Firstly, companies change logos, change names, force password resets or your password will leak in a breach. How does this system accommodate those scenarios?
Also, once a few of your passwords are in data breaches, anyone who wants to target you can derive your admittedly less complex formula.
It is not out of the question that there already is or will be technology to compare multiple passwords from the same person from different leaks and derive the formula.
It's true that's a hypothetical attack vector.
In practice, so few people use this approach that you're not going to see a general-purpose deployable attack suite based on this principle. It'd be useful for a targeted attack, but if you're under a persistent, targeted threat there are better tools (spear-phishing, social engineering, physical compromise, etc.).