Skip to content

margin_leverage.md

Latest commit

 

History

History
79 lines (58 loc) · 2.08 KB

margin_leverage.md

File metadata and controls

79 lines (58 loc) · 2.08 KB
Raw

Margin And Leverage

QuantBT follows Binance-style futures accounting for notional buying power.

buying_power = current_equity * leverage
initial_margin_required = abs(order_notional) / leverage
maintenance_margin = abs(position_notional) * maintenance_ratio

Important rule:

leverage does not multiply alloc_per_trade.

If the account has initial_capital=10_000 and leverage=10, the initial buying power is about 100_000. A strategy can still choose to allocate 50_000; leverage only controls required margin and rejection/liquidation thresholds.

bt = BacktestEngine(
    Datetime=dt,
    Position=signal,
    Close=close,
    initial_capital=10_000,
    leverage=10,
    alloc_per_trade=50_000,
    hedge_type="signal_notional",
)

Fees And Slippage

Legacy BacktestEngine expects fee as round-trip and halves it internally to one-way. V2 backends expect fee_rate as one-way.

Legacy slippage is a fraction:

slippage=0.0001  # 1 bp

V2 ExecutionConfig.slippage_bps is in basis points:

ExecutionConfig(slippage_bps=1.0)

Liquidation

Intrabar liquidation uses high/low worst-case prices:

  • long positions are checked against Low;
  • short positions are checked against High;
  • if High and Low are not provided, close is used as fallback.

For robust crypto intraday backtests, always pass High and Low.

Quantity Constraints

Crypto fractional trading is controlled by exchange quantity filters, not by contract_size.

Use:

QuantBTEndpoint.signal_notional(
    contract_size=1.0,   # PnL/notional multiplier for linear USDT contracts
    qty_step=0.001,      # venue lot increment
    min_qty=0.001,
    min_notional=10.0,
)

qty_step, lot_size, and the compatibility alias slot_size describe the same venue quantity increment. QuantBT rounds target/order quantity down before execution and zeroes orders below min_qty or min_notional. The same shared constraint layer is used by legacy single-symbol, native vectorized, native event/orders, native portfolio, and Nautilus validation routes.