Implicit conversion from float -61.15384615384616 to int loses precision