Struct TradingEnv 

pub struct TradingEnv { /* private fields */ }

A steppable, leak-free trading environment over a frozen dataset. The caller drives it: reset returns the first observation, then each step applies the supplied decision and advances one bar.

Implementations

impl TradingEnv

pub fn new(data: Dataset, window: Window, costs: CostModel, seed: u64) -> Self

Build an environment that steps window over data with seeded execution noise and the given cost model.

pub fn reset(&mut self) -> MarketObservation

Reset to the start of the window and return the first point-in-time observation. Re-seeds the book, so it is safe to call repeatedly.

pub fn step(&mut self, decision: Decision) -> StepResult

Apply decision at the current bar and advance one step. The returned reward is the bar’s portfolio return; done is set once the window is exhausted (further calls re-apply the final bar harmlessly).

pub fn clone_state(&self) -> EnvState

Snapshot the mutable sim state (time cursor + book) in O(1) — no replay. The returned EnvState is a value: clone it, stash it, restore it later with restore_state to fork the env from this exact point (e.g. tree search / what-if rollouts over the same frozen data). The immutable config (data, window, costs, seed) is not snapshotted — it never changes — so a snapshot is cheap and a restore is exact.

pub fn restore_state(&mut self, state: EnvState)

Restore a snapshot taken by clone_state: the env resumes producing the exact byte-identical trajectory it would have from the snapshot point, including seeded execution noise (the RNG cursor is part of the snapshot).