Alright, I understand how it’s intended to work now. Might be worth clarifying on MP page.
Unfortunately this isn’t any kind of solution to replicates movement. A solution to replicates movement would be the exact same, except with client-side prediction and interpolated corrections. This is feasible but difficult without RPCs.
What you’re trying to do is an entirely different system.
Furthermore, it’s not a solution for it’s actual intended purpose in any remotely competitive or persistent environment because it is prone to cheating.