Addressing hate speech on social media platforms is a pressing concern, particularly on platforms like Twitter and Facebook. To date, extensive research and numerous projects on the subject of hate speech detection and prevention have been conducted, as well as for crime detection and its prevention. However, there is a major void in the current research terrain whereby there exists no specific effort to study how hate speech and crime are related globally. This research work proposes a multi-series methodology to investigate the complex linkages between online hate speech and real world acts of violence. Multi series BERT model from transformer was used for detecting the hate speech in the tweets of popular politicians, celebrities, peoples tweets on popular cases and showing the linkage of these to the real world crime happenings. When compared with the other available popular models like CNN. KNN, SVM, RF, LR, NB, DT which are generally used for detecting hate speech and crime detection, the proposed BERT model gave great results with 93% in accuracy as well as 92% in F1 Score outperforming all the other models. Furthermore the results showed that states like Delhi, Maharashtra and Tamil Nadu have the higher chances of crime happening due to hate speech based on the prediction done. However the data may vary for different social media platforms as the current research work is only based on twitter.