Echo and Compass help run B and A-books. Echo helps us identify the best way to manage counterparties, including by identifying the counterparties most appropriate for STPing or B-booking. With Compass we can also use the automated classifier that can move counterparties directly onto a different hedging regime, for example moving counterparties from the B to A book (this can have human oversight as required).

We can also run a bounded B-book, meaning we can set NOP and VaR limits and increase hedging activity as the portfolio approaches these limits. This allows the book's PnL to develop naturally, but with reduced volatility, so the PnL will tend to develop at a much more steady rate.

For everything not A or B book, we can manage in a separate “Compass book”. Predictive signals means we hold risk when risk is moving in our favour, or reduce risk when the market is not moving in our favour. We reduce market impact by using the DTA (Dynamic Time Algorithm) to intelligently source liquidity at the tightest spreads possible.

Hedging

Compass has 2 types of hedgers available to manage risk, both work to reduce equivalent positions (equivalent VaR of the portfolio) reducing the risk of losses and exposure to market movement:

1. The arbitrage hedger - looks for discrepancies in pricing to remove risk at a better rate. 
   • • The Arbitrage hedger can be set up to run in risk reducing and/or risk increasing mode.
       
       
2. The hybrid hedger - fully configurable hedger defined by a set of rules that are activated on a number of triggering criteria 
   • • The hedger is able to make decisions on how and and when to start reducing a position, controlling the speed and quantity of VaR reduction per second.
     • This means the hedger is able to flexibly execute trades based on the style of flow coming through the book or on predicted market movement. For example it is able to speed up or slow down execution based on trade size to avoid making market impact or slow the order execution when a price is expected mature in your favour or increase risk clearance rate if it’s going to go against you.
     • The hybrid hedger can also work alongside the automated classifier to dynamically execute on changing counterparty yield profiles. Meaning that if a current counterparties flow worsens or a new counterparty with toxic flow begins to trade in the book, it will be quickly picked up and hedged in the most efficient way possible. 
     • As part of the product we have a suite of predictive signals that are used for price prediction. We apply adaptive signal composition a method of composition using an online reinforcement learning and online recalibration to capture changes in market state.
     • Learn more about the predicates controlling the hybrid hedger here