mirror of
https://github.com/OpenBB-finance/OpenBB.git
synced 2026-11-07 09:52:13 +00:00
234 lines
6.8 KiB
Python
234 lines
6.8 KiB
Python
"""Yahoo Finance Comparison Model"""
|
|
__docformat__ = "numpy"
|
|
|
|
import logging
|
|
from datetime import datetime, timedelta
|
|
from typing import List, Optional, Tuple
|
|
import warnings
|
|
|
|
import numpy as np
|
|
import pandas as pd
|
|
import yfinance as yf
|
|
from sklearn.manifold import TSNE
|
|
from sklearn.preprocessing import normalize
|
|
|
|
from openbb_terminal.decorators import log_start_end
|
|
from openbb_terminal.rich_config import console
|
|
|
|
logger = logging.getLogger(__name__)
|
|
|
|
d_candle_types = {
|
|
"o": "Open",
|
|
"h": "High",
|
|
"l": "Low",
|
|
"c": "Close",
|
|
"a": "Adj Close",
|
|
"v": "Volume",
|
|
}
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_historical(
|
|
similar: List[str],
|
|
start_date: Optional[str] = None,
|
|
candle_type: str = "a",
|
|
) -> pd.DataFrame:
|
|
"""Get historical prices for all comparison stocks
|
|
|
|
Parameters
|
|
----------
|
|
similar: List[str]
|
|
List of similar tickers.
|
|
Comparable companies can be accessed through
|
|
finnhub_peers(), finviz_peers(), polygon_peers().
|
|
start_date: Optional[str], optional
|
|
Initial date (e.g., 2021-10-01). Defaults to 1 year back
|
|
candle_type: str, optional
|
|
Candle variable to compare, by default "a" for Adjusted Close. Possible values are: o, h, l, c, a, v, r
|
|
|
|
Returns
|
|
-------
|
|
pd.DataFrame
|
|
Dataframe containing candle type variable for each ticker
|
|
"""
|
|
|
|
if start_date is None:
|
|
start_date = (datetime.now() - timedelta(days=366)).strftime("%Y-%m-%d")
|
|
|
|
candle_type = candle_type.lower()
|
|
use_returns = False
|
|
if candle_type == "r":
|
|
# Calculate returns based off of adjusted close
|
|
use_returns = True
|
|
candle_type = "a"
|
|
|
|
# To avoid having to recursively append, just do a single yfinance call. This will give dataframe
|
|
# where all tickers are columns.
|
|
similar_tickers_dataframe = yf.download(
|
|
similar, start=start_date, progress=False, threads=False
|
|
)[d_candle_types[candle_type]]
|
|
|
|
returnable = (
|
|
similar_tickers_dataframe
|
|
if similar_tickers_dataframe.empty
|
|
else similar_tickers_dataframe[similar]
|
|
)
|
|
|
|
if use_returns:
|
|
# To calculate the period to period return,
|
|
# shift the dataframe by one row, then divide it into
|
|
# the other, then subtract 1 to get a percentage, which is the return.
|
|
shifted = returnable.shift(1)[1:]
|
|
returnable = returnable.div(shifted) - 1
|
|
|
|
df_similar = returnable[similar]
|
|
|
|
if np.any(df_similar.isna()):
|
|
nan_tickers = df_similar.columns[df_similar.isna().sum() >= 1].to_list()
|
|
console.print(
|
|
f"NaN values found in: {', '.join(nan_tickers)}. Backfilling data"
|
|
)
|
|
df_similar = df_similar.fillna(method="bfill")
|
|
|
|
df_similar = df_similar.dropna(axis=1, how="all")
|
|
|
|
return df_similar
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_correlation(
|
|
similar: List[str],
|
|
start_date: Optional[str] = None,
|
|
candle_type: str = "a",
|
|
) -> Tuple[pd.DataFrame, pd.DataFrame]:
|
|
"""
|
|
Get historical price correlation. [Source: Yahoo Finance]
|
|
|
|
Parameters
|
|
----------
|
|
similar : List[str]
|
|
List of similar tickers.
|
|
Comparable companies can be accessed through
|
|
finnhub_peers(), finviz_peers(), polygon_peers().
