Topic 1 – Basic Stock Price Time Series (S&P 500)

Dataset

• Source: S&P 500 stock data – Kaggle
• Link: https://www.kaggle.com/datasets/camnugent/sandp500

Download Instructions

1. Open https://www.kaggle.com/datasets/camnugent/sandp500
2. Log in to Kaggle.
3. Click "Download".
4. Extract ZIP to data/.
5. Use all_stocks_5yr.csv.

Data Loading

import pandas as pd
df = pd.read_csv("data/all_stocks_5yr.csv")
aapl = df[df["Name"] == "AAPL"].copy()
aapl["date"] = pd.to_datetime(aapl["date"])
aapl = aapl.set_index("date").sort_index()

Implementation Steps

1. Data Exploration

• Load the dataset and select a company (e.g., AAPL, MSFT, GOOGL)
• Visualize the time series (close price over time)
• Check for missing values and handle them appropriately
• Examine basic statistics (mean, std, min, max)

2. Exploratory Data Analysis (EDA)

• Plot time series with trend lines
• Perform time series decomposition (trend, seasonality, residual)
• Calculate and visualize ACF/PACF plots
• Check for stationarity using visual inspection and ADF test

3. Data Preprocessing

• Handle missing values (forward fill, interpolation, or removal)
• If non-stationary, apply differencing (1st or 2nd order)
• Split data into train/validation/test sets (temporal split)

4. Model Building

• ARIMA Models:
• Use ACF/PACF to identify initial (p, d, q) parameters
• Try multiple ARIMA configurations
• Use AIC/BIC for model selection
• SARIMA Models (if seasonality exists):
• Identify seasonal patterns
• Fit SARIMA models with seasonal components

5. Model Evaluation

• Generate forecasts on validation/test sets
• Calculate metrics: MAE, RMSE, MAPE
• Visualize actual vs predicted values
• Analyze residual diagnostics (ACF of residuals, normality tests)

6. Forecasting

• Generate future forecasts (e.g., next 30 days)
• Include confidence intervals
• Visualize forecasts with historical data

Expected Deliverables

1. EDA Report:

• Time series plots
• Decomposition plots
• ACF/PACF plots
• Stationarity test results

1. Model Results:

• Best model parameters (p, d, q)
• Model diagnostics
• Forecast accuracy metrics
• Forecast plots with confidence intervals

1. Code:

• Well-documented Python notebook
• Functions for data loading, preprocessing, modeling, and evaluation

Tips

• Start with a well-known stock (AAPL, MSFT) for easier interpretation
• Stock prices are typically non-stationary; differencing is usually required
• Consider log transformation to stabilize variance
• Use walk-forward validation for time series cross-validation
• Compare multiple models and select the best based on validation performance