Numinous
SN6AI agents that predict future events, aiming to outperform human forecasters
A forecasting protocol where AI agents compete inside sandboxes to predict real-world events, scored ruthlessly on accuracy. Winner takes all.
// Forecasting agents, not forecasting models
Numinous is Subnet 6 on Bittensor. It runs an open-source competition where miners submit Python-based forecasting agents that predict real-world events: geopolitical developments, sports outcomes, commodity prices, and crypto markets. Validators execute these agents in isolated sandboxes and score them on prediction accuracy.
The simple version: It is like a prediction tournament where AI agents compete daily, and only the most accurate one gets paid.
Centralized equivalent: Metaculus or Polymarket's prediction markets, but the forecasters are autonomous AI agents competing on a decentralized network.
How it works:
- Miners submit forecasting agent code that runs once per event, producing probability predictions (0.0 to 1.0) for binary outcomes
- Validators execute miner agents in Docker sandboxes with no internet access, score predictions using Brier score, and set weights based on forecasting performance
- The problem it solves: There is no open, competitive infrastructure for evaluating AI forecasting agents against each other under controlled conditions. Existing prediction markets measure crowd consensus, not agent quality. Numinous isolates the reasoning architecture itself and benchmarks it against real outcomes.
- The opportunity: Accurate forecasting has downstream applications across finance, insurance, geopolitical risk assessment, and supply chain planning. If Numinous can consistently produce better agents and more useful aggregated signals, the output becomes commercially valuable. The official product surface already includes a forecasting website and broader forecasting infrastructure, though public customer traction still appears early.
- The Bittensor advantage: Decentralization enables a permissionless competition where anyone can submit an agent. The open-source requirement creates a self-improving ecosystem: miners build on each other's code, and the collective performance ratchets upward over time. The sandbox model ensures fair evaluation without miners gaming data access.
- Traction signals: The leaderboard and public product surface are live, the subnet has official architecture and rules documentation, and the team is building around a real forecasting workflow rather than a thin demo. Current market metrics are weaker than when this article was first written, with price at 0.005838 TAO, market cap around 26,743 TAO, and negative 7-day net flow of -564 TAO. That combination says the product is shipping, but market conviction is currently soft.
Category: Financial Forecasting and Trading Signals | Centralized Competitor: Metaculus, Polymarket, Good Judgment Open
Forecasting is one of the hardest problems in AI. Traditional prediction markets rely on human participants or opaque proprietary models. There is no open, competitive marketplace where autonomous forecasting agents can be evaluated on equal footing with standardized tools and data access. Numinous attempts to fill that gap by turning forecasting into a meritocratic, winner-takes-all competition on Bittensor.
Mechanism:
Miners write Python agents that are uploaded to the network and executed by validators inside Docker containers. These sandboxes have no direct internet access. Instead, agents interact with the outside world through a signing proxy that routes requests to approved services. Official docs explicitly mention Chutes, Desearch, OpenAI, Vericore, LunarCrush, OpenRouter, and Numinous Signals as supported gateway services. Each agent has a 240-second execution timeout. TAO.app says the protocol processes approximately 100 events daily across geopolitical, political, and Polymarket-style events. Predictions are scored using the Brier score, and official subnet docs describe a winner-takes-all mechanism based on average performance over a rolling window of events. All agent code is open-source, meaning miners can see and learn from each other's approaches.
The subnet is building out a richer forecasting stack around the core competition. Current public materials highlight Numinous Signals as a native scored-news input, alongside broader gateway access to research, inference, and verification services. The architecture and rules docs also make clear that tracks, gateway allowlists, and execution constraints are becoming more structured, which matters because the product is not just a leaderboard, it is an operating environment for forecasting agents.
- Zero emissions. Numinous currently receives 0.00% emissions with negative net TAO flows (-564.43 TAO over 7 days), meaning the subnet is not attracting new capital under the Taoflow model. Without a reversal in staking sentiment, the economic incentive for miners and validators erodes.
- Small team, high surface area. Public repo history still shows a small contributor base, and the product spans agent execution, gateway tooling, event generation, scoring, and external-facing surfaces. That is a lot of moving parts for a young subnet.
- Code copying tension. The open-source, winner-takes-all design creates a real free-rider risk. If strong agents can be cheaply imitated, the subnet has to work harder to preserve incentives for original research and better forecasting architectures.
- Miner profitability. Cost control remains a live issue in any forecasting subnet that relies on sandboxed agent execution plus external APIs. With market cap now around 26,743 TAO and emissions currently at zero, miners need a credible path to positive expected value, not just leaderboard prestige.
