StockAnalysisAgent / src /technical_analysis.py
OnurKerimoglu's picture
use the df_hist for FetchForecast
067b5d0 unverified
import logging
import os
import dotenv
import matplotlib.pyplot as plt
import matplotlib.dates as mdates
from matplotlib.axes import Axes
import numpy as np
import pandas as pd
from ta.volume import volume_weighted_average_price
from ta.momentum import RSIIndicator, StochasticOscillator
from ta.trend import MACD
class TechnicalAnalysis():
def __init__(
self,
ticker: str,
df_hist: pd.DataFrame,
df_past=None,
df_fcst=None,
plot_ta:bool=True,
savefig:bool=False,
debug=False):
# input arguments
"""
Initialize TechnicalAnalysis object.
Args:
ticker : str
stock ticker to analyze
df_hist: pd.DataFrame
historical price data for ticker
df_past: pd.DataFrame, optional, default: None
Closeing price of the ticker for the past few days
df_fcst: pd.DataFrame, optional, default: None
Forecasted closing price and relative returns nextf few days
plot_ta : bool, optional, default: True
whether to generate plots of technical analysis metrics. Plot will be created under plots/{ticker}.png
debug : bool, optional, default: False
whether run in debug mode, so that logging should be produced at debug level
"""
# set up logging
if debug:
self.logger_level = logging.DEBUG
else:
self.logger_level = logging.INFO
self.logger = logging.getLogger(__name__)
logging.basicConfig(level=self.logger_level) # filename='TechnicalAnalysis.log',
# input arguments
self.ticker = ticker
self.df_hist = df_hist
self.df_past = df_past
self.df_fcst = df_fcst
self.plot_ta = plot_ta
self.savefig = savefig
# done initializing
self.logger.info(f'Initialized TechnicalAnalysis object for ticker: {ticker}')
def run(
self
) -> None:
"""
Main entry point for the TechnicalAnalysis object.
This method:
- computes the technical analysis metrics
- plots the price and TA metrics.
"""
df = self.df_hist
# add the features based on technical analysis
if df.shape[0] > 0:
df = self.tech_analysis(df)
# Merge with forecast data
df_merged = self.merge_hist_with_forecast(df, self.df_past, self.df_fcst)
# plot the results
if self.plot_ta:
os.makedirs('plots', exist_ok=True)
fig = self.plot_stock_metrics(
df_merged,
datasets={
'Volume': ['Volume'],
'Indices': ['RSI', 'StochOsc'],
'Trend': ['MACD', 'MACDsig', 'MACDdif'],
'Prices': ['Close', 'VWAP'] # 'High','Low',
},
savefig=self.savefig
)
else:
fig = None
else:
if self.plot_ta:
fig = self.get_fetcherror_fig(message='failed fetching data')
else:
fig = None
return df, fig
def tech_analysis(
self,
df: pd.DataFrame
) -> pd.DataFrame:
"""
Calculates technical analysis indicators for the fetched stock price data.
This method takes the fetched stock price data and calculates several
technical analysis indicators. The following indicators are calculated:
- Additional Price Indicators:
- Volume-Weighted Average Price (VWAP)
- Momentum Indicators:
- Relative Strength Index (RSI)
- Stochastic Oscillator
- Trend Indicators:
- Moving Average Convergence Divergence (MACD)
The calculated indicators are added to the DataFrame as new columns.
Args:
df: pd.DataFrame
The DataFrame containing the fetched stock price data.
Returns:
pd.DataFrame
The DataFrame with the calculated technical analysis indicators.
