The $ is just a character in the string. Is there a reason you think it would need to be escaped?

How is there a test for year 0 when there isn't a year 0 in the calendar?

I'm not sure that rule applies correctly to BC dates. Not having a year 0 would offset the rule by 1 year. The correction wasn't implemented in BC and therefore any testing of it for negative dates is not valid.

in the Ruby version:

1. the instructions specify that the hash keys will be symbols but they are actually strings
2. the instructions give a different case for the hash keys (camel case in the instructions, snake case in the tests)

expected error "404..." does not match testing "Invalid Value"

My comment was not about any difficulty in solving the problem.
For accuracy, the problem statement should match the actual problem and tests.
The problem stated that the loss was guaranteed to be a float, but one test passed in an int.
In mathematics integers are a subset of real numbers.
In programming INT and FLOAT are distinct types and you shouldn't claim to be passing one and then use the other.
The problem statement should remove the "guarantee" of a float, or the test should pass in a float.

Ruby. Having an integer 1 as apposed to a float 1.0 affects rounding in the calculations.

The parameters are not all as specified,
parameter (positive float, guaranteed) percentLossByMonth
the loss is given as an integer in at least one test.