One thing they do is track what subs you subscribe to. So if a new account is created and subscribes to all the same communities as a banned account, they can fingerprint you that way.
They can also fingerprint you based on what browser and OS you use. Not just cookies. When you connect to websites there is info shared about you that goes beyond cookies.
Just because you got a new IP from your ISP doesn't mean that they don't share data about you with reddit, especially if reddit asks for info from your ISP. You don't know what info your ISP shares about you. Maybe your ISP has some kind of an EULA but I don't know.
If you really wanted to focus on bypassing IP ban and are willing to go to any effort, install something like Oracle Virtualbox, create a VM with a fresh install of windows, install a decentralized VPN in that virtual machine (Mysterium VPN is an example) because most VPN's have fixed exit nodes that will be banned by reddit. A decentralized VPN allows any user on the VPN to act as an exit node, so Mysterium has literally thousands of exit nodes that change every day. It's impossible to ban them all.
Then after connecting your VM to a decentralized VPN, you would want to use a different browser every time you sign up for reddit, and use a different email address every time. Then never subscribe to the same communities every time you sign up a new account.
And even then, they can use AI to scan comments and detect similarities between you and other accounts based on how you type.
At some point you have to ask, is it really worth it? They're going to an absurd length to ban people. I want a community where I can be me. I believe in free speech. I want to be free to shitpost. Saying something rude shouldn't result in automatic permaban.