Pirimid Fintech

BOT Trader Pattern Detection AI Model

There are thousands of automated traders (bots, strategies) triggering millions of transactions globally on a daily basis. We started out this project of building AI models to take few known bots and reverse engineer through machine learning to predict their actions I.e. buy, sell or idle. For initial phase, we only focused on BOTS doing EUR/USD forex transactions. Models are built to predict if BOT trader would proceed with transaction in next time interval given the history. Thus our models solve classification problem for three classes: idle (no transaction), buy and sell. These type of models are really beneficial for sell sides trying to figure out buys side's trading behaviour.