You're right that the total forging balance matters. I meant that the probability looks normal to me taking into account the number of forgers after the recent upgrade.
I believe forging on a public node has a slight advantage than on a private node behind NAT like in Nxt, because a public node is connected to more peers and can propagate a generated block a few msec faster, orphaning competitor block. This should be more so in HEAT because HEAT has faster blocks and propagation speed is more important than in Nxt.
That could be the case, but I dont know enough about it to say for sure.
Its always hard when you have to try and seperate what an issue and what is just luck. Looking at my node, its been doing very poorly after the upgrade to 1.0.1, so of course I wonder if its luck or I have to tweak some settings. At least the community has made some pretty good tools to track forging and other network stats.