Concepts

Understanding how Rez makes trading decisions.

---

Decision Engine

The agent uses a ReAct (Reasoning + Acting) loop:

1

Observe

Gather market data, positions, indicators

2

Think

LLM analyzes context and forms thesis

3

Act

Execute trade or hold

4

Reflect

Log outcome for learning

---

Technical Indicators

Rez calculates these indicators in real-time:

Indicator	Timeframes	Usage
RSI	5m, 4h	Overbought/oversold detection
EMA	5m, 4h	Trend direction
MACD	5m, 4h	Momentum and crossovers
ATR	5m	Volatility measurement

---

Risk Profiles

Conservative

• Smaller position sizes

• Tighter stop-losses

• Focus on high-confluence setups

Moderate

• Balanced risk/reward

• Standard position sizing

• Accepts moderate drawdowns

Aggressive

• Larger positions

• Wider stops

• Pursues higher returns

---

Position Management

Entry

Market orders with slippage tolerance (default 5%).

Take Profit

Trigger orders placed automatically based on risk/reward ratio.

Stop Loss

Protective orders to limit downside.

Exit

Positions can be closed:

• Automatically by TP/SL triggers

• Manually via dashboard

• By agent decision based on market change

---

Data Flow

Market Data → Indicators → Context Builder → LLM → Decision → Execution
                              ↑                         ↓
                         Account State ←──────── Reconciliation

---

Session Management

Each wallet address is a unique session. The backend:

• Persists agent configuration across restarts

• Maintains one agent per session

• Syncs state across multiple browser tabs

• Stores trade history in per-user diary files