Implicit conversion from float -78.2608695652174 to int loses precision