Nulls can be a little challenging when considered as 'data'.
In general I would always seek an explicit value or value set to filter with. In this case I think it would probably be using ISNULL to set a flag in a separate calculated field and then filter on that since you know it will have one of two possible values. It also make it easier to check the results against expectations.
Monarch can be a little fiddly with nulls - for example adding "abc" and a null string results in a null string. When you work with database sources, there's an option to replace null values with defaults (0,"",01/01/1970), but I frequently still find myself defining calculated fields like if(isnull()=1;0;[field]).
With filters, it might be helpful to think of an expression having three answers, not two. a=b can be either true (1) if both a and b are non-null and are equal, or false (0) if they are both non-null but unequal, or null (1/0) if one or both of them is null.
So <>1 can be true if field is non-null and not equal to 1, false if non-null and equal to 1, or null if field is null. The filter only returns the true matches, so in your case you'd need to include the nulls with: