Indicator Engine

The Indicator Engine (src/indicators/) is responsible for providing the deterministic "ground truth" that the AI uses for its reasoning. It transforms raw price movements into technical signals.

Local vs. API-Based Calculation

Unlike many trading bots that rely on external Technical Analysis (TA) APIs, Rez calculates indicators locally.

Why Local?

1. Reduced Latency: No network round-trips for every indicator lookup.

2. Cost Efficiency: Eliminates dependencies on paid TA API tiers.

3. Accuracy: Rez controls exactly how indicators are calculated, ensuring consistency with the exchange's data.

The LocalIndicatorCalculator

The calculation process follows a three-step pipeline:

1. Raw Data Fetching

Rez pulls raw OHLCV (Open, High, Low, Close, Volume) data directly from the Binance /klines API (the industry standard for depth and reliability).

2. Warmup & Series Generation

To ensure accuracy, Rez implements a "warmup period" for recursive indicators like EMA and RSI.

• Formula: If 10 results are requested, Rez fetches 10 * 3 = 30 candles.

• Rationale: This ensures that the initial "cold" values of the exponential averages have vanished, and the current value matches what a trader would see on an exchange chart.

3. Processing with pandas-ta

Rez leverages the pandas-ta library, a high-performance technical analysis library built on top of Pandas. This allows for complex calculations like MACD or ATR to be completed in milliseconds.

Supported Indicators

Indicator	Short Name	Key Use Case
Exponential Moving Average	ema	Trend identification and Support/Resistance.
Relative Strength Index	rsi	Overbought/Oversold levels.
MACD	macd	Momentum crossovers.
Average True Range	atr	Volatility-adjusted Stop-Loss placement.

Tool-Calling Integration

The Indicator Engine is exposed to the TradingAgent via the fetch_indicator tool. When an LLM calls this tool, it passes the asset, indicator_name, and interval (e.g., 5m, 1h, 4h). The engine calculates it on the fly and returns the precise value.