mirror of
https://github.com/OpenBB-finance/OpenBB.git
synced 2026-10-28 04:52:07 +00:00
cd7d602b8f
* refactor quantitative for v5 * add factors endpoint and add a few charts --------- Co-authored-by: deeleeramone <>
90 lines
2.8 KiB
Python
90 lines
2.8 KiB
Python
"""Shared fixtures for the quantitative test suite."""
|
|
|
|
import numpy as np
|
|
import pandas as pd
|
|
import pytest
|
|
from openbb_core.app.utils import df_to_basemodel
|
|
|
|
SEED = 42
|
|
N_ROWS = 250
|
|
FACTOR_N_ROWS = 800
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def prices_df() -> pd.DataFrame:
|
|
"""Build a deterministic OHLCV price series with a date column."""
|
|
rng = np.random.default_rng(SEED)
|
|
t = np.arange(N_ROWS)
|
|
close = 100 + 0.1 * t + rng.normal(0, 1, N_ROWS).cumsum() * 0.5
|
|
dates = pd.date_range("2022-01-03", periods=N_ROWS, freq="B")
|
|
|
|
return pd.DataFrame(
|
|
{
|
|
"date": dates,
|
|
"open": close + rng.normal(0, 0.3, N_ROWS),
|
|
"high": close + np.abs(rng.normal(0.8, 0.3, N_ROWS)),
|
|
"low": close - np.abs(rng.normal(0.8, 0.3, N_ROWS)),
|
|
"close": close,
|
|
"volume": (1e6 + rng.normal(0, 5e4, N_ROWS)).astype(float),
|
|
}
|
|
)
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def prices_data(prices_df) -> list:
|
|
"""Expose the price DataFrame as a ``list[Data]``."""
|
|
return df_to_basemodel(prices_df)
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def factor_dates() -> pd.DatetimeIndex:
|
|
"""Daily business-day index long enough for rolling 252-day windows."""
|
|
return pd.date_range("2018-01-01", periods=FACTOR_N_ROWS, freq="B")
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def factor_matrix_df(factor_dates) -> pd.DataFrame:
|
|
"""Two synthetic factors plus a constant risk-free column.
|
|
|
|
Used together with ``target_returns_df`` so the true betas are known
|
|
(`y = 0.7*f1 - 0.3*f2 + rf + noise`).
|
|
"""
|
|
rng = np.random.default_rng(SEED)
|
|
n = len(factor_dates)
|
|
return pd.DataFrame(
|
|
{
|
|
"date": factor_dates,
|
|
"f1": rng.normal(0, 0.01, n),
|
|
"f2": rng.normal(0, 0.01, n),
|
|
# rf varies slightly so statsmodels keeps it as its own regressor
|
|
# instead of folding it into the intercept term.
|
|
"rf": 0.0001 + rng.normal(0, 1e-6, n),
|
|
}
|
|
)
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def target_returns_df(factor_dates, factor_matrix_df) -> pd.DataFrame:
|
|
"""Target series whose returns are a known linear combo of the factors."""
|
|
rng = np.random.default_rng(SEED + 1)
|
|
n = len(factor_dates)
|
|
y = (
|
|
0.7 * factor_matrix_df["f1"].to_numpy()
|
|
- 0.3 * factor_matrix_df["f2"].to_numpy()
|
|
+ factor_matrix_df["rf"].to_numpy()
|
|
+ rng.normal(0, 0.002, n)
|
|
)
|
|
return pd.DataFrame({"date": factor_dates, "close": y})
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def factor_matrix_data(factor_matrix_df) -> list:
|
|
"""Factor matrix as ``list[Data]``."""
|
|
return df_to_basemodel(factor_matrix_df)
|
|
|
|
|
|
@pytest.fixture(scope="session")
|
|
def target_returns_data(target_returns_df) -> list:
|
|
"""Target returns as ``list[Data]``."""
|
|
return df_to_basemodel(target_returns_df)
|