I am reading through about transaction isolation levels (serializable, repeatable read, read committed, read uncommitted) and am a little unclear about their particular uses. Will I comprehend it factual that for instance in banking applications (money transfers etc) the isolation level used is most likely serializable since no relaxation of the rule is permitted?

It can be you to decide on the isolation level for the way you would like the application to operate. You might require different isolation levels for various db calls, based on your work using the data.

You will find no solid rules for just about any specific industries for example banking, although some people might data (for instance financial transaction data) is clearly important and for you to do all you are able to prevent loss of data and data inconsistency.

Isolation levels are described much better than I'm able to here.
http://en.wikipedia.org/wiki/Isolation_%28database_systems%29