The goal of the research was to fix a problem with estimating daily variance in stock prices when there's noisy data. The researchers came up with a new method to correct this bias by looking at the general pattern of the noise. They tested it with computer simulations and found that it works well even when the noise changes.