I just discovered this problem on 18.0.1 and a branch of develop updated today.
My test system is a VM with CentOS 7, and OpenNMS with the default configuration. I only added some nodes to the system (nothing else).
On the screenshot, you can see that the availability shown is negative for a node, even knowing that there are not outages for the node in question.
The negative availability appears as soon as the nodes are being properly categorized and dashboard is rendered.