How It Works

​

System flow

BETTER combines data ingestion, signal processing, and low-latency execution into a single pipeline.

​

1. Data ingestion

The system is designed to monitor blockchain mempools and prediction market order books in real time:

• On-chain transactions: Wallet movements, contract interactions, insider activity
• Order book data: Bids, asks, liquidity depth across Polymarket’s CLOB
• External feeds: News APIs, social sentiment, event data

For Terminal launch, production routing is co-located in the same data centre facility as the relevant order books.

​

2. Signal processing (BRAID Protocol)

BRAID (Bounded Reasoning for Autonomous Inference and Decisions) is the constrained decision framework used by BETTER’s OpenServ agents. Rather than “free-form” automation, BRAID requires each trade to pass a verifiable sequence of checks (for example: liquidity depth, slippage, correlation, and risk limits) before it can be executed. Raw data is processed through the OpenServ AI agents running the BRAID protocol. The system builds a Reasoning DAG (Directed Acyclic Graph) that:

• Assigns Insider Scores to wallets based on historical accuracy
• Calculates liquidity depth to estimate slippage
• Maps risk correlations across existing positions
• Generates actionable signals with confidence scores

Signals are processed continuously to generate trade decisions.

​

3. Terminal display

Processed signals flow to the Terminal in real time. Users with access can see:

• Live signal feed with context (wallet scores, liquidity, correlation)
• Which signals the AI agents are acting on
• Copy-trading opportunities for manual execution

​

4. Vault execution

For vault depositors, the execution engine automatically acts on qualified signals:

• Signals pass through logic gates (liquidity check, risk check)
• Qualified trades execute via the Rust-based high-frequency trading engine
• Positions are managed 24/7 without human intervention

---

​

User flow

1. Deposit (Base): Users deposit USD Coin into a BETTER vault on Base.
2. Execute (Prediction Markets): The pipeline processes signals and executes trades on behalf of vault depositors.
3. Account (Optional): A mirrored structure can issue receipt tokens representing vault performance.

​

Execution engine

The execution system is built in Rust for deterministic performance and low-latency execution.

• Latency (public claim): approximately 0.6 milliseconds tick-to-trade, measured from mempool read to transaction hash on Base, then to order finality on Polymarket (Polygon)
• Co-location: production routing is co-located in-region for Terminal launch
• Uptime: 24/7 execution with automated monitoring

​

AI agent stack (BRAID Protocol)

OpenServ AI agents use the BRAID protocol (Bounded Reasoning for Autonomous Inference and Decisions) to process on-chain signals through a Reasoning DAG:

1. Ingest: Signal received. Wallet 0x123 (Insider Score 99/100) bought YES on Contract Y.
2. Liquidity Gate: Is depth sufficient to enter without >2% slippage? If NO → ABORT.
3. Risk Gate: Does this correlate >0.8 with existing positions? If YES → ABORT.
4. Execution: Fire transaction.

​

Mirror vault design

The Mirror Vault separates where capital lives from where liquidity and composability are best:

• Feeder (Real Capital): The Base vault holds funds and drives trading.
• Master (Public Receipt): A Polygon-side vault observes performance and mints an ERC-20 receipt (e.g. vBETTER).

​

Cross-chain execution

BETTER uses Enclave MagicSpend++ to coordinate cross-chain execution. Users deposit USD Coin on Base while the system handles bridging and rebalancing as needed.

​

Why receipt tokens matter

Receipt tokens make strategy exposure:

• Easier to transfer
• Tradable 24/7
• Compatible with DeFi primitives (borrowing, LPing, etc.)