VectorBT – An Introductory Guide

The article “VectorBT – An Introductory Guide” first appeared on AlgoTrading101 blog.

Excerpt

What is VectorBT?

VectorBT is an open-source Python library for quantitative analysis and backtesting.

Link: https://vectorbt.dev/

What is VectorBT used for?

VectorBT is used by algorithmic traders and investors to perform quantitative analysis, strategy testing, and research. It is built and optimized for performance and uses NumPy and Numba under the hood.

Why should I use VectorBT?

• VectorBT is open-source
• VectorBT is easy to use
• Is fast
• Integrates with Telegram
• Offers interactive charting via Jupyter notebooks

Why shouldn’t I use VectorBT?

• VectorBT could use more features
• It doesn’t have the best documentation
• If you want to get serious with VectorBT, you will need to acquire a proprietary VectorBT Pro membership that gives you access to the upgraded library

Is VectorBT free?

VectorBT is an open-source library and it is free to use. Do keep in mind that there is VectorBT Pro which is a successor of VectorBT and comes with advanced features and performance which is a premium product on an invite-only GitHub basis.

What are some VectorBT alternatives?

VectorBT can be replaced with other software that can be more suitable for your needs. Here are some of them:

• QuantConnect
• TradingView (Pine Script)
• Backtrader
• Interactive Brokers
• Thikscript
• QuantRocket
• HaasOnline
• cTrader
• Trality, and more

How to get started with VectorBT?

To get started with VectorBT, you will need to download the Python library via pip with the following command (you might want to have a new environment):

pip install -U vectorbt

If you want all the dependencies and features that VectorBT has, you will want to run this command:

pip install -U "vectorbt[full]"

If you are a fan of using Docker for your development, you have the option of spinning up a docker container that hosts a Jupyter Lab with VectorBT inside of it:

docker run --rm -p 8888:8888 -v "$PWD":/home/jovyan/work polakowo/vectorbt

The code block above pulls the latest polakowo/vectorbt image from Docker Hub. It then starts a container running a Jupyter Notebook server and exposes the server on host port 8888.

Visiting http://127.0.0.1:8888/?token=<token> in a browser loads JupyterLab, where the token is the secret token printed in the console.

When you’re done with using the Docker container, Docker destroys the container after the notebook server exit, but any files written to the working directory in the container remain intact in the working directory of the host.

I’ll personally use it inside of a Google Colab notebook and install it via pip.

Now that we have VectorBT ready, let us explore in the following headers what it has to offer in terms of features and performance.

How to get data with VectorBT?

To get data with VectorBT, you will need to utilize the function that connects to Yahoo Finance to download the data and provide it with the asset you wish to obtain the data. For example, let’s obtain the ETH-USD ticker closing price data:

eth_price = vbt.YFData.download('ETH-USD').get('Close')
eth_price[:5]

Date
2017-11-09 00:00:00+00:00    320.884003
2017-11-10 00:00:00+00:00    299.252991
2017-11-11 00:00:00+00:00    314.681000
2017-11-12 00:00:00+00:00    307.907990
2017-11-13 00:00:00+00:00    316.716003
Freq: D, Name: Close, dtype: float64

How to use technical indicators with VectorBT?

To use technical indicators with VectorBT, you will need to use in-built functions that host indicators such as the MA, MSTD, BBANDS, RSI, and more. For example, let’s calculate a fast and slow MA for ETH and the RSI.

fast_ma = vbt.MA.run(eth_price, 10)
slow_ma = vbt.MA.run(eth_price, 50)
rsi = vbt.RSI.run(eth_price)

How to define entries and exists with VectorBT?

To define entries and exists with VectorBT, all you need to do is to define the logic of those conditions that need to be satisfied in order to be marked as an entry or exit.

For example, let’s define the entry to be when the fast MA crosses the slow MA while the RSI is over 50 and the exit when the slow MA crosses above the fast MA and the RSI is under 50:

entries = fast_ma.ma_crossed_above(slow_ma) & rsi.rsi_above(50)
exits = slow_ma.ma_crossed_above(fast_ma) & rsi.rsi_below(50)

Note that this strategy is just an example and strategies like this almost never work in real-life.

How to perform backtesting with VectorBT?

To perform backtesting with VectorBT, you can use the Portfolio function and its modules to define the trading requirements such as the entry and exit conditions, initial cash, and more. For example, let’s implement our exits and entries from the header above and backtest them:

pf = vbt.Portfolio.from_signals(eth_price, entries, exits, init_cash=10000)
pf.total_profit()

55760.53

Let us observe the overall statistics of our simple trading strategy that is only for showcase purposes by running the pf.stats() command:

Start                          2017-11-09 00:00:00+00:00
End                            2022-11-07 00:00:00+00:00
Period                                1825 days 00:00:00
Start Value                                      10000.0
End Value                                   65760.531431
Total Return [%]                              557.605314
Benchmark Return [%]                           393.42466
Max Gross Exposure [%]                             100.0
Total Fees Paid                                      0.0
Max Drawdown [%]                               61.262033
Max Drawdown Duration                  513 days 00:00:00
Total Trades                                          18
Total Closed Trades                                   17
Total Open Trades                                      1
Open Trade PnL                                1156.40722
Win Rate [%]                                   41.176471
Best Trade [%]                                318.026171
Worst Trade [%]                               -21.805347
Avg Winning Trade [%]                          71.673003
Avg Losing Trade [%]                            -10.7351
Avg Winning Trade Duration    83 days 06:51:25.714285715
Avg Losing Trade Duration               18 days 14:24:00
Profit Factor                                   3.388864
Expectancy                                   3212.007307
Sharpe Ratio                                    0.919382
Calmar Ratio                                    0.746707
Omega Ratio                                      1.23331
Sortino Ratio                                   1.378526

Visit AlgoTrading101 blog to read the full article.