Spam comments are common on platforms like YouTube and are often used to scam users through spam campaigns masquerading as investment advice from so-called experts, particularly in finance-themed videos. This study aimed to develop and evaluate a Multinomial Naive Bayes model to classify spam comments. For this, the approach used defined rules for classifying spam, created a dataset of over 25,000 comments from 30 finance-related YouTube videos, and generated and evaluated a proposed model using assertiveness metrics to validate its performance. The results showed that the proposed model classified Spanish-language spam comments in finance videos with 98% Recall rate, indicating strong performance based on standard metrics. In conclusion, a viable alternative to reduce spam comments was proposed that could be incorporated into a web tool.