/vectorbt-expert

VectorBT Simulation Modes

1. from_signals (Signal-Based) - Most Common

Entry/exit boolean arrays. VectorBT processes signals sequentially - after entry, waits for exit before next entry (unless accumulate=True).

import vectorbt as vbt
import numpy as np

pf = vbt.Portfolio.from_signals(
    close,                      # Price series (required)
    entries,                    # Boolean Series - True = buy signal
    exits,                      # Boolean Series - True = sell signal
    init_cash=1_000_000,        # Starting capital
    fees=0.00111,               # Indian delivery equity (see indian-market-costs.md)
    fixed_fees=20,              # Rs 20 per order
    slippage=0.0005,            # 0.05% slippage
    size=0.75,                  # Position size
    size_type="percent",        # How to interpret size
    direction="longonly",       # longonly, shortonly, both
    freq="1D",                  # Data frequency
    min_size=1,                 # Minimum order size
    size_granularity=1,         # Round to whole shares
    sl_stop=0.05,               # 5% stop loss (optional)
    tp_stop=0.10,               # 10% take profit (optional)
    accumulate=False,           # True = allow pyramiding
)

2. from_orders (Order-Based) - Direct Orders

Provide explicit order arrays. Best for portfolio rebalancing and target-weight strategies.

pf = vbt.Portfolio.from_orders(
    close=close,
    size=0.15,                  # Target 15% allocation
    size_type='targetpercent',  # Rebalances to target weight
    group_by=True,              # Group columns as one portfolio
    cash_sharing=True,          # Share cash across assets
    fees=0.00111, fixed_fees=20,
    init_cash=1_000_000,
    freq='1D',
    min_size=1,
    size_granularity=1,
)

3. from_order_func (Custom Callback) - Most Powerful

Numba-compiled functions called at each bar with full portfolio state access. Use for complex logic (e.g., dynamic position sizing based on portfolio state, multi-asset coordination). flexible=True allows multiple orders per symbol per bar.

4. from_holding (Buy-and-Hold Benchmark)

pf_benchmark = vbt.Portfolio.from_holding(close, init_cash=1_000_000, fees=0.00111, freq="1D")

Direction

Key Parameters Reference

Random Signal Baseline

pf_random = vbt.Portfolio.from_random_signals(close, n=50, init_cash=1_000_000, fees=0.00111, freq="1D")

Save/Load Portfolio

pf.save("my_backtest.pkl")
pf_loaded = vbt.Portfolio.load("my_backtest.pkl")

When to Use Which Mode