Execution Model

The engine simulates realistic execution:

Latency: configurable delay between order submission and fill
Slippage: basis‑point or custom models
Commission: per‑share or percentage models
Order types: market, limit, stop, stop‑limit
Time in force: GTC, day, IOC, and FOK handling in the backtest loop
Liquidity participation: fills are capped by execution_settings.max_volume_participation, so oversized orders can partially fill instead of teleporting through the bar
Lifecycle events: backtest results now retain submitted / filled / canceled / rejected / expired order events for auditability

The simulator processes events in time order across symbols to avoid look‑ahead bias.

Order lifecycle slice

GlowBack now applies a first execution-realism slice in the engine itself:

orders are marked submitted when strategies place them
GTC orders can remain open after a partial fill
IOC orders cancel any remainder immediately after a partial fill
FOK orders cancel when the full requested size cannot be filled inside the configured participation limit
day orders expire when their conditions are not met on the execution bar or when only a partial slice is available

This keeps the current engine deterministic while making order outcomes visible to Python and API consumers.

GlowBack supports two fee models via FeeModel:

Model	Use Case	Fields
PerShare	Equities, commodities	`per_share`, `percentage`, `minimum`
MakerTaker	Crypto, FX	`maker_fee_pct`, `taker_fee_pct`

Use ExecutionConfig::for_asset_class() for sensible defaults per asset class.

Market hours are asset-class-aware via MarketHours::for_asset_class():