Implicit conversion from float -13.236459709379119 to int loses precision