As a general rule, a good way to build complex regexes is to be long-winded. Don't optimize. List each case in full:

  $interest !~ (
    ^  # start of line
    -? # optionally negative
    (?:
        \d+       # numbers with no decimal point
      | \d+\.\d+  # numbers with decimal point in middle
      | \.\d+     # numbers with decimal point at start
      | \d+\.     # numbers with decimal point at end
    )
    $  # end of line
   /x;
[download]

From this starting point, you can then attempt to optimize some the cases, if you want. But there's no need to: perl will optimize them for you. --Dave