As of September 2025, there are around 128 Bittensor subnets, but only a subset focuses explicitly on prediction, forecasting, or probabilistic modeling.
Based on recent ecosystem analyses, the primary prediction-related subnets are listed below. These specialize in areas like financial markets, sports, weather, and events.
| Subnet NetUID | Name | What It Predicts | Examples |
|---|---|---|---|
| SN6 | Infinite Games | Binary outcomes of future events on decentralized prediction markets, using AI to generate probabilistic forecasts. Miners submit predictions, validated against real outcomes. | Election results (e.g., “Will Candidate X win?”), sports events (e.g., “Will Team Y score over 2.5 goals?”) on platforms like Polymarket or Azuro. |
| SN8 | PTN (Proprietary Trading Network / Taoshi) | Intraday financial price movements and trading signals, focusing on decentralized financial forecasting. | Bitcoin (BTC) price changes within trading hours (e.g., “Will BTC rise 2% in the next hour?”). |
| SN18 | Zeus | Environmental and climate variables, using machine learning for high-resolution weather and climate forecasts. Outperforms traditional models by ~40% in error reduction. | 2-meter surface temperature (T2m) for specific regions/time windows (e.g., “Temperature in New York at 3 PM tomorrow”); expanding to precipitation or wind speeds for energy trading or racing. |
| SN41 | Sportstensor | Sports event outcomes, incentivizing AI models to beat market odds through collaborative analytics. Validators sync real-time data every 30 minutes. | Soccer match winners/scores (e.g., “Manchester United vs. Liverpool: Predicted score 2-1”); NFL game spreads. |
| SN44 | Score (ScorePredict / Score Vision) | Soccer (football) match results, focusing on scoreline and outcome predictions via AI vision models. | Premier League fixtures (e.g., “Arsenal vs. Chelsea: Exact score prediction”); goal totals in La Liga games. |
| SN50 | Synth | Probabilistic cryptocurrency price paths, generating synthetic data distributions rather than single-point estimates. Evaluated via Continuous Ranked Probability Score (CRPS). | Bitcoin or Ethereum price trajectories over 24+ hours (e.g., “Distribution of BTC prices from $60K-$70K with 80% confidence over next day”). |
| SN123 | Mantis | Short-term prices of digital assets, fiat currencies, and trading signals (“alpha”) for financial markets. Miners build models for signal processing and prediction markets. | Crypto short-term swings (e.g., “ETH price in next 15 minutes”); fiat-crypto pairs like USD/BTC for day trading. |
These subnets represent the core of Bittensor’s prediction ecosystem, enabling decentralized alternatives to centralized oracles and models.
Did we miss any prediction-related subnet? Let’s know in the comments!
Be the first to comment