Do stock prices go down on Fridays?

It seems that stock prices have tendency to go up on Mondays and go down Friday. Here is quick analysis of S&P500 weekday prices from 2015 to 2020.

Getting S&P 500 data

import pandas as pd
import yfinance as yf
import os
from pandas.core.frame import DataFrame
from more_itertools.recipes import flatten
# Save csv file from 
# https://datahub.io/core/s-and-p-500-companies to 'sp500.csv'
df = pd.read_csv("stocks/week/sp500.csv")
df.head()

First 5 rows of S&P companies

Step 2: Get stock price histories for 2015–2020

START_DATE = "2015-08-28"
END_DATE = "2020-08-28"
def get_history(symbol):
    ticker = yf.Ticker(symbol)
    df = ticker.history(period="1d", start=START_DATE, end=END_DATE)
    if not df.empty:
        df["Symbol"] = symbol
    return df
get_history(df["Symbol"][0]).head()

First 5 days of stock prices for MMM

Step 3: Get differences between Monday and other weekday

def get_weekly_diffs(df):
    symbol = df["Symbol"][0]
    df = df.reset_index()
    df["weekday"] = df["Date"].apply(lambda x: x.weekday())
    if df[df["weekday"] == 0].empty:
        return {}
    # getting first row that starts with Monday
    df = df[df[df["weekday"] == 0].index[0] :]
    prev_monday_price = 0
    full_weeks = []
    is_monday = False
    is_friday = False
    week = []
    for _, row in df.iterrows():
        weekday = row["weekday"]
        # filter weeks were all days were traiding days
        is_monday = weekday == 0
        is_friday = weekday == 4
        if is_monday:
            prev_monday_price = row["Open"]
            price_diff = 1
            week = [row]
        else:
            if prev_monday_price == 0:
                continue
            price_diff = row["Open"] / prev_monday_price
            week.append(row)
        row["diff"] = price_diff
        if is_friday and len(week) == 5:
            full_weeks += week
     if not full_weeks:
         return {}
     df = DataFrame(full_weeks)
     df["Symbol"] = symbol
         return df.to_dict("r")
DataFrame(get_weekly_diffs(get_history(df["Symbol"][0]))).head()

First 5 days of stock prices for MMM with weekdays differences comparing to Monday in percentage

Step 4: Find mean weekday differences for all stocks

os.sys.path.append('../github/python_recipes/')
from utils.misc import concurrent_map
histories = []
# Next check is needed because some stocks 
# might not contain data at all
for df_h in concurrent_map(get_history, df["Symbol"]):
    if not df_h.empty:
        histories.append(df_h)
symbol_diffs = flatten(concurrent_map(get_weekly_diffs, histories))
df = DataFrame(symbol_diffs)
df = (
    df[["weekday", "diff"]]
    .groupby("weekday", group_keys=False)
    .mean()
    .reset_index(drop=True)
)
df = df * 100
df.head()

Mean weekdays differences comparing to Monday(0 — Monday, 5 — Friday)

Step 5: Visualize data

from seaborn import barplot
index2weekday = dict(
  zip(
    range(0,5),
    ['Monday', 'Tuesday', 'Wednesday', 'Thursday','Friday']
  )
)
df_percent_diff = df - 100
df_percent_diff.index = [
  index2weekday[w] 
  for w in df_percent_diff.index
]
barplot(
  x=df_percent_diff.index, 
  y=list(flatten(df_percent_diff.values))
)
df_percent_diff

Mean weekdays differences comparing to Monday

Weekdays change in percentage comparing to Monday

You can see from table and chart that on Friday stock prices go up to ~0.19% on “average”.