Implicit conversion from float -15.488850771869636 to int loses precision