Implicit conversion from float -24.419475655430716 to int loses precision