Portfolio Optimization & Risk Analysis

This project built a comprehensive portfolio management system that simulates random asset allocation across major stocks and calculates key financial metrics including returns, volatility, and risk-adjusted performance. I developed a complete portfolio analytics framework using Python to evaluate investment strategies and portfolio performance over time.

💻 Tech Stack:

• Python for financial calculations and portfolio modeling
• Pandas for time series data manipulation and financial computations
• NumPy for random weight generation and mathematical operations
• Plotly & SciPy for interactive portfolio performance visualisation, statistical analysis and risk metrics

🧪 Data Pipeline:

• Data Preparation: Loaded and sorted stock data chronologically using sort_values() by Date to ensure proper time series analysis for portfolio calculations.
• Random Portfolio Generation: Used np.random.seed() and np.random.seed(9) to create randomized asset allocation weights, then normalized them using weights / np.sum(weights) to ensure they sum to 100%.
• Portfolio Normalisation: Applied a custom normalize() function to standardize all stock prices to their initial values, creating a baseline for relative performance comparison across different price ranges.
• Portfolio function development: Built a reusable portfolio_allocation() function that encapsulates the entire workflow for testing different weight combinations and portfolio strategies.
• Risk metrics calculation: Computed cumulative return, standard deviation (volatility), average daily return, and Sharpe ratio (assessesment of the risk-adjusted returns of an investment) using np.sqrt(252) for annualization.

📊 Code Snippets & Visualisations:

np.random.seed()

# Create random weights for the stocks
weights = np.array(np.random.random(9))

# Random Asset Allocation & Calculate Portfolio Daily Return
weights = weights / np.sum(weights)
print(weights)

# Define Normalization function
def normalize(df):
    x = df.copy()
    for i in x.columns[1:]:
        x[i] = x[i] / x[i][0]
    return x

# Enumerate returns the value and a counter as well
for counter, stock in enumerate(df_portfolio.columns[1:]):
    df_portfolio[stock] = df_portfolio[stock] * weights[counter]
    df_portfolio[stock] = df_portfolio[stock] * 1000000

# Calculate the portfolio daily return
df_portfolio['portfolio daily % return'] = 0.0000

for i in range(1, len(stocks_df)):
    # Calculate the percentage of change from the previous day
    df_portfolio['portfolio daily % return'][i] = (
        (df_portfolio['portfolio daily worth/$'][i] - df_portfolio['portfolio daily worth/$'][i - 1])
        / df_portfolio['portfolio daily worth/$'][i - 1]
    ) * 100

# Create a function for stock portfolio allocation
# Assume $1000000 is total amount for portfolio
def portfolio_allocation(df, weights):
    df_portfolio = df.copy()

    # Normalize the stock values
    df_portfolio = normalize(df_portfolio)

    for counter, stock in enumerate(df_portfolio.columns[1:]):
        df_portfolio[stock] = df_portfolio[stock] * weights[counter]
        df_portfolio[stock] = df_portfolio[stock] * 1000000

    df_portfolio['portfolio daily worth in $'] = df_portfolio[df_portfolio.columns[1:]].sum(axis=1)

    df_portfolio['portfolio daily % return'] = 0.0000

    for i in range(1, len(stocks_df)):
        # Calculate the percentage of change from the previous day
        df_portfolio['portfolio daily % return'][i] = (
            (df_portfolio['portfolio daily worth in $'][i] - df_portfolio['portfolio daily worth in $'][i - 1])
            / df_portfolio['portfolio daily worth in $'][i - 1]
        ) * 100

    # Set the value of first row to zero, as previous value is not available
    df_portfolio['portfolio daily % return'][0] = 0
    return df_portfolio

# Plot the portfolio daily return (Figure 1)
fig = px.line(
    x=df_portfolio.Date,
    y=df_portfolio['portfolio daily % return'],
    title='Portfolio Daily % Return',
    labels={"x": "Date", "y": "Daily Percentage Return"}
)
fig.show()

# Cumulative return of the portfolio
cumulative_return = (
    (df_portfolio['portfolio daily worth/$'][-1:] - df_portfolio['portfolio daily worth/$'][0])
    / df_portfolio['portfolio daily worth/$'][0]
) * 100
print('Cumulative return of the portfolio is {} %'.format(cumulative_return.values[0]))

# Calculate the average daily return
print('Average daily return of the portfolio is {} %'.format(df_portfolio['portfolio daily % return'].mean()))

# Portfolio Sharpe ratio
sharpe_ratio = (
    df_portfolio['portfolio daily % return'].mean()
    / df_portfolio['portfolio daily % return'].std()
    * np.sqrt(252)
)
print('Sharpe ratio of the portfolio is {}'.format(sharpe_ratio))
						

🌟 Key Insights:

• Random portfolio allocation provides baseline performance benchmarks for comparing against optimized strategies, revealing how diversification across 9 stocks performs under equal-weight and random-weight scenarioss
• Daily return volatility patterns indicate portfolio risk characteristics with higher volatility periods corresponding to market stress, while Sharpe ratio quantifies whether returns adequately compensate for risk takeno
• Normalization to Day 1 baseline enables fair comparison across stocks with different price levels, allowing proper weight allocation based on percentage changes rather than absolute dollar amounts
• Cumulative returns demonstrated the compound effect of daily performance over the investment period

🧗🏾 Challenge Faced:

I initially encountered an indexing error when calculating portfolio daily returns because I was trying to access the previous day's value for the first row, which doesn't exist. The calculation df_portfolio['portfolio daily % return'][i-1] failed on the first iteration. I solved this by explicitly setting the first day's return to 0 using df_portfolio['portfolio daily % return'][0] = 0 after the loop, and ensuring the loop started from index 1 rather than 0. This approach properly handled the edge case while maintaining accurate percentage calculations for all subsequent trading days.

View on GitHub

← Back to Projects