Crate rustrade_backtest 

rustrade-backtest

Deterministic backtest engine for rustrade_core::Brains. The same Brain trait used by rustrade for live trading drives the backtest — no special “backtest-mode” code paths in the strategy, no duplicate decision logic to keep in sync.

See the crate-level README.md for a quickstart.

Determinism

Given the same (Brain, Vec<Candle>, BacktestConfig) inputs the engine produces the same BacktestResult every run. There is no random number source and no thread interleaving — the loop is single-threaded and synchronous through every step that affects state.

Risk-gate parity with live

The live ExecutionService blocks decisions behind a per-symbol session-PnL halt and circuit breaker. Thread the same configs through BacktestConfig::session_pnl / BacktestConfig::circuit_breaker and the replay engine applies the same gates — driven by candle time (the daily halt rolls over at 00:00 UTC in replay time, the breaker’s window and cooldown run on candle timestamps), so a brain that relies on live gating backtests the way it trades. Blocked decisions are counted in BacktestResult::orders_blocked. Both gates default to off.

Fill models

BacktestConfig::fill_model (builder: .fill_model(FillModel::…)) selects how orders become fills. The default, FillModel::TakerAtClose, keeps the engine’s historical single-candle semantics — existing backtests are bit-for-bit unchanged. FillModel::Resting opts in to honest resting-order semantics: non-marketable limits rest across candles and fill when a later candle crosses their level (at the limit price, or at the candle’s open when it gaps through), stops trigger on a cross and fill at the level or worse, standalone (stop-only / TP-only) protective orders are honoured, and same-candle ambiguity resolves conservatively. See FillModel for the full convention list.

Perp funding

BacktestConfig::funding (builder: .funding(FundingModel::…)) books perp funding cashflows against open positions at settlement timestamps — a historical FundingModel::Series or a constant rate + interval fallback. Positive rate → longs pay, shorts receive; cashflows land in the equity curve, accrue onto each closed trade’s TradeOutcome::funding, and total up in BacktestResult::funding_received / BacktestResult::funding_paid. Defaults to FundingModel::None — existing backtests are bit-for-bit unchanged.

Status

Phase 4b — adds CSV candle loader, Sharpe/Sortino metrics, and multi-symbol replay. Parquet loaders and book-walk slippage remain deferred. See the workspace TODO.md.

Re-exports

pub use config::BacktestConfig;

pub use config::BacktestConfigBuilder;

pub use config::FillModel;

pub use engine::Backtest;

pub use error::Error;

pub use error::Result;

pub use fees::FeeModel;

pub use funding::EIGHT_HOURS_MS;

pub use funding::FundingModel;

pub use loaders::load_csv;

pub use loaders::load_csv_str;

pub use loaders::sort_chronological;

pub use metrics::TradeOutcome;

pub use result::BacktestResult;

pub use result::EquityPoint;

pub use slippage::SlippageModel;

Modules

config

Backtest configuration + builder.

engine

The replay engine.

error

Errors specific to the backtest engine.

fees

Fee models applied to every simulated fill in the backtest engine.

funding

Perpetual-futures funding cashflows for the replay engine.

loaders

Candle series loaders for backtests.

metrics

Trade-level outcome captured during the replay loop.

result

Backtest result — aggregated metrics + the full trade ledger.

slippage

Slippage models applied between a brain’s signal and the simulated fill price.