One consequence of climate change is that summer heat waves increase in frequency and intensity, and so do uncomfortably hot summer nights. One effective way to counter this effect is night ventilation. As connected temperature sensors are growing more common, they provide easily accessible data, from which nocturnal window openings can be inferred. Machine or deep learning algorithms, common in the literature, tend to require many labelled data. This article instead proposes an algorithm relying on expert rules and knowledge of the thermal behaviour of the building. It detects night openings or closings based, respectively, on a decrease or increase of indoor temperature. In our case study of an apartment equipped with connected window and temperature sensors, the algorithm is able to estimate the number of hours of night ventilation to within ±15%. It was more reliable in the hot summer of 2022 than in the cooler summer of 2021: the hypothesis of an effect of opening on temperature was then more realistic. This knowledge will eventually be used to quantify the potential temperature decrease from opening windows during summer nights, so as to provide residents with personalised advice on their window management.