The precise measurement of the luminosity is one of the key requirements for every ATLAS analysis at the Large Hadron Collider (LHC) at CERN. Particularly in high precision measurements, the uncertainty on the luminosity can be one of the main limitations. Therefore, its reduction is the prime goal of the ATLAS luminosity programme, requiring a precise understanding of the contributing factors. The two largest individual components are the calibration transfer (extrapolating the measurement from the calibration regime to the physics regime) and the long term stability (stability of the measurement typically over a whole year), both determinations involving the track counting luminosity measurement. This technique uses charged particle tracks to measure the delivered luminosity. The uncertainty of this measurement is dependent on the track selection and therefore, the performance of three selections is compared over different LHC fill configurations. The goal of this study is to determine the stability of the measurement and if possible correct observed effects.