|
|
start_date : Optional[str], optional
|
|
Initial date (e.g., 2021-10-01). Defaults to 1 year back
|
|
candle_type : str, optional
|
|
OHLCA column to use for candles or R for returns, by default "a" for Adjusted Close
|
|
|
|
Returns
|
|
-------
|
|
Tuple[pd.DataFrame, pd.DataFrame]
|
|
Dataframe with correlation matrix, Dataframe with historical prices for all comparison stocks
|
|
"""
|
|
|
|
if start_date is None:
|
|
start_date = (datetime.now() - timedelta(days=366)).strftime("%Y-%m-%d")
|
|
|
|
df_similar = get_historical(similar, start_date, candle_type)
|
|
|
|
correlations = df_similar.corr()
|
|
|
|
return correlations, df_similar
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_volume(
|
|
similar: List[str],
|
|
start_date: Optional[str] = None,
|
|
) -> pd.DataFrame:
|
|
"""Get stock volume. [Source: Yahoo Finance]
|
|
|
|
Parameters
|
|
----------
|
|
similar : List[str]
|
|
List of similar tickers.
|
|
Comparable companies can be accessed through
|
|
finnhub_peers(), finviz_peers(), polygon_peers().
|
|
start_date : Optional[str], optional
|
|
Initial date (e.g., 2021-10-01). Defaults to 1 year back
|
|
|
|
Returns
|
|
-------
|
|
pd.DataFrame
|
|
Dataframe with volume for stock
|
|
"""
|
|
|
|
if start_date is None:
|
|
start_date = (datetime.now() - timedelta(days=366)).strftime("%Y-%m-%d")
|
|
|
|
df_similar = get_historical(similar, start_date, "v")
|
|
df_similar = df_similar[df_similar.columns]
|
|
return df_similar
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_1y_sp500() -> pd.DataFrame:
|
|
"""
|
|
Gets the last year of Adj Close prices for all current SP 500 stocks.
|
|
They are scraped daily using yfinance at https://github.com/jmaslek/daily_sp_500
|
|
|
|
Returns
|
|
-------
|
|
pd.DataFrame
|
|
DataFrame containing last 1 year of closes for all SP500 stocks.
|
|
"""
|
|
return pd.read_csv(
|
|
"https://raw.githubusercontent.com/jmaslek/daily_sp_500/main/SP500_prices_1yr.csv",
|
|
index_col=0,
|
|
)
|
|
|
|
|
|
# pylint:disable=E1137,E1101
|
|
|
|
|
|
@log_start_end(log=logger)
|
|
def get_sp500_comps_tsne(
|
|
symbol: str,
|
|
lr: int = 200,
|
|
) -> pd.DataFrame:
|
|
"""
|
|
Runs TSNE on SP500 tickers (along with ticker if not in SP500).
|
|
TSNE is a method of visualing higher dimensional data
|
|
https://scikit-learn.org/stable/modules/generated/sklearn.manifold.TSNE.html
|
|
Note that the TSNE numbers are meaningless and will be arbitrary if run again.
|
|
|
|
Parameters
|
|
----------
|
|
symbol: str
|
|
Ticker to get comparisons to
|
|
lr: int
|
|
Learning rate for TSNE
|
|
|
|
Returns
|
|
-------
|
|
pd.DataFrame
|
|
Dataframe of tickers closest to selected ticker
|
|
"""
|
|
|
|
# Adding the type makes pylint stop yelling
|
|
close_vals: pd.DataFrame = get_1y_sp500()
|
|
if symbol not in close_vals.columns:
|
|
df_symbol = yf.download(symbol, start=close_vals.index[0], progress=False)[
|
|
"Adj Close"
|
|
].to_frame()
|
|
df_symbol.columns = [symbol]
|
|
df_symbol.index = df_symbol.index.astype(str)
|
|
close_vals = close_vals.join(df_symbol)
|
|
|
|
close_vals = close_vals.fillna(method="bfill")
|
|
rets = close_vals.pct_change()[1:].T
|
|
# Future warning from sklearn. Think 1.2 will stop printing it
|
|
warnings.filterwarnings("ignore", category=FutureWarning)
|
|
model = TSNE(learning_rate=lr, init="pca")
|
|
tsne_features = model.fit_transform(normalize(rets))
|
|
warnings.resetwarnings()
|
|
xs = tsne_features[:, 0]
|
|
ys = tsne_features[:, 1]
|
|
|
|
data = pd.DataFrame({"X": xs, "Y": ys}, index=rets.index)
|
|
x0, y0 = data.loc[symbol]
|
|
data["dist"] = (data.X - x0) ** 2 + (data.Y - y0) ** 2
|
|
data = data.sort_values(by="dist")
|
|
|
|
return data
|