Performance Analysis
Evaluation of the Ex-Post Performance of Securities and Rules-Based Strategies on an Absolute, Relative and Risk-Adjusted Basis.
Paul McAteer, MSc, MBA
pcm353@stern.nyu.edu
Contents¶
- Import libraries.
- Specify investment universe and historic time period.
- Define parameters for raw data import and storage.
- Import price data over sample period, save and perform preliminary inspection.
- Pass price data to dataframe. Clean data structure.
- Visualize asset price evolution. Calculate and inspect daily and monthly returns.
- Construct cap-weighted benchmark.
- Construct equal-weighted benchmark.
- Compute expected return vector and sample covariance matrix.
- Compute idiosyncratic risk statistics.
- Construct efficient frontier based on classical Markowitz model.
- Shrink covariance matrix.
- Construct efficient frontier based on robust covariance estimate.
- Design Risk Parity portfolio.
- Design Black-Litterman optimized portfolio.
- Create security weighting scheme for Black-Litterman portfolios..
- Generate strategy backtest returns applying security weighting schemes.
- Generate risk-adjusted performance metrics for strategies.
1) Import libraries.¶
In [7]:
# Import the python libraries
import pandas as pd
import numpy as np
import math
from datetime import datetime
import matplotlib.pyplot as plt
2) Specify investment universe and historic time period.¶
In [8]:
Securities = "MSFT AAPL AMZN GOOG NVDA BRK-A JNJ V PG JPM UNH MA INTC VZ HD T PFE MRK PEP WMT BAC XOM DIS KO CVX CSCO CMCSA WFC BA ADBE"
Start = "2016-05-30"
End = "2020-06-30"
#Start = "2010-06-30"
#End = "2020-06-30"
3) Define parameters for raw data import and storage.¶
In [9]:
# Select File Type for upload of Security Data
filetype =".csv"
# Specify Local Storage Location
path = r"C:\Users\delga\Desktop\NYU\CQF_Work\Portfolio_Management"
# Convert data parameters to string for file naming purposes
Sec_Dates = Securities,Start, End
def convertTuple(tup):
str = '_'.join(tup)
return str
conv = convertTuple(Sec_Dates)
# Converted data parameters + filetype = filename
filename = conv+filetype
# Join path, filename & filetype for single reference "File"
import os
File = os.path.join(path,filename)
print(File)
4) Import price data over sample period, save and perform preliminary inspection.¶
In [10]:
import yfinance as yf
data = yf.download(Securities, start=Start, end=End)
In [11]:
# View data
data
Out[11]:
In [12]:
# Save data to named file
data.to_csv(File)
In [13]:
# Inspecting the first 3 lines and all columns lines of the saved CSV file
f =open(File,"r")
f.readlines()[:3]
Out[13]:
5) Pass price data to dataframe. Clean data structure.¶
In [14]:
#The filename passed to the pd.read_csv() function creates the daily price dataframe.
#Specified that the first two rows shall be handled as headers.
#Specified that the first column shall be handled as an index.
#Specified that the index values are of type datetime
df_csv = pd.read_csv(File, header= [0,1], index_col=0, parse_dates=True,)
df_csv.info()
In [15]:
df_csv
Out[15]:
In [16]:
# Define string and substring to count securities in portfolio. Reduce Dataframe to daily adj close for 30 securities
string = Securities
substring = " "
Sec_count = string.count(substring)+1
df_csv = df_csv.iloc[:,0:Sec_count]
df_csv
Out[16]:
In [17]:
# Check Column names
for col in df_csv.columns:
print(col)
In [18]:
# Create single level header from multilevel header
df_csv.columns = df_csv.columns.map('|'.join).str.strip('|')
df_csv.columns = df_csv.columns.str.replace(r'Adj Close|$', '')
df_csv.columns = df_csv.columns.str.lstrip('|') # strip suffix at the right end only.
# Check column names
print(df_csv.columns)
In [19]:
# View Adj Close Dataframe
df_csv
Out[19]:
In [20]:
# View metadata
df_csv.info()
In [21]:
# Identify null values in dataset
df_csv.isnull().any()
Out[21]:
In [22]:
# Drop null values in dataset
df_csv = pd.DataFrame(df_csv.dropna().round(4))
df_csv.info()
6) Visualize asset price evolution. Calculate and inspect daily and monthly returns.¶
In [23]:
# Plot Daily Price Evolution
df_csv.plot(figsize=(12, 60), subplots=True);
In [24]:
# Calculate and plot daily returns
returns_daily = df_csv.pct_change()
returns_daily.plot(figsize=(12, 60), subplots=True);