Abstract-Financial botnets, those specifically aimed at carrying out financial fraud, represent a well-known threat for banking institutions all around the globe. Unfortunately, these malicious networks are responsible for huge economic losses or for conducting money laundering operations. Contrary to DDoS and spam malware, the stealthy nature of financial botnets requires new techniques and novel research in order to detect, analyze and even to take them down. This paper presents a work-in-progress research aimed at creating a system able to mitigate the financial botnet problem. The proposed system is based on a novel architecture that has been validated by one of the biggest savings banks in Spain.Based on previous experiences with two of the proposed architecture building blocks -the Dorothy framework and a blacklistbased IP reputation system-, we show that it is feasible to map financial botnet networks and to provide a non-deterministic score to its associated zombies. The proposed architecture also promotes intelligence information sharing with involved parties such as law enforcement authorities, ISPs and financial institutions.Our belief is that these functionalities will prove very useful to fight financial cybercrime.

Every day, hundreds or even thousands of computers are infected with financial malware (i.e. Zeus) that forces them to become zombies or drones, capable of joining massive financial botnets that can be hired by well-organized cybercriminals in order to steal online banking customers' credentials. Despite the fact that detection and mitigation mechanisms for spam and DDoS-related botnets have been widely researched and developed, it is true that the passive nature (i.e. low network traffic, fewer connections) of financial botnets greatly hinder their countermeasures. Therefore, cyber-criminals are still obtaining high economical profits at relatively low risk with financial botnets.In this paper we propose the use of publicly available IP blacklists to detect both drones and Command & Control nodes that are part of financial botnets. To prove this hypothesis we have developed a formal framework capable of evaluating the quality of a blacklist by comparing it versus a baseline and taking into account different metrics.The contributed framework has been tested with approximately 500 million IP addresses, retrieved during a one-month period from seven different well-known blacklist providers. Our experimental results showed that these IP blacklists are able to detect both drones and C&C related with the Zeus botnet and most important, that it is possible to assign different quality scores to each blacklist based on our metrics.Finally, we introduce the basics of a high-performance IP reputation system that uses the previously obtained blacklists' quality scores, in order to reply almost in real-time whether a certain IP is a member of a financial botnet or not. Our belief is that such a system can be easily integrated into e-banking anti-fraud systems.