Investigate this fourth fact by creating the ACF plots of squared and absolute returns.

1. Create the ACF plots:

fig, ax = plt.subplots(2, 1, figsize=(12, 10))

smt.graphics.plot_acf(df.log_rtn ** 2, lags=N_LAGS, 
                      alpha=SIGNIFICANCE_LEVEL, ax = ax[0])

ax[0].set(title='Autocorrelation Plots',
          ylabel='Squared Returns')

smt.graphics.plot_acf(np.abs(df.log_rtn), lags=N_LAGS, 
                      alpha=SIGNIFICANCE_LEVEL, ax = ax[1])
ax[1].set(ylabel='Absolute Returns',
          xlabel='Lag')

Executing the preceding code results in the following plots:

We can observe the small and decreasing values of autocorrelation for the squared and absolute returns, which are in line with the fourth stylized fact.