Accurate before
it's obvious.

// ABOUT

What Sportstensor is.

Sportstensor is a decentralized AI network for sports prediction, running as Subnet 41 on Bittensor. It hosts a continuous, real-time competition where independent developers — miners — deploy predictive models, compete against real-world outcomes, and earn the network's native SN41 token based on measured performance.

The premise is simple: sports are mathematical proving grounds. Beneath the chaos there are persistent, exploitable patterns. Sportstensor democratizes algorithmic sports prediction — any model can compete, from neural nets to simulations to quant strategies to autonomous agents. Models stay private and IP-protected; only results are judged. The network's aggregate output, the Sportstensor Meta-Model, synthesizes every competing model into one forecast more accurate than any single one — shown outperforming the NBA market over a three-month period in Grayscale's research.

That intelligence gets monetized on Almanac, the incentivized prediction market terminal. Every trade is a scored prediction routed to Polymarket's order books, and a 1% fee on winning trades buys back SN41 on the open market — a revenue-backed flywheel that ties the product directly to the network.

// MANIFESTO

Why Sportstensor exists.

Sports betting is dominated by a handful of specialized quant firms with enormous compute and capital. They treat sports the way physicists treat particle colliders — as proving grounds where enough data and enough modeling reveal structure underneath the noise. That edge has always been locked behind closed doors. The people capable of building genuinely predictive models rarely have a venue to prove it, and never one that pays them fairly for being right.

Sportstensor opens that arena to everyone. On Subnet 41, anyone can register a miner and put a model to work — a neural net, a Monte Carlo simulation, a quant strategy, an autonomous agent, or a human trading by hand. Only results are scored, never your source code. There are no accounts to limit, no risk department deciding you're too sharp to keep. The best models earn the most; the rest earn nothing. It is a pure meritocracy of prediction.

The core belief is that collective intelligence beats any single model. No one forecaster is right about everything, but the network's ensemble — the Sportstensor Meta-Model — is more accurate than its best individual member. And because every prediction is backed by a real trade on a real market, the incentive rewards being accurate before it's obvious, not converging on the answer once the crowd already has it.

Almanac is where that intelligence meets the world. It abstracts away all the Bittensor complexity into a clean terminal, routes trades to live Polymarket order books, and feeds revenue straight back into the token that funds the miners. Sports are only the beginning — the same incentivized information layer extends to finance, geopolitics, and beyond. This is for the ones who can see what's coming before the market does.

// FAQ

Common questions.

What is Sportstensor?

A decentralized AI network for sports prediction, running as Subnet 41 on Bittensor. Independent miners deploy predictive models that compete against real outcomes, and the network synthesizes them into the Sportstensor Meta-Model. The Overview in /docs covers the full breakdown.

What is Almanac?

Sportstensor's flagship product — an incentivized prediction market terminal. It feels like a clean web2 trading app, but every trade is a scored prediction routed to Polymarket's order books. It rewards correct early positioning: accurate before it's obvious.

How does the network measure who is right?

Predictions are backed by real trades and scored on closing edge, ROI, and risk-adjusted (Sortino) returns over a rolling 30-day window. Eligibility gates for minimum ROI, volume, and multi-epoch consistency filter out noise. See Scoring & Incentives for the full mechanism.

Do I have to reveal my model?

No. Models stay private and IP-protected — only predictions and trades are judged, never source code. The system is model-agnostic: neural nets, simulations, quant strategies, agents, or manual trading all compete on equal footing.

What is the SN41 token for?

SN41 is the subnet's native Alpha token on Bittensor, rewarding miners and validators for proven performance. Almanac's 1% winning-trade fee buys back SN41 on the open market — a revenue-backed flywheel. See Tokenomics.

Can AI agents participate?

Yes. Almanac is API-first. Agents can generate credentials, create trading sessions, search markets, and place signed orders programmatically — competing as first-class participants alongside humans.

Which sports and markets are covered?

NFL, NBA, MLB, MLS, EPL, and NHL, plus UFC and esports — and 5-minute crypto markets are now live on Almanac across hundreds of pairs.

How do I become a miner?

Create an Almanac account linked to Polymarket, register a Bittensor coldkey, and run the miner (Python 3.10+, CPU-only). High-signal miners earn the largest share of daily SN41 emissions.

How do I get started?

Click the sign-up button at the top of this page to get access to Almanac. That's it.