Implicit conversion from float -9.01574803149606 to int loses precision