Ah, no, I probably wouldn't do it any simpler in Racket. Racket tends to de-emphasize more complicated function composition, which tends to serve it just fine. I've always adopted a somewhat pro-functional approach in my usage of the language, but even here there wouldn't really be a distinctly better way to do it.

I guess I was just hoping that Haskell, being a more composition-oriented language in many ways, would have some nice built-in to do that for me.

In concatenative programming languages, composition is the default, and it makes point-free code much easier to write. In Factor, for example, you could write your example as:

    : inexact-matches ( x y -- n )
      [ matches ] [ exact-matches ] 2bi - ;

That is, run both “matches” and “exact-matches” on both inputs, then take the difference of the results. If Haskell had functions of multiple parameters and results, 2bi’s type would look like this:

    (a, b, ((a, b) -> c), ((a, b) -> d)) -> (c, d)