Transaction velocity counting for fraud detection

1. A method for transaction counting in a fraud detection system, the method comprising:

storing in a memory of a computer, a plurality of initial-buckets associated with a first time interval and a plurality of roll-up buckets associated with a second time inter-val, wherein the first time interval is contained within the second time interval;

at the computer, receiving a transaction from a network, the transaction including a transaction timestamp;

formatting the transaction timestamp to generate a formatted transaction time-stamp;

based on said transaction timestamp, determining a particular initial-bucket of said plurality of initial-buckets in which to store the transaction;

wherein said particular initial-bucket is associated with a sub-time interval of said first time interval;

wherein said particular initial-bucket has a bucket count that indicates a number of one or more previously received transactions that occurred during said sub-time interval of said first time interval;

in response to the computer determining that a time indicated by said transaction timestamp is not within said sub-time interval of said first time interval, the computer adding the bucket count for the particular initial-bucket to a bucket count of a particular roll-up bucket; and

storing the transaction in the particular initial-bucket in the memory of the computer.

2. The method of claim 1 , wherein said determining a particular initial-bucket of said plurality of initial-buckets to store the transaction is based on said formatted transaction timestamp.

3. The method of claim 1 , wherein said particular initial-bucket is associated with a bucket timestamp.

4. The method of claim 3 , wherein said particular roll-up-bucket is determined based on said bucket timestamp associated with said particular initial-bucket.

5. The method of claim 1 , further comprising:

storing a plurality of second roll-up buckets associated with a third time interval, wherein the first and second time intervals are contained within the third time interval;

wherein the particular roll-up bucket is associated with a sub-time interval of said second time interval;

wherein said particular roll-up bucket has a roll-up bucket count that indicates a number of one or more previously received transactions that occurred during said sub-time interval of said second time interval; and

in response to the computer determining that a time indicated by said transaction timestamp is not within said sub-time interval of said second time interval, the computer adding the roll-up bucket count for the particular roll-up bucket to a bucket count of a roll-up bucket of said plurality of second roll-up buckets.

6. The method of claim 3 , further comprising:

formatting said bucket timestamp to produce a formatted bucket timestamp; and

wherein said particular roll-up bucket is determined based on said formatted bucket timestamp.

7. The method of claim 1 , further comprising:

setting said bucket count of said particular-initial bucket to 1 so that said bucket count of said particular initial bucket no longer indicates a number of one or more previously received transactions that occurred during said sub-time interval of said first time interval.

8. The method of claim 3 , further comprising:

setting said bucket timestamp associated with said particular initial-bucket to a time that is after than the end of said first time interval.

9. The method of claim 1 , further comprising:

formatting the transaction timestamp to generate a formatted transaction time-stamp;

wherein said particular initial-bucket is associated with a bucket timestamp; and

wherein determining that a time indicated by said transaction timestamp is not within said sub-time interval of said first time interval includes comparing said formatted transaction timestamp to said bucket timestamp.

10. The method of claim 1 , wherein said plurality of initial-buckets and said plurality of roll-up buckets store a plurality of transactions that include said transaction, and wherein each transaction of said plurality of transactions is concurrently stored in no more than one bucket of said plurality of initial-buckets and said plurality of roll-up buckets.

11. The method of claim 1 , further comprising:

classifying a transaction as a potentially fraudulent transaction.

12. The method of claim 11 , wherein classifying a transaction as a potentially fraudulent transaction comprises performing rule-based fraud detection based on the counts stored in the initial-buckets and roll-up buckets.