Abstract

Context. The O’Connell effect is a phenomenon in eclipsing binary (EB) systems that consists of unequal maxima in a light curve when out of eclipse. Despite being known for decades and with several theories proposed over the years, this effect is still not fully understood. Aims. Our goal is to find different O’Connell effect properties using a multiband approach, compare them with each other, and find correlations between these properties and the physical parameters of the systems. Methods. We present the analysis of 14 new EBs that show the O’Connell effect using multiband data extracted from the Asteroid Terrestrial-impact Last Alert System (ATLAS) and Zwicky Transient Facility (ZTF) all-sky surveys. We measured the difference in maximum amplitudes (?m) alongside different light curve features in different passbands via a new modeling process that uses Gaussian fits. We created a brand-new phenomenological model for O’Connell effect systems whose analysis had previously been hampered by overfitting. Results. Although the magnitude of the O’Connell effect seems to be more pronounced at shorter effective wavelengths, supporting the idea that cool starspots cause the effect, a conclusive correlation is not found. On the other hand, we do find strong correlations between the magnitude of the effect and both the temperature and the period, both of which are inconsistent with a previous study. We also find that in systems that show both positive and negative O’Connell effects, there are different correlations with the aforementioned parameters. Conclusions. We conclude that, even though starspots may be one cause of the O’Connell effect, it is likely a multipronged phenomenon; for instance, the physical interaction between the components of close binary stars may be another important factor. © The Authors 2025.