScaleAI,” combining:

  • Community-driven data creation
  • Decentralized reinforcement learning (RLAF)
  • Low-cost, scalable training with QLoRA adapters

This aligns with crypto principles of community ownership, permissionless participation, and verifiable systems.

Why Decentralization Matters in AI

Traditional AI development is highly centralized. Tech giants like Google, OpenAI, and Meta control:

  • Expensive labeled datasets
  • Proprietary infrastructure
  • Advanced training models accessible only to large teams

Fraction AI seeks to decentralize this process, much like DeFi decentralized banking. By opening AI agent training and competition to anyone with simple prompts, it reduces barriers to entry while creating transparent results.

Conclusion

Fraction AI has unveiled the crypto industry's first AI agent prediction market, showing how artificial intelligence performance is evaluated and compensated. Powered by NEAR Protocol's robust infrastructure, the platform delivers real-time portfolio tracking, transparent performance rankings, and user-driven prediction markets with tangible rewards. The system prioritizes results over marketing hype by implementing merit-based incentives for AI builders.

This fusion of prediction markets and AI agents establishes a transparent, equitable ecosystem where trust stems from verifiable, measurable outcomes rather than promotional claims. The platform establishes new performance benchmarks for decentralized artificial intelligence and showcases NEAR Protocol's capabilities as the backbone for sophisticated, real-time AI-powered financial applications.

Resources:

  1. Fraction AI’s AI Agent prediction market launch announcement: https://x.com/FractionAI_xyz/status/1962916033976795484?t=b68iJEbsvEO21Sc7-2EI_w&s=19

  2. Fraction AI Lightpaper - "Fraction AI: Decentralized Auto-Training Platform for AI Agents"

  3. Fraction AI Official Website - fractionai.xyz

  4. Fraction AI Official X Account - @FractionAI_xyz

Frequently Asked Questions

Disclaimer

Disclaimer: The views expressed in this article do not necessarily represent the views of BSCN. The information provided in this article is for educational and entertainment purposes only and should not be construed as investment advice, or advice of any kind. BSCN assumes no responsibility for any investment decisions made based on the information provided in this article. If you believe that the article should be amended, please reach out to the BSCN team by emailing [email protected].

Author

Soumen Datta

Soumen has been a crypto researcher since 2020 and holds a master’s in Physics. His writing and research has been published by publications such as CryptoSlate and DailyCoin, as well as BSCN. His areas of focus include Bitcoin, DeFi, and high-potential altcoins like Ethereum, Solana, XRP, and Chainlink. He combines analytical depth with journalistic clarity to deliver insights for both newcomers and seasoned crypto readers.

Project & Token Reviews

Learn about the hottest projects & tokens

Join our newsletter

Sign up for the very best tutorials and the latest Web3 news.

Subscribe Here!
BSCN

BSCN

BSCN RSS Feed

BSCN is your go-to destination for all things crypto and blockchain. Discover the latest cryptocurrency news, market analysis and research, covering Bitcoin, Ethereum, altcoins, memecoins, and everything in between.