At the core off this, even with port forwarding, you would have to determine if you are on symmetric nat or full cone nat or port restricted nat, or CGNAT etc from your home router/ISP. If you are behind a router and in a full cone nat topology, you should be good (its a clean port forward similar to being on a VPS/Cloud VM).
If you are on a symmetric nat or port restricted nat, or CGNAT, you will likely have to use --relay option to be able to connect and earn, if the other two options (public VM or full cone nat) are not feasible for you.
As mentioned in prior update here: (Update 19th June, 2025 - #23 by Shu) (see below):
There are a series off PRs in the works to rely on connection information than identify, including 180 second dial backs, and eventually automatically reach a conclusion on the type of node/network topology you are on so flags such as --no-upnp, --relay, --ip etc would not be required.
The process would be super streamlined including auto-detecting which network adapter is transmitting and receiving traffic for antnodes running within the OS as well etc.
All of the above needs to be rolled out in phases with majority off the code base for node operators to be on the latest versions to have maximum impact towards a healthier network.
With today’s release, some of these parts/phases have been rolled out, but more to come in future releases, though as the update states, the priority has been on improving downloader & uploader behavior especially with CLN, especially on mainnet and for the IF program.