2-part solution:
Using IPTABLES, you can do some PREROUTING to capture all traffic to a.b.c.d (the remote server), then mask that traffic and redirect it to e.f.g.h (your noip.com IP)
Using a cron script, running every N minutes, if the IP has changed, purge the IPTABLES rules and re-insert them with the new IP.
I haven't tested it, but this is how I would begin:
iptables -t nat -I PREROUTING -s [localwanip] -d [remoteip] -p -m tcp --dport [port] --to-destination [newremoteip]
...and the script, again proofing required (results may vary depending on versions and system flavour):
#!/bin/bash
NOIPNAME=yourname.noip.com
# Your IP on your WAN interface
LOCALIP=1.2.3.4
# The IP the software is _mistakenly_ trying to talk to
REMOTEIP=4.3.2.1
# The TCP port the software is using to connect to the remote IP
PORT=1234
# Just a file to keep track of what the last IP was...
REMEMBERFILE=/var/run/oldip.txt
# and now the magic, if it works...
HOSTLINE=$(host $NOIPNAME ns1.no-ip.com | grep 'has address')
HOSTLEN=$(echo $HOSTLINE | wc -m)
# Make sure the return string is > 8 characters (1.2.3.4\n)
if [ $HOSTLEN -lt 8 ]; then
# Host resolve failure
echo "Bad host"
exit 1
fi
# Extract the IP from the return string.
DYNIP=$(echo $HOSTLINE | sed -rn 's/^.* ([0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}).*/\1/p')
OLDIP=$(cat $REMEMBERFILE)
if [ "x$DYNIP" = "x$OLDIP" ]; then
# Nothing to do.
exit 0
else
echo $DYNIP > $REMEMBERFILE
fi
# Flush
iptables -t nat -F
# and re-write
iptables -t nat -I PREROUTING -s $LOCALIP -d $REMOTEIP -p -m tcp --dport $PORT --to-destination $DYNIP
exit 0
Lastly, add it to cron; the below runs every 10 minutes (via this part: */10).
echo "*/10 * * * * root /your/path/to/script" >> /etc/crontab
Caveats:
Now the IPTABLES will be tricky, if the command I gave doesn't work (probably won't - it was just a guess) - have a dig in google, for example:
http://ribbalicious.com/rewrite-destination-ip-address-using-iptables/
Good luck.