CCXT Exchange Adapter

The CCXT Exchange Adapter provides a standardized interface between the backtest-kit framework and the CCXT library, specifically configured for Binance. It implements a reusable ccxt-exchange schema that handles market data retrieval, order book fetching, and precision formatting for prices and quantities. This adapter is shared across data dumping, strategy optimization, and forecast testing modules.

Singleton Exchange Initialization

To prevent redundant network connections and repeated market loading, the adapter utilizes a singleshot singleton pattern to initialize the Binance exchange instance.

Feature	Configuration	File Reference
Exchange	ccxt.binance
Market Type	spot
Time Sync	adjustForTimeDifference: true
Rate Limiting	enableRateLimit: true

The getExchange function ensures that exchange.loadMarkets() is called exactly once before any trading or data fetching operations occur.

Exchange Schema Implementation

The adapter is registered using addExchangeSchema, defining how the system interacts with the physical exchange.

OHLCV Data Mapping

The getCandles function maps CCXT's fetchOHLCV output (an array of arrays) into the structured object format required by the backtest-kit engine.

Order Book with Backtest Guard

The getOrderBook implementation includes a safety check to prevent execution during backtests, as historical order book data is not supported in the default schema. In live contexts, it fetches and formats asks and bids as string-based price/quantity pairs.

Precision and Tick Formatting

To ensure orders are accepted by the exchange, the adapter provides formatPrice and formatQuantity. These functions retrieve tickSize and stepSize from the market metadata.

• Priority 1: Uses market.limits or market.precision to find the minimum increment.
• Priority 2: Applies roundTicks to align the value with the exchange's required increments.
• Fallback: Uses CCXT's built-in priceToPrecision or amountToPrecision methods.

Code Entity Mapping: Exchange Integration

This diagram maps the natural language requirements of exchange interaction to the specific code entities and CCXT methods used in the implementation.

Exchange Data Flow

Usage in Forecast Testing

The run_forecast.ts script utilizes the ccxt-exchange schema within a runInMockContext wrapper. This allows developers to test the LLM's forecasting logic against real-world historical data by mocking the execution environment (symbol and timestamp) while providing the actual exchange adapter for data resolution.

Forecast Mock Context Flow

Implementation Summary Table

Module	Schema Role	Key Functions
dump.module.ts	Data Extraction	getCandles
walker.module.ts	Strategy Optimization	getCandles, getOrderBook, formatPrice, formatQuantity
pine.module.ts	Signal Generation	getCandles
run_forecast.ts	Logic Validation	getCandles, formatPrice, formatQuantity