Implicit conversion from float -11.919040479760099 to int loses precision