We present a yield simulator to predict the number and characteristics of planets, false positives and false alarms in transit surveys. The simulator is based on a galactic model and the planet occurrence rates measured by the Kepler mission. It takes into account the observation window function and measured noise levels of the investigated survey. Additionally, it includes vetting criteria to identify false positives. We apply this simulator to the Next Generation Transit Survey (NGTS), a wide-field survey designed to detect transiting Neptune-sized exoplanets. We find that red noise is the main limitation of NGTS up to 14 mag, and that its obtained level determines the expected yield. Assuming a red noise level of 1 mmag, the simulation predicts the following for a 4-yr survey: 4 plus or minus 3 Super-Earths, 19 plus or minus 5 Small Neptunes, 16 plus or minus 4 Large Neptunes, 55 plus or minus 8 Saturn-sized planets and 150 plus or minus 10 Jupiter-sized planets, along with 4688 plus or minus 45 eclipsing binaries and 843 plus or minus 75 background eclipsing binaries. We characterize the properties of these objects to enhance the early identification of false positives and discuss follow-up strategies for transiting candidates.