"""
# Price Indicators
# Volume-Weighted Average Price (VWAP)
# https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-overlays/volume-weighted-average-price-vwap
df['VWAP'] = volume_weighted_average_price(
high=df['High'],
low=df['Low'],
close=df['Close'],
volume=df['Volume'],
)
# Indices
# RSI:
# https://www.investopedia.com/terms/r/rsi.asp
df['RSI'] = RSIIndicator(
df['Close'],
window=14).rsi()
# Stochastic Oscillator:
# https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-indicators/stochastic-oscillator-fast-slow-and-full
df['StochOsc'] = StochasticOscillator(
df['High'],
df['Low'],
df['Close'],
window=14).stoch()
# Trend signals
# Moving Average Convergence Divergence (MACD):
# https://chartschool.stockcharts.com/table-of-contents/technical-indicators-and-overlays/technical-indicators/macd-moving-average-convergence-divergence-oscillator
macd = MACD(
df['Close'],
window_slow=26,
window_fast=12,
window_sign=9)
df['MACD'] = macd.macd()
df['MACDsig'] = macd.macd_signal()
df['MACDdif'] = macd.macd_diff()
return df
def merge_hist_with_forecast(self, df_hist: pd.DataFrame, df_past: pd.DataFrame | None, df_fcst: pd.DataFrame | None) -> pd.DataFrame:
# make sure we are merging the right thing
"""
Merge historical data with forecast data. If forecast data is available, merge it with historical data based on date.
If forecast data is not available, return the historical data as is.
Args:
df_hist: pd.DataFrame
Historical data
df_past: pd.DataFrame
Recent data used for comparison
df_fcst: pd.DataFrame or None
Forecast data
Returns:
df_merged: pd.DataFrame
Merged data
"""
if df_fcst is not None:
# Make sure that the previous hist close price is matching to that of the past close price
assert df_hist.Close.iloc[-2] == df_past.Close.iloc[-2]
df_hist.reset_index(inplace=True)
# in case there are overlapping dates, make sure to remove them
df_fcst = df_fcst.loc[~df_fcst["Date"].isin(df_hist["Date"]), ["Date", "Close"]]
date_diff = df_fcst.Date.iloc[0] - df_hist.Date.iloc[-1]
if date_diff > pd.Timedelta('3 days'):
self.logger.warning(f'Date diff between the first forecast and the last hist is {date_diff}')
df_merged = pd.concat([df_hist, df_fcst], ignore_index=True)
df_merged.set_index("Date", inplace=True)
else:
df_merged = df_hist
return df_merged
def plot_stock_metrics(
self,
df,
datasets,
savefig=False
) -> None:
"""
Plots the given stock metrics datasets as subplots.
This method takes in a DataFrame and a dictionary of datasets, where
each key is a dataset name and the value is a list of column names.
The method creates a figure with subplots for each dataset and plots
the corresponding columns of the DataFrame.
The figure is then saved to a file in the 'plots' directory in png format
with the ticker symbol as the filename.
Args:
df (pd.DataFrame)
The DataFrame to plot
datasets (dict)
A dictionary of datasets, where each key is a dataset name and
the value is a list of column names to be plotted
savefig (bool)
Whether to save the figure to a file
"""
numax = len(datasets)
fig, axes = plt.subplots(
nrows=numax,
ncols=1,
figsize=(6, 3*numax))
for i, ax in enumerate(axes.flat):
dataset = list(datasets.keys())[i]
colstoplot = datasets[dataset]
self.plot_stock_metrics_ax(
ax,
dataset,
df,
colstoplot)
plt.tight_layout()
if savefig:
rootdir = os.path.dirname(os.path.dirname(__file__))
fname = os.path.join(rootdir, 'plots', f'{self.ticker}.png')
plt.savefig(fname)
self.logger.info(f'Saved figure into: {fname}')
plt.close()
fig = None
return fig
def plot_stock_metrics_ax(
self,
ax:Axes,
dataset:str,
df:pd.DataFrame,
colstoplot:list) -> None:
"""
Plots specified stock metrics on the provided Axes object.
This function takes in an Axes object and plots the specified columns from
the DataFrame `df` on it. It formats the x-axis with major ticks set to
every Monday and minor ticks set to every day. The plot includes a title,
x and y labels, and optionally a legend if more than one column is plotted.
Additional shaded regions are added for certain datasets.
Args:
ax (matplotlib.axes.Axes): The flattened axes to plot on.
dataset (str): The name of the dataset, used for the title and y-label.
df (pd.DataFrame): The DataFrame containing the data to plot.
colstoplot (list): A list of column names to plot from the DataFrame.
Note:
- If `dataset` is 'Index' or 'Indices', the y-axis is limited to [0, 100]
and a shaded region between y=30 and y=70 is added.
- If `dataset` is 'Price' or 'Prices', a shaded region between 'Low' and
'High' columns is added.
"""
self.logger.info(f'plotting {colstoplot} in {dataset}')
colorcycle = ['black', 'blue', 'green', 'orange']
for i, col in enumerate(colstoplot):
ax.plot(
df.index,
df[col],
color=colorcycle[i],
label=col,
linewidth=2)
if dataset in ['Index', 'Indices']:
ax.set_ylim([0, 100])
# Add a transparent shaded region between y=30 and y=70
ax.fill_between(df.index, 30, 70, color='gray', alpha=0.3, label='30-70 Range')
elif dataset in ['Price', 'Prices']:
# Add a transparent shaded region daily lows and highs
ax.fill_between(df.index, df['Low'], df['High'], color='gray', alpha=0.3, label='Price Range')
# extract the Price rows for which 'High's are NaN
nanind = np.where(df.High.isna())
df_fcst = df['Close'].iloc[nanind]
if df_fcst.shape[0] > 0:
# append the last day before the forecasts
nonnanind = np.where(df.High.notna())
df_now = df['Close'].iloc[[nonnanind[0][-1]]]
df_now_fcst = pd.concat([df_now, df_fcst])
# connect the last available day and forecasts with a red line
ax.plot(df_now_fcst.index, df_now_fcst, color='red')
# plot only the forecasts with markers
ax.plot(df_fcst.index, df_fcst, color='red', marker='*', label='Forecast')
# else:
# # plot a transparent line across the full df.index just to make sure that x-axis limits are identical for all panels
# ax.plot(df.index, df[col], color='gray', alpha=0.0)
# # ax.fill_between(df.index, 30, 70, color='gray', alpha=0.0)
ax.set_xlim([df.index.min()-pd.Timedelta(days=5), df.index.max()+pd.Timedelta(days=5)])
# Format major ticks with year
# Set major ticks (every Monday with labels)
ax.xaxis.set_major_locator(mdates.WeekdayLocator(byweekday=mdates.MO))
# ax.xaxis.set_major_formatter(mdates.DateFormatter('%b %d'))
ax.xaxis.set_major_formatter(mdates.DateFormatter('%m.%d'))
# Set minor ticks (every day, but without labels)
ax.xaxis.set_minor_locator(mdates.DayLocator())
plt.setp(ax.get_xticklabels(), rotation=90, ha='center')
ax.set_ylabel(dataset)
ax.grid(True, linestyle='--', alpha=0.7)
ax.set_title(dataset)
# ax.set_xlabel('Date')
if len(colstoplot) > 1:
ax.legend(loc='upper left')
def get_fetcherror_fig(
self,
message
) -> plt.Figure:
"""
Fetches images/plot_error.png, annotates it and returns it as a matplotlib.pyplot.Figure object
Args:
message (str): message to be annotated on the displayed image
Returns:
plt.Figure: figure object containing the annotated image
"""
fig, ax = plt.subplots(
figsize=(5, 5)
)
# Load and display the image
parentdir = os.path.dirname(os.path.dirname(__file__))
fname = os.path.join(parentdir, 'images', 'plot_error.png')
img = plt.imread(fname)
ax.imshow(img)
# Remove axes ticks and labels
# ax.set_xticks([])
# ax.set_yticks([])
ax.axis('off') # removes both ticks and axes lines completely
ax.text(0.5, 0.05, message, fontsize=20, ha='center', va='center', transform=ax.transAxes)
return fig
if __name__ == '__main__':
ticker = 'AAPL'
# testing
from src.fetch_forecast import FetchForecast
from src.fetch_data import FetchData
dotenv.load_dotenv(dotenv.find_dotenv())
df_hist = FetchData(ticker, fetchperiodinweeks=12).run()
df_past, df_fcst = FetchForecast(ticker, df_hist).run()
df, fig = TechnicalAnalysis(ticker, df_hist=df_hist, df_past=df_past, df_fcst=df_fcst, plot_ta=True, savefig=True, debug=False).run()
# print(f'columns: {df.columns}')