Fraudulent crimes have been a major concern in criminal patterns for a long time. Investment fraud, in particular, has become increasingly rampant in recent years, since it causes substantial financial losses in a single case. Fraud gang can gain high amount of proceeds of fraud crime with low physical harm and time expenditure. This type of fraud is often perpetrated through online channels, in the virtual world, so that the law enforcement agencies are difficult to investigate, trace, identify, and catch the fraud gang. One of the challenges in combating online investment scams is the ability to identify malicious websites. Domain names of fraudulent websites are often changed frequently and even sometimes shut-downed for preventing from law enforcement detection. In this paper, we propose a text analysis algorithm for identifying domain names of fraudulent investment websites. The proposed algorithm extracts a set of linguistic features from domain names, such as length, number of hyphens, number of digits, presence of keywords associated with investment fraud, and presence of typos or grammatical errors. The algorithm then uses these features to train a machine learning model to classify domain names as malicious or legitimate. The proposed text analysis algorithm provides a promising approach for law enforcement agencies to proactively combat online investment scams. The algorithm can be used to inspect domain names and flag suspected fraudulent domains, enabling the authorities to take timely action to block these websites and protect potential victims.