Implicit conversion from float -13.058823529411768 to int loses precision