Implicit conversion from float -92.50950570342206 to int loses precision