Empirical Properties of SPY Returns

Statistical analysis of daily log returns for the SPDR S&P 500 ETF (SPY) over 10 years. These stylized facts of equity returns motivate the use of stochastic models beyond simple Geometric Brownian Motion.

Data source: Yahoo Finance adjusted close prices (daily frequency)

Why These Properties Matter

Model Selection

The empirical properties of returns—volatility clustering, fat tails, and negative skewness—determine which stochastic models are appropriate. GBM assumes none of these, while Heston captures clustering, jump-diffusion models capture fat tails, and Lévy processes offer flexible tail behavior.

Risk Management

The tail quantiles (1% and 99%) show the magnitude of extreme daily moves. Fat tails mean these extremes occur more frequently than a normal distribution predicts—critical for VaR calculations and portfolio risk assessment.

Volatility Dynamics

The autocorrelation of squared returns reveals volatility clustering: periods of high volatility tend to persist. This motivates stochastic volatility models like Heston that explicitly model time-varying volatility.

Option Pricing

The implied volatility smile observed in options markets reflects these empirical properties. Models that capture skewness and kurtosis produce more realistic option prices than Black-Scholes.

Explore how different stochastic models capture these properties: View all models →