The CRC is very strong for detecting random corruption, because every bit of the data may affect any bit of the CRC.

The CRC is very weak for encryption, because it is a reversible mathematical operation. Just as you can solve a linear equation system by iterating out one variable/iteration, you can solve the CRC equation one bit at a time.

CRC was originally designed for use with a shift register with parallel access and a number of XOR gates. The intention was to be able to compute a very very good checksum with almost no hardware at a time when the world did not have any microprocessors.

For cryptography, you have to look at solutions that are not trivially reversible.

Anyway - you have still not described your problem. You mention RTC time. What RTC time? To/from a known product or component? And why didn't you start with telling that you had listened to a communication stream and had two bytes in the packet that you could not find how to compute?

There are many different 16-bit CRC alternatives. Several polynomials. With or without pre- or post-conditioning (setting bits initial bits to 1 or 0, inverting or not inverting the answer), and some implementations are intended to be fed the most significant bit first, while some are intended to be fed the most significant bit first.