An opinionated trading strategy backtesting framework for python.

Generally speaking, mega_money_millions expects Pandas Dataframes of Price information indexed by either time or multi-indexed on time AND ticker.

There is also a fair amount of unit test coverage so working examples can be found in tests/**/test_*.py.

mega_money_millions.indicators.singleasset includes functions that add the given indicator to a Dataframe indexed by time.

Add a Simple Moving Average to the given prices DataFrame for a single asset. prices will be mutated and the moving average will be found in a new column named f'{prefix}{window}. Eg: add_sma(prices, 45) will add a new column SMA45 to prices.

Add a Rate of Change to the given prices DataFrame for a single asset. prices will be mutated and the rate of change will be found in a new column named f'{prefix}{window}. Eg: add_roc(prices, 20) will add a new column ROC20 to prices.

Add two boolean fields to prices based on when on_stat1 crosses over on_stat2 and vice versa. For example:

# Adds two columns to `prices`: 'SMACrossUp' and 'SMACrossDown'.  On the bar where the SMA45 crosses over
# (becomes greater than) the SMA90, 'SMACrossUp' will be true.  On the bar when the SMA90 crosses over the SMA45,
# 'SMACrossDown' will be true.  Note that the value will only be true _on the bar_ when the crossover occurs; it
# will become false again on the next bar.  This is meant to track when the change occurs, not every tick when one
# bar is greater than the other.
add_crossover(prices, 'SMA45', 'SMA90', 'SMACrossUp', 'SMACrossDown')

Takes two single asset price DataFrames and combines them into a new DataFrame with a multi-index on time and ticker. Expects prices1 and prices2 to each have a time and ticker column.

Includes Exchange subclasses that represent places where securities are traded. These are responsible for things like estimating trading fees. The only current option is CoinBase().

Defines an object that helps simulate trades.

A new Portfolio can be initialized with Portfolio(exchange, initial_cash). exchange is an Exchange instance as defined above. initial_cash is an integer or float. It exposes the following methods:

Buy either quantity shares at the given price or calculate the number of shares to purchase based on percentage_of_cash. Exchange/trading fees will be taken into account based on the Exchange object (eg: Coinbase()). percentage_of_cash should be between 1 and 100 (not 0 and 1).

Sell either quantity shares or calculte the number of shares to sell based on percentage_of_shares. Exchange/trading fees will be taken into account based on the Exchange object. percentage_of_shares should be between 1 and 100 (not 0 and 1).

Returns the current cash balance of the portfolio based on initial_cash and trade history.

Returns the number of shares currently owned of the given ticker.

A property that returns a DataFrame that represents a chronological log of buys and sells.

Exposes a function intended for use in backtesting trading strategies.

Accepts a multi-asset prices (multi-indexed on time and ticker), an initial cash amount, and an on_tick callback. The on_tick method should accept the current time, the Portfolio object, and the prices_for_date for ALL assets on the given date. Note that the tick frequency is currently hardcoded to be days.

https://github.com/davebenvenuti/mega-money-millions/blob/main/tests/test_backtester.py gives a good example of how one might backtest a strategy to move 100% of a portfolio's cash into BTC when the SMA45 crosses the SMA90, and sell the entire position then the SMA90 once again tops the SMA45.

pipenv install [package]

Might also have to add it to pyproject.toml manually. lol.

bin/shell

Which actually does this:

pipenv shell # vscode might do this automatically? maybe?

bin/lint
bin/lint-autocorrect

Linter and formatter are separate and both configured in pyproject.toml.

• https://packaging.python.org/en/latest/guides/writing-pyproject-toml/
• https://packaging.python.org/en/latest/tutorials/packaging-projects/
• https://dev.to/vorsprung/vscode-pipenv-python-34pf used to get pipenv and vscode working together
• https://pipenv.pypa.io/en/latest/