FX Rate Storage

Comparison of FX rate storage methods: In-Memory vs HDF5.

Storage Comparison

Feature	InMemoryFXRateReader	HDF5FXRateReader
Speed	⚡⚡⚡ Fastest	⚡⚡ Fast
Capacity	Limited by RAM	~TB scale
Persistence	❌ No	✅ Yes
Setup	Simple dictionary	HDF5 file required
Use Case	Testing, prototyping	Production

In-Memory Storage

from rustybt.data.fx import InMemoryFXRateReader
import pandas as pd

# Create rate data structure
data = {
    'mid': {  # Rate name
        'USD': pd.DataFrame({  # Quote currency
            'EUR': [1.08, 1.09, 1.10],  # Base currencies
            'GBP': [1.25, 1.26, 1.27],
        }, index=pd.date_range('2024-01-01', periods=3))
    }
}

reader = InMemoryFXRateReader(data=data, default_rate='mid')

Pros: - Instant access - No I/O overhead - Simple setup

Cons: - Not persistent - Limited by memory - Rebuild on restart

HDF5 Storage

from rustybt.data.fx import HDF5FXRateWriter, HDF5FXRateReader

# Write rates
writer = HDF5FXRateWriter('/path/to/fx_rates.h5')
writer.write(
    rate='mid',
    quote='USD',
    rates=df  # DataFrame with dates index, currency columns
)

# Read rates
reader = HDF5FXRateReader('/path/to/fx_rates.h5')

Pros: - Persistent storage - Efficient compression - Fast random access - Scalable to large datasets

Cons: - File management required - Initial write overhead - More complex setup

HDF5 File Structure

fx_rates.h5
├── /data
│   ├── /mid
│   │   ├── /USD
│   │   │   └── /rates (2D array)
│   │   ├── /EUR
│   │   │   └── /rates (2D array)
│   ├── /bid
│   │   └── ...
│   └── /ask
│       └── ...
└── /index
    ├── /dts (timestamps)
    └── /currencies (currency codes)

See FX Overview for usage examples.