If you have the pool member defined on port anything but port 0, you shouldn't need to specify a port in the monitor. The monitor will use the pool member's port.
I'd assumed that RTMP was running over HTTP (
Click here), so I suggested an HTTP monitor. If you don't configure a send string in the TCP or HTTP monitor, the monitoring daemon will only check to see if it gets a response to a TCP handshake request. If you capture a tcpdump of a successful request, you could configure the monitor send string with the request. Add a string found in a successful response as the receive string, and you can check that the application is responding correctly before marking the pool member up.
Aaron