The article discusses how high-frequency financial data can be used to estimate price variations. By calculating realized variance and realized absolute variation from intra-day prices, we can estimate integrated variance and spot volatility. These estimators can be improved using filtering and smoothing techniques. Realized multipower variation can help identify jumps in the price process, while realized bipower, tripower, and quadpower variation are more robust to market noise. However, there is a trade-off between bias and variance, with bias from market noise at high frequencies and variance from assumptions at low frequencies. Subsampling and averaging can help reduce this effect, allowing for calculations at higher frequencies.