Implicit conversion from float -9.271844660194176 to int loses precision