NIST seems to have it as a guideline for memorised secrets:

Verifiers SHOULD NOT impose other composition rules (e.g., requiring mixtures of different character types or prohibiting consecutively repeated characters) for memorized secrets. Verifiers SHOULD NOT require memorized secrets to be changed arbitrarily (e.g., periodically). However, verifiers SHALL force a change if there is evidence of compromise of the authenticator.

The most prominent source is NIST, which states:

Verifiers SHOULD NOT require memorized secrets to be changed arbitrarily (e.g., periodically). However, verifiers SHALL force a change if there is evidence of compromise of the authenticator. (source)

I found an explanation on a different site:

It’s difficult enough to remember one good password a year. And since users often have numerous passwords to remember already, they often resort to changing their passwords in predictable patterns, such as adding a single character to the end of their last password or replacing a letter with a symbol that looks like it (such as $ instead of S).