Caching

CachedProvider wraps any data provider and stores bars in memory after the first iteration. Useful for parameter sweeps where you run the same data multiple times.

Usage

from replaybt import CachedProvider, CSVProvider

data = CachedProvider(CSVProvider("ETH_1m.csv", symbol_name="ETH"))

# First iteration: reads from disk, caches in memory
engine1 = BacktestEngine(strategy=Strategy1(), data=data, config=config)
results1 = engine1.run()

# Second iteration: reads from memory (fast)
data.reset()
engine2 = BacktestEngine(strategy=Strategy2(), data=data, config=config)
results2 = engine2.run()

With Exchange Fetchers

Especially useful with API-based providers to avoid redundant network calls:

from replaybt import CachedProvider, BinanceProvider

data = CachedProvider(
    BinanceProvider("ETHUSDT", start="2024-01-01", end="2024-12-31")
)

# First run fetches from Binance API
# Subsequent runs use cached bars

In Parameter Sweeps

ParameterSweep handles caching internally — you don't need to wrap your provider manually:

from replaybt.optimize.sweep import ParameterSweep

sweep = ParameterSweep(
    strategy_class=MyStrategy,
    data=CSVProvider("ETH_1m.csv"),  # loaded once, shared across workers
    base_config=config,
    param_grid=grid,
)