Hi,
I’m excited to announce OnlineResamplers.jl, a high-performance Julia package for real-time resampling of financial market data! 
What is OnlineResamplers.jl?
OnlineResamplers.jl provides efficient streaming algorithms for aggregating tick-level market data into OHLC candlesticks and other time-based formats. Built on top of OnlineStatsBase.jl, it offers constant memory usage and zero-allocation operations for processing financial data streams.
Key Features
Real-time Processing: Stream market data with constant memory usage - no need to store historical data in memory
Multiple Resampling Methods:
- OHLC (Open, High, Low, Close) candlesticks
- Mean price aggregation
- Volume sum aggregation
Parametric Types: Full support for custom numeric types like FixedPointDecimals.jl for precise financial calculations
High Performance: Type-stable operations with zero allocations during steady-state processing
Parallel Processing: Built-in merge operations for distributed computing scenarios
OnlineStatsBase Integration: Seamless compatibility with Julia’s online statistics ecosystem
Quick Example
using OnlineResamplers OnlineStatsBase, Dates
# Create a 1-minute OHLC resampler
resampler = MarketResampler(Minute(1))
# Process streaming market data
data = MarketDataPoint(DateTime(2024, 1, 1, 9, 30, 0), 100.0, 1000.0)
fit!(resampler, data)
# Get results
result = value(resampler)
println("OHLC: $(result.price.ohlc)") # OHLC(100.0, 100.0, 100.0, 100.0)
println("Volume: $(result.volume)") # 1000.0
Why OnlineResamplers.jl?
Traditional approaches to market data resampling often require loading entire datasets into memory or using complex windowing mechanisms. OnlineResamplers.jl takes a different approach by:
- Processing data point-by-point as it arrives
- Automatically handling time window transitions
- Maintaining constant memory usage regardless of data volume
- Supporting high-precision numeric types for financial applications
This makes it ideal for:
- Real-time trading systems
- Market data processing pipelines
- Financial analysis applications
- Any scenario requiring efficient time-series aggregation
Advanced Usage
The package supports sophisticated scenarios like custom numeric types:
using FixedPointDecimals
# High-precision resampler
resampler = MarketResampler{DateTime, FixedDecimal{Int64,4}, FixedDecimal{Int64,2}}(
Minute(1), price_method=:ohlc
)
# Process high-precision data
data = MarketDataPoint{DateTime, FixedDecimal{Int64,4}, FixedDecimal{Int64,2}}(
DateTime(2024, 1, 1, 9, 30, 0),
FixedDecimal{Int64,4}(100.5012),
FixedDecimal{Int64,2}(1000.50)
)
Installation
using Pkg
Pkg.add(url="https://github.com/femtotrader/OnlineResampler.jl")
Documentation & Examples
Documentation: Available in the docs/ directory
API Reference: Comprehensive function documentation
Tutorial: Step-by-step guide from basic to advanced usage
Examples: Real-world usage scenarios
Performance
OnlineResampler.jl is designed for high-frequency data processing. The streaming approach means:
- O(1) memory usage per resampler instance
- Zero allocations during steady-state processing
- Type-stable operations for maximum performance
- Parallel processing support via merge operations
Get Involved
The package is open source and welcomes contributions! Whether you’re working with financial data, time-series analysis, or online statistics, I’d love to hear about your use cases and feedback.
Repository: GitHub - femtotrader/OnlineResamplers.jl: high-performance Julia package for real-time resampling of financial market data
Issues: GitHub · Where software is built
Built with 
for the Julia community
Would love to hear your thoughts and feedback! Has anyone been working on similar streaming data aggregation problems?