Implicit conversion from float -24.840499306518723 to int loses precision