We use a theoretical model of the γd→K+ K- np reaction adapted to the experiment done at LEPS where a peak was observed and associated with the Θ+ (1540) pentaquark. The study shows that the method used in the experiment to assign momenta to the undetected proton and neutron, together with the chosen cuts, necessarily creates an artificial broad peak in the assumed K+ n invariant mass in the region of the claimed Θ+ (1540), such that the remaining strength seen for the experimental peak is compatible with a fluctuation of 2σ significance.