Using Grinold's Fundamental Law of Active Management, the expected standard deviation (volatility) of alpha is driven by the Information Coefficient (IC) and the residual risk ($\omega$). Standard deviation of alpha ($\sigma_\alpha$) = $IC \times \omega$. Given $IC = 10\%$ and $\omega = 18\%$, the standard deviation of alpha is $\sigma_\alpha = 0.10 \times 0.18 = 0.018$ or $1.8%$.

Assuming alphas are normally distributed with a mean of 0%, we want to find the proportion of stocks with an alpha greater than 3.6% or less than -3.6%. A value of 3.6% is exactly 2 standard deviations away from the mean ($\pm 2 \times 1.8\% = \pm 3.6\%$). In a normal distribution, approximately 5% of observations lie outside $\pm 2$ standard deviations from the mean (using the empirical rule). Number of stocks = $5% \times 500 = 25$.