We present a novel approach to the determination of the pointing of Imaging Atmospheric Cherenkov Telescopes (IACTs) using the trajectories of the stars in their camera's field of view. The method starts with the reconstruction of the star positions from the Cherenkov camera data, taking into account the point spread function of the telescope, to achieve a satisfying reconstruction accuracy of the pointing position. A simultaneous fit of all reconstructed star trajectories is then performed with the orthogonal distance regression (ODR) method. ODR allows us to correctly include the star position uncertainties and use the time as an independent variable. Having the time as an independent variable in the fit makes it better suitable for various star trajectories. This method can be applied to any IACT and requires neither specific hardware nor interface or special data-taking mode. In this paper, we use the Large-Sized Telescope (LST) data to validate it as a useful tool to improve the determination of the pointing direction during regular data taking. The simulation studies show that the accuracy and precision of the method are comparable with the design requirements on the pointing accuracy of the LST (≤14″). With the typical LST event acquisition rate of 10 kHz, the method can achieve up to 50 Hz pointing monitoring rate, compared to O(1) Hz achievable with standard techniques. The application of the method to the LST prototype (LST-1) commissioning data shows the stable pointing performance of the telescope.