Implicit conversion from float -41.198347107438025 to int loses precision