Subnet
Introduction
TensorUSD Subnet SN113 that enables Decentralized Market Maker (DMM) for TAO-backed stablecoins. Built on the Bittensor network, it uses a miner-validator architecture to create a trustless marketplace where miners act as a Decentralized Market Maker (DMM) to fulfill buy and sell orders for ALPHA tokens in exchange for rewards.
TensorUSD Subnet is a specialized network that:
Incentivizes participation through performance-based rewards
Ensures quality using validator scoring
Maintains decentralization through cryptographic verification
Subnet Architecture
The TensorUSD subnet consists of four main components.
Core Components
Miners
Miners hold TAO and ALPHA tokens, fulfill orders, execute on-chain transactions, and earn rewards based on performance.
Validators
Validators score miners using EMA-based metrics and set weights on the Bittensor chain.
Platform
The TensorUSD platform distributes orders, authenticates miners, tracks order states, and supplies performance data to validators.
Bittensor Network
The blockchain layer that registers participants, executes staking operations, and distributes emissions.
Benifits of Subnet to:
Miners
Earn Rewards: Generate income by being Decentralized Market Maker
Flexible Participation: Sets own limits and availability
Transparent Scoring: Understand exactly how you're evaluated
Decentralized: No central authority controlling your participation
Validators
Support Network: Contribute to subnet health and fairness
Earn Emissions: Receive rewards for validation work
Influence Quality: Help maintain high standards across the network
Net Flow–Based Performance Evaluation
To ensure reliable market making provision and consistent order execution within the TensorUSD subnet, miner performance is evaluated using a Net Flow–based scoring system. This mechanism directly measures a miner’s ability to accept and successfully complete alpha swap transactions.
Net Flow Calculation
Exponential Moving Average (EMA)
To avoid penalizing miners for short-term disruptions while still capturing long-term behavior, the subnet applies an Exponential Moving Average (EMA) to net flow values.
Validator Evaluation Cycle = 150 Blocks (30 Mins)
Alpha (⍺) = 0.001
EMA[t] = Alpha * Net FLow + (1 - Alpha) * EMA[t-1]
Last updated