Execution modes

Three modes, three levels of execution realism. Pay only for the realism you need.

Minute (1× cost)

The default mode. Strategy receives 1-minute OHLCV bars. Orders submitted on bar t fill at bar t+1's open plus a constant slippage (default 2 bps for crypto, lower for FX/ETFs).

When to use: rapid iteration, parameter exploration, longer horizons (years). Minute mode is deliberately optimistic about execution — your real fills will be worse, but you can iterate without paying for that realism while exploring.

Hybrid (3× cost)

Strategy still receives 1m bars but order fills are computed within the OHLC range with adverse selection: buys lean toward bar.high, sells lean toward bar.low. Coefficient is calibrated against actual tick fills (recalibrated quarterly).

When to use: once your strategy looks promising in minute mode and you want execution slippage baked in. Especially valuable for higher-frequency strategies where minute-mode optimism overstates returns meaningfully.

Tick (10× cost)

Strategy still receives 1m bars (callback unchanged) but fills are routed to the actual tick stream. Buy orders fill at the first tick where the ask side gets hit; sell orders at the first bid hit. Closer to real execution than any heuristic.

Available for: BTC, ETH, SOL crypto only. FX tick mode lands as a follow-up after we calibrate against the crypto top 3.

When to use: you're testing high-frequency, latency-sensitive, or microstructure-aware strategies. For longer-horizon trend-following, tick mode is overkill — hybrid is enough.

Comparing modes on the same strategy

The pattern: run minute first, then re-run with improve_strategy using use_hybrid_mode or use_tick_mode:

# 1. Initial submit
> Run sma-cross on BTCUSDT for 2024 in minute mode

# 2. Validate slippage realism
> Re-run that backtest in hybrid mode using improve_strategy

# 3. Tick-validate (Quant tier)
> Re-run again in tick mode using improve_strategy

Cost example