VectorBT

Common errors

• AttributeError: 'Function' object has no attribute '_Function__info' when using vbt.IndicatorFactory.from_talib()

  cause This error has been reported in older versions when attempting to dynamically create indicators from TA-Lib functions using `IndicatorFactory`. It indicates an incompatibility or internal API change.
  fix Ensure `vectorbt` and `TA-Lib` are up-to-date. If the issue persists, consider manually defining indicators or using `vectorbt`'s built-in indicator wrappers if available, instead of `IndicatorFactory.from_talib` for the specific indicator.

• vectorbt heatmap displays wrong values or not all values when using a Pandas Series directly.

  cause The heatmap plotting functionality might have unexpected behavior or require specific data structures (like DataFrames) for correct visualization, especially when dealing with multi-parameter results which are typically structured as DataFrames.
  fix Transform your Pandas Series into a Pandas DataFrame, even if it's a single column, before passing it to the `vbt.heatmap` method. This ensures proper indexing and value mapping.

• Backtesting runs very slowly with large datasets or many strategy permutations.

  cause While VectorBT is designed for speed, users might inadvertently write non-vectorized Python loops or perform operations that prevent Numba from optimizing. Incorrect data structuring (e.g., not leveraging MultiIndex for parameter sweeps) can also lead to performance bottlenecks.
  fix Ensure your strategy logic is vectorized as much as possible, leveraging pandas and NumPy operations. For custom complex logic, use Numba's `njit` decorator. Structure hyperparameter sweeps into MultiIndex DataFrames for `vectorbt` to process efficiently in a vectorized manner.

Warnings

• gotcha Older versions of vectorbt (prior to 0.28.5) had compatibility issues with pandas 2.x, specifically an `AttributeError` when using `Styler.render()`. While fixed in 0.28.5, users on older `vectorbt` versions with pandas 2.x may encounter this.
  Fix: Upgrade to vectorbt version 0.28.5 or newer: `pip install --upgrade vectorbt`.
• gotcha Early documentation and tutorials (e.g., from March 2022) incorrectly stated that vectorbt did not support Python 3.10. The library now officially supports and requires Python >=3.10 as per PyPI metadata. Relying on outdated information may lead to incorrect assumptions about compatibility.
  Fix: Ensure you are using Python 3.10 or newer and `vectorbt` version 0.28.0 or later. Always check the official PyPI or GitHub for current Python requirements.
• gotcha When combining `vbt` accessor methods with other arrays or pandas objects, ensure the `vbt` accessor operand is on the left-hand side (e.g., `df.vbt * 2` rather than `2 * df.vbt`) for consistent behavior and performance benefits.
  Fix: Always place the `.vbt` accessor on the left side of binary operations.

Install

• pip install vectorbt Core library
• pip install "vectorbt[full]" With optional dependencies (e.g., TA-Lib, Plotly)

Imports

• vbt

  import vectorbt as vbt

  Standard alias for the library.
• YFData

  vbt.YFData

  Used for downloading historical data from Yahoo Finance.
• Portfolio

  vbt.Portfolio

  Core class for creating and analyzing trading portfolios from signals or orders.
• MA

  vbt.MA

  Provides moving average indicators.
• RSI

  vbt.RSI

  Provides Relative Strength Index indicator.

Quickstart

This quickstart demonstrates how to download historical price data, simulate a basic buy-and-hold strategy, and then implement a simple moving average crossover strategy with entry/exit signals using `vectorbt`.

import vectorbt as vbt
import pandas as pd

# Download historical daily price data for Bitcoin (BTC-USD)
# Use a specific date range for reproducibility
price = vbt.YFData.download(
    'BTC-USD',
    start=pd.Timestamp('2020-01-01', tz='UTC'),
    end=pd.Timestamp('2021-01-01', tz='UTC')
).get('Close')

# Simulate a simple buy-and-hold portfolio with $100 initial cash
pf = vbt.Portfolio.from_holding(price, init_cash=100)

# Calculate total profit
profit = pf.total_profit()
print(f"Total profit from buy-and-hold: ${profit:.2f}")

# Calculate a simple moving average crossover strategy
fast_ma = vbt.MA.run(price, 10)
slow_ma = vbt.MA.run(price, 50)

# Generate entry and exit signals
entries = fast_ma.ma_crossed_above(slow_ma)
exits = slow_ma.ma_crossed_above(fast_ma)

# Backtest the MA crossover strategy
pf_ma_crossover = vbt.Portfolio.from_signals(
    price, entries, exits, init_cash=100, fees=0.001
)

# Print strategy statistics
print("\nMA Crossover Strategy Statistics:")
print(pf_ma_crossover.stats())