Astrid
SN127Evaluates trading strategies in competition where the best performers rise to the top
A trading strategy evaluation platform that wants to be Bittensor's capital axis, with a working validator stack and a mining layer still being prepared.
// Trading strategy contests, sandboxed onchain.
Astrid is subnet 127 on Bittensor, also branded as SigmaArena, a platform for evaluating trading strategies in sandboxed Docker environments using real market data. The on-chain identity describes it as "the capital axis for Bittensor."
The simple version: Imagine an open arena where coders submit trading strategies, the network runs them on real market data in isolated containers, and the best performers earn rewards.
Centralized equivalent: QuantConnect or Numerai, but with the contest and the rewards living on a subnet rather than inside one company.
How it works:
- Miners are not yet publicly documented. The team's TAO.app about page lists this side as "Coming soon."
- Validators run the Astrid validator daemon, which executes trading strategy simulations in Docker sandboxes via the SigmaArena VM, validates on-chain transactions, and manages weight commitments on subnet 127.
- The problem it solves: Strategy backtesting and evaluation today happens inside closed platforms or private trading firms. Astrid aims to make the contest open, with verifiable execution and on-chain rewards for the strategies that actually perform.
- The opportunity: Trading strategy evaluation is a real industry surface, used by quant shops, prop traders, and signal services. A subnet-native version turns the leaderboard into an incentive layer.
- The Bittensor advantage: Strategies are executed in isolated, deterministic containers, with slippage and order modeling, so results are reproducible and the scoring can be settled on-chain rather than trusted to a single host.
- Traction signals: The validator codebase is in active development, with the most recent commit on 2026-04-27. Two contributors are visible on the main repo, and the SigmaArena VM lives in a second repository. Public on-chain emission allocation is currently 0%, reflecting Taoflow net flows, while the AMM pool holds roughly 4,192 TAO.
Category: Financial Forecasting and Trading Signals | Centralized Competitor: QuantConnect, Numerai
Astrid sits in a crowded but real category. Trading strategy evaluation has obvious demand inside quant funds, prop desks, and signal marketplaces, and several Bittensor subnets already touch adjacent ground in forecasting and market prediction. Astrid's framing, "the capital axis for Bittensor," positions it as financial infrastructure rather than a pure inference subnet.
Mechanism:
According to the subnet's GitHub README, validators run a TypeScript daemon that registers with the Astrid coordination service, polls for tasks, and executes them in isolated Docker containers. The task types described in the README include Docker image execution, NPX runs, simulated trades using the SigmaArena VM with OHLCV market data, and on-chain transaction validation. Validators report results back to the coordinator, maintain heartbeats, and commit weights on subnet 127 at configurable intervals.
The SigmaArena VM, in a separate repository, is described on the team's TAO.app page as a deterministic simulation engine with order execution modeling, slippage simulation, and CCXT-based exchange integration. The published roadmap lists multi-exchange expansion, advanced order types, real-time execution, AI signal integration, portfolio optimization, and risk management as forward items.
The miner side is the open question. The TAO.app about page lists "Miners: Coming soon," and the README focuses on the validator role. On-chain data shows zero active miners on TaoSwap at the time of this snapshot, consistent with a subnet whose mining incentive layer has not yet gone live. The pool itself is mostly organic, with a root proportion around 0.23, meaning most of the depth is alpha demand rather than protocol subsidy. Price is 0.00464 TAO with a market cap near 15,147 TAO, up about 16% over 90 days but down 15% over the last 30.
The team listed on the official about page is a single founder, Leo Mercier. GitHub shows two contributors on the main subnet 127 repository. Community discussion on X around SN127 references a wider operational entity and bridge plans, but those claims are not yet present on the team's own about page or in the main repo's documentation, so this article treats them as unverified.
- Pre-launch mining: The TAO.app about page explicitly lists miners as "Coming soon" and the public README documents the validator role rather than miner submissions. Until the mining incentive layer is specified and live, the subnet's competition surface is incomplete.
- Execution: Building a credible trading evaluation platform requires exchange relationships, reliable market data feeds, and careful handling of strategy IP. The roadmap is ambitious, with live trading and AI integration listed as future items.
- Competition: Trading and financial forecasting is a contested category, both inside Bittensor and against well-funded centralized backtesting and signal platforms. Astrid will need a defensible edge beyond running strategies in containers.
- Emission allocation: Current Taoflow emission share for subnet 127 is 0%, meaning recent net flows have not earned the subnet an allocation. This is a function of the flow-based emission model and can shift if staking inflows pick up, but it limits near-term miner economics if and when the mining layer opens.
Into the next one.
