qis.bootstrap_price_fundamental_data

qis.bootstrap_price_fundamental_data(price_datas, fundamental_datas, bootstrap_type=BootstrapType.STATIONARY, bootstrap_output=BootstrapOutput.DF_TO_LIST_ARRAYS, num_samples=10, index_length=1000, block_size=30, min_block_size=1, is_log_returns=False, seed=1, is_price_weighted_fundamentals=False, init_to_end=True, is_positive=True)[source]

resample price panels and AR(1) fundamentals jointly with one shared index array.

One index array is drawn over the len(prices.index) - 1 return rows of the first price panel and passed to bootstrap_price_data() for every price panel and to bootstrap_ar_process() for every fundamental panel, so the return and the AR innovation drawn at the same index row move together.

The two kinds of path start differently and are offset by one step. A price path starts at its anchor (the last positive price with init_to_end=True, the first row otherwise): row 0 is the anchor and row t has applied the returns drawn at index rows 1..t. A fundamental path starts from the full-sample mean of its data, which is not part of the output: row t has applied the innovations drawn at index rows 0..t. With is_price_weighted_fundamentals=True each fundamental path is multiplied element by element by the matching path of the first price panel, row by row and column by column.

Parameters:

  • price_datas (Dict[str, Series | DataFrame]) – price panels keyed by name; the first one sets the index array and must be aligned with every fundamental panel. Every panel must have as many rows as the first

  • fundamental_datas (Dict[str, Series | DataFrame]) – fundamental panels keyed by name, with the index, and for DataFrames the columns, of the first price panel

  • bootstrap_type (BootstrapType) – resampling scheme; see BootstrapType

  • bootstrap_output (BootstrapOutput) – shape of the results; see BootstrapOutput

  • num_samples (int) – number of independent draws

  • index_length (int) – length of each path

  • block_size (int) – mean block length for STATIONARY, exact length for FIXED_BLOCK

  • min_block_size (int) – floor on the drawn block length under STATIONARY

  • is_log_returns (bool) – resample log returns rather than arithmetic ones

  • seed (int) – seed for the numba random state

  • is_price_weighted_fundamentals (bool) – multiply each fundamental path by the matching path of the first price panel

  • init_to_end (bool) – anchor of the price paths, forwarded to bootstrap_price_data()

  • is_positive (bool) – positivity rule of the fundamental paths, forwarded to bootstrap_ar_process()

Returns:

Tuple of (price paths, fundamental paths), each a dict keyed like its input

Raises:

ValueError – if a fundamental panel and the first price panel are of different types, if a price panel has fewer rows than the first, or if gaps leave a fundamental panel fewer residual rows than the shared index array reaches

Return type:

Tuple[Dict[str, DataFrame | List], Dict[str, DataFrame | List]]