portfolio

Classes

Portfolio

Portfolio class for managing trading positions and cash flow.

Module Contents

class portfolio.Portfolio(cfg, symbols, position_manager=None, time_step=(TimeFrameUnit.Day, 1))

Portfolio class for managing trading positions and cash flow. Very much Stateful TODO there is not a clean seperation between portfolio logic and the underlying broker api and state. It is good to have the portfolio state mirrored internally via the df, but self.cash, self.initial_cash, etc… are not reflections of the underlying broker api, (ALPACA, LOCAL, REMOTE). ? position manager acts as the broker api here, interacting with the positions, trades, and orders, should we offload many of the portfolio attributes to the position manager? (and rename it)

Parameters:

  • cfg (trading.cli.alg.config.PortfolioConfig)

  • symbols (list[str])

  • position_manager (trading.src.portfolio.position.PositionManager | None)

  • time_step (tuple[alpaca.data.timeframe.TimeFrameUnit, int])

cfg
vbt_pf: vectorbt.Portfolio | None = None
position_manager = None
time_step
persistent_df: pandas.DataFrame | None = None
as_vbt_pf(df=None)

Update the portfolio with new data.

Parameters:

df (pandas.DataFrame | None)

Return type:

vectorbt.Portfolio

state()

Get the current state of the portfolio. :returns: [internal_cash, positions]. :rtype: pd.Series

Return type:

numpy.ndarray

enforce_trade_rules(df, prices, trade_mode=TradeMode.CONTINUOUS)

Enforce trade rules on the actions. From trade sizes to actual trade sizes based on configuration and current portfolio state. @TODO clean up and make more efficient

Parameters:
Return type:

numpy.ndarray

scale_actions(df, prices, trade_mode=TradeMode.CONTINUOUS)

Scale the actions to the maximum position size. From signal strength to a desired trade size @TODO clean up and make more efficient

Parameters:
Return type:

numpy.ndarray

update_position_batch(df)

Update positions for a batch of tickers at a given timestamp. @TODO clean up and make more efficient

Parameters:

df (pandas.DataFrame)

Return type:

tuple[float, list[alpaca.trading.requests.MarketOrderRequest] | None]

step(df, normalized_actions=False)

Scale the actions and take a step in the portfolio environment. 1. scale actions to trade size 2. enforce trade limits 3. update positions @TODO clean up and make more efficient

Parameters:

  • df (pandas.DataFrame)

  • normalized_actions (bool)

Return type:

dict

classmethod load(cfg, file_path)

Load backtesting results from a JSON file.

Parameters:
Return type:

Portfolio

save(file_path, df=None)
Parameters:

  • file_path (str)

  • df (pandas.DataFrame | None)

save_plots(backtest_dir, tickers=None)
Parameters:

  • backtest_dir (pathlib.Path)

  • tickers (list[str] | None)

reset()

Reset the portfolio to an empty state.

set_vbt(pf)

Set the portfolio object with backtest results.

Parameters:

pf (vectorbt.Portfolio)

_get_plots()

Get the plots for the portfolio.

Return type:

dict[str, vectorbt._typing.BaseFigure]

plot()

Plot the results of the backtest.

stats()

Return the statistics of the backtest.

orders()

Return the orders of the backtest.

trades()

Return the trades of the backtest.

__repr__()

String representation of the Portfolio.

Return type:

str

get_positions()

Return the current positions in the portfolio.

Return type:

trading.src.portfolio.position.PositionManager

analysis(analysis_config, df=None)
Parameters:

df (pandas.DataFrame | None)