It's not HTML. HTML doesn't even have namespaces. It claims to be XHTML, so you should be using parse_file.

Unfortunately, it's not well-formed XHTML. Lots of unescaped ampersands and less-than symbols, for starters.

The following fixes the escaping errors, but there's still a bunch of duplicate ids and unmatched tags:

#!/usr/bin/perl -w
use strict;
use warnings;

use XML::LibXML    qw( );
use LWP::UserAgent qw( );

sub text_to_xml {
   my ($text) = @_;
   $text =~ s/&/&/g;
   $text =~ s/</&lt;/g;
   return $text;
}

my $url = "http://www.marketwatch.com/story/brazil-mexico-stocks-face-
+quarterly-slides-2010-06-30?siteid=rss&rss=1";

my $ua = LWP::UserAgent->new();
my $response = $ua->get($url);
$response->is_success()
   or die "Request failed: ". $response->status_line() ."\n";

my $xhtml = $response->decoded_content(charset => "none");
$xhtml =~ s{(<script[^>]*>)(.*?)(</script>)}{"$1".text_to_xml("$2")."$
+3"}sieg;
$xhtml =~ s/&(?!#|[a-zA-Z0-9]{1,8};)/&amp;/g;

my $parser = XML::LibXML->new;
my $doc = $parser->parse_string($xhtml);
[download]

Remaining errors

There's a parser option that directs libxml to recover from errors if possible. It might help. If not, HTML::Parser is designed to handle garbage.