Implicit conversion from float -20.106382978723403 to int loses precision