Originally Posted by EDepth

A few additions:
1. Collect decline data as well as approval data, usually someone who is using a stolen CC will get declined a few times before finding a working card. Save every bit of data you can and keep it. Previous history from last year can come in handy.
2. Compare join page IP with the IP of the first user to login under the account.
3. Use a geo-ip database to compare the location of the user's IP vs. their credit card details.
4. Even if the carder is switching IPs / names / cards / etc, you can usually find some similarities between recent declines and a new sale.
5. Create an alert system that looks for certain things, such as certain email addresses, female names, etc.