rangebar

High-performance non-lookahead bias range bar construction for Binance UM Futures data.

[!NOTE] This crate processes 137M+ trades/second using fixed-point arithmetic for financial-grade precision.

[!WARNING] Range bars use non-lookahead bias - thresholds are computed only from bar opening prices, never from evolving high/low ranges.

Quick Start

Add this to your Cargo.toml:

[dependencies]
rangebar = "0.4.1"

Basic Usage

use rangebar::{RangeBarProcessor, AggTrade, FixedPoint};

// Create processor with 0.8% threshold (8000 basis points)
let mut processor = RangeBarProcessor::new(8000);

// Create sample trade data
let trade = AggTrade {
    agg_trade_id: 123456789,
    price: FixedPoint::from_str("50000.12345").unwrap(),
    volume: FixedPoint::from_str("1.50000000").unwrap(),
    first_trade_id: 100,
    last_trade_id: 105,
    timestamp: 1609459200000,
    is_buyer_maker: false,
};

// Process trades into range bars
let trades = vec![trade];
let bars = processor.process_trades(&trades).unwrap();

for bar in bars {
    println!("Bar: O={} H={} L={} C={} V={}",
             bar.open, bar.high, bar.low, bar.close, bar.volume);
}

Algorithm

Range bars close when price moves ±threshold% from the bar's opening price:

1. Non-lookahead bias: Thresholds computed only from bar open price
2. Breach inclusion: Breaching trade included in closing bar
3. Fixed thresholds: Never recalculated during bar lifetime

Features

• statistics (default): Comprehensive statistical analysis via Polars
• data-integrity (default): Data validation and checksums
• arrow-support: Apache Arrow and Parquet export
• python-bindings: PyO3 Python bindings (optional)

Performance

• 137M+ trades/second processing (2025 benchmarks)
• Fixed-point arithmetic (no floating-point errors)
• Memory efficient streaming processing
• Zero-copy design where possible

Data Source

Designed for Binance UM Futures aggTrades data:

// Sample aggTrade format
{
    "a": 123456789,     // Aggregate trade ID
    "p": "50000.12345", // Price
    "q": "1.50000000",  // Quantity
    "f": 100,           // First trade ID
    "l": 105,           // Last trade ID
    "T": 1609459200000, // Timestamp
    "m": false          // Is buyer maker
}

License

MIT license. See LICENSE for details.