Implicit conversion from float -7.363083164300207 to int loses precision