Implicit conversion from float -41.0850439882698 to int loses precision