Implicit conversion from float -92.47058823529412 to int loses precision