This paper investigates the microstructural origins of two well-known stylized facts in financial markets: the leverage effect (negative correlation between returns and volatility) and rough volatility (volatility process with low regularity). The authors construct a microscopic price model based on a bi-dimensional Hawkes process that captures four key features of modern high-frequency trading: high endogeneity (most orders are reactions to other orders), absence of statistical arbitrage (balanced upward and downward jump intensities), asymmetry between bid and ask liquidity (more liquidity on the ask side), and the presence of metaorders (large orders split over time). By encoding these features into the kernel matrix of the Hawkes process, the authors show that under a nearly unstable regime (spectral radius close to 1), the rescaled price process converges to a Heston stochastic volatility model when only the first three features are present. The asymmetry parameter β > 1 generates a negative correlation between the Brownian motions driving the price and volatility, thus producing the leverage effect. When metaorders are additionally incorporated via heavy-tailed kernels, the limit becomes a rough Heston model, where the volatility process is driven by a fractional Brownian motion with Hurst parameter less than 0.5, matching empirical findings. This work provides a natural, non-ad-hoc explanation for both leverage effect and rough volatility, rooted in the microstructure of financial markets.