Pre-trade risk – a unified approach for various trading infrastructures Vladimir Kurlyandchik Director of business development arqa.ru arqa.ru Sources of financial errors arqa.ru Key factors to succeed understanding the client’s trading model mathematical tools of risk management taking into account peculiarities of particular markets understanding of various infrastructure options arqa.ru Risk server requirements control by broker multi venues various risk control models multi asset, multi currency and portfolios with different settlement conditions fast and scalable arqa.ru Risk server Control by broker Trading Venue Risk Server Broker’s Infrastructure Client arqa.ru Risk server Multi venues arbitrage netting obligations higher volumes arqa.ru Risk server Various risk control models arqa.ru Risk server Assets, currencies, settlement conditions arqa.ru Risk server Fast and scalable consecutive calculation number of clients number of transactions drop-copy processing arqa.ru Risk calculation approaches Genuine pre-trade Client Trading Venue Pre-trade control Risk Server arqa.ru Risk calculation approaches Fast pre-trade Client Trading Venue Pre-trade Control Module yes / no Risk Server arqa.ru Latency defines approach Client Trading Venue Pre-trade control Trading Venue Pre-trade Control Module Risk Server Market data latency 2-3 ms risk server at co-location ≤5 ms risk server in broker’s data center Risk Server Client arqa.ru Pre-trade & Algo Trading Venue Pre-trade Control Module Algo Engine Clients yes no Risk Server arqa.ru Pre-trade & OMS Broker’s OMS Trading Venue FIXPreTrade FIX FIX yes / no Risk Server FIX arqa.ru Risk awareness of clients monitoring risks what-if calculator closing