VoyagerFi

"AI-powered autonomous trading agent. Deposit, sleep, profit."

VoyagerFi is an autonomous quant trading platform for Southeast Asia's retail crypto market. Users deposit USDC into a shared vault, select a risk profile, and the AI agent — powered by quant indicators and DeepSeek v3 reasoning via 0G Compute TEE — trades perpetuals on Hyperliquid automatically. Every decision is cryptographically anchored on 0G Chain, making the agent's behavior fully verifiable.

---

How It Works

User deposits USDC (Arbitrum Vault)
        ↓
Agent: OBSERVE → THINK → ACT → LOG
  OBSERVE:  Pyth Oracle prices + news sentiment
  THINK:    RSI/MACD/Bollinger + DeepSeek v3 via 0G Compute (TEE)
  ACT:      Execute trades on Hyperliquid (real liquidity, real fills)
  LOG:      Decision hash → 0G Chain + full trace → 0G Storage
        ↓
User withdraws USDC + proportional profit

No manual orders. Every decision is AI-generated and on-chain verifiable.

---

Architecture

Two-Layer Design

Layer	Purpose	Chain
Vault	User custody (deposit/withdraw USDC)	Arbitrum Sepolia
Execution	Trade perpetuals (real liquidity)	Hyperliquid Testnet
Verifiability	Decision hash + audit trail	0G Chain Mainnet
AI Inference	DeepSeek v3 via TEE	0G Compute
Storage	Full decision trace (verifiable)	0G Storage

Why Hyperliquid for Execution?

Building a custom perpetuals engine requires building trust from scratch — liquidity, audit history, protocol risk. Hyperliquid is already battle-tested with billions in volume and a fully programmatic API. VoyagerFi's moat is the AI agent + verifiability layer, not the matching engine. We use the best available infrastructure for each layer.

Why Arbitrum for the Vault?

0G Chain does not have native USDC. Hyperliquid settles in USDC from Arbitrum. Putting the vault on Arbitrum avoids cross-chain bridging complexity and lets users deposit/withdraw USDC natively with no bridge risk.

0G Integration (mandatory components)

• 0G Compute — DeepSeek v3 (deepseek-chat-v3-0324) provides AI reasoning for every trade signal via TEE. This is the verifiable AI inference layer.
• 0G Storage — Full decision trace (indicators, reasoning, signal strength, price, timestamp) uploaded and content-addressed for every decision.
• 0G Chain Mainnet — DecisionLog.sol anchors decision hashes on-chain. StorageAnchor.sol links on-chain records to 0G Storage root hashes.

---

Tech Stack

Component	Technology
Frontend	Next.js 15 + React 19 + TailwindCSS + shadcn/ui
Wallet	RainbowKit + wagmi
Backend	Go 1.25 + Gin + GORM
Database	Supabase PostgreSQL (cache/index only)
Quant Engine	Custom RSI + MACD + Bollinger + MA
AI	DeepSeek v3 via 0G Compute (TEE)
Oracle	Pyth Network (Hermes REST)
Execution	Hyperliquid Testnet (perpetuals)
Vault	Arbitrum Sepolia (USDC)
Verifiability	0G Chain Mainnet + 0G Storage
Smart Contracts	Solidity 0.8.24 + Foundry

---

Business Model

0% management fee. 20% performance fee on net realized profit only.

• Fee only activates when users profit — aligned incentives
• High-water mark rule: users must recover to previous peak before fee applies again
• Standard hedge fund practice, fully trustless via smart contract

Why this works:

• Retail SEA has crypto but lacks trading skill/time
• VoyagerFi = Robinhood of on-chain algo trading: AI-managed, TEE-verified, 0% management fee

B2B2C Roadmap:

• B2C: retail users deposit → bot trades → 20% on profit
• B2B2C: exchanges/wallets white-label VoyagerFi engine
• Institutional: custom risk params, $50k+ minimum

---

Agent Loop

Every 10 seconds, for each trading pair (ETH, BTC, SOL, ARB, BNB):

1. OBSERVE: Fetch live price from Pyth, macro/micro context, news sentiment
2. THINK: Run quant analysis (RSI, MACD, Bollinger, MA) + DeepSeek v3 reasoning
3. COMBINE: 40% quant signal + 60% AI signal = combined direction + strength
4. ACT: If no open position + signal strength > 0.3 → open Hyperliquid position
5. MANAGE: Close position if take profit (≥3%) or stop loss (≤-5%)
6. LOG: Hash decision → 0G Chain + upload trace → 0G Storage

Risk Controls

• Signal threshold: 0.3 (weak signals rejected)
• Take profit: 3%
• Stop loss: 5%
• Leverage: 8–15x (balanced profile)
• Max drawdown: 20%

---

Smart Contracts

Arbitrum Sepolia (Vault)

Contract	Purpose
Vault.sol	USDC custody, share-based NAV, deposit/withdraw

0G Chain Mainnet (Verifiability)

Contract	Purpose
AgentRegistry.sol	Whitelist of authorized agent wallets
DecisionLog.sol	On-chain decision hash anchor
StorageAnchor.sol	Links on-chain records to 0G Storage root hashes

Deployment

# Deploy Vault on Arbitrum Sepolia
forge script script/DeployArbitrum.s.sol --rpc-url $ARBITRUM_RPC_URL --broadcast

# Deploy verifiability contracts on 0G Chain Mainnet
forge script script/DeployOG.s.sol --rpc-url https://evmrpc.0g.ai --broadcast

---

Local Development

Backend

cd backend
cp .env.example .env
# Fill in AGENT_PRIVATE_KEY, DEEPSEEK_API_KEY, DEEPSEEK_URL, contract addresses
go run main.go

Frontend

cd frontend
cp .env.example .env.local
npm install
npm run dev

Smart Contracts

cd smart-contracts
forge build
forge test

---

Project Structure

voyagerfi/
├── backend/
│   ├── service/
│   │   ├── agent/          # Agent loop: OBSERVE → THINK → ACT → LOG
│   │   ├── quant/          # RSI, MACD, Bollinger, MA
│   │   ├── external/
│   │   │   ├── hyperliquid/ # Execution layer (trade placement + signing)
│   │   │   ├── deepseek/    # 0G Compute AI inference
│   │   │   ├── pyth/        # Price oracle
│   │   │   ├── storage/     # 0G Storage upload
│   │   │   └── news/        # News sentiment (RSS + CoinGecko)
│   │   └── chain/           # 0G Chain bindings (DecisionLog, StorageAnchor)
│   └── http/                # Gin REST API + WebSocket
├── frontend/                # Next.js dashboard
└── smart-contracts/         # Foundry (Vault + verifiability)

---

Trading Pairs

ETH/USD · BTC/USD · SOL/USD · ARB/USD · BNB/USD

Risk profiles: Balanced (active) · Conservative (coming soon) · Aggressive (coming soon)

---

Hackathon

0G APAC Hackathon — Track 2: Agentic Trading Arena (Verifiable Finance) Deadline: May 9, 2026 23:59 UTC+8