EagleAi™ TradeWatch Trading Risk module is an ensemble of AI engines that observes trading transactions and learns to identify outliers that matters for risk management. The advanced AI engines powering EagleAi Trading Risk modules have been designed to detect issues to prevent loss in trading due to operational issues and regulatory fines.
EagleAi monitors the patterns of trading behaviors of different entities involved in trading such as the clients, traders, desk, symbol and even the system that is used for trading. Factors such as gross notional value, net notional value, market exposures such as realized and unrealized pnl are analyzed as a time series for detecting any anomalies.
Trading Risk anomalies :
Anomalous trading: Insert / Update of an order at odd time for
Anomalous trading: Insert / Update of an order at odd times
from the user’s pattern
Anomalous trading: Trader trading a new product
Unusual tagging of Order Capacity
Large order quantity – Liquidity check
Large price away – PriceAway check
Large sized order – Notional Value
Unusual Number/NV of Short Positions
EagleAi detects various risky anomalies for different asset classes such as stocks, ETFs, options, bonds and FX.
EagleAi has a very high accuracy rate thanks to its ensemble of AI algorithms employed to do its job. With its relying on multiple different approaches to confirm an issue before it is raised as an alert, EagleAi is highly accurate in its identification of issues.
EagleAi uses innovative methods to triangulate across different models to be able to identify the exact problem such as the specific client, trader, system name etc. that would give enough information to fix any problem instantaneously. There is no need to second guess with issues identified by EagleAi. EagleAi’s models are adaptive to changing trends as it constantly updates its models and hence reduces false positives significantly. EagleAi TradeWatch computes an anomaly score for every event. Alerts can be custom configured to alert at various levels for different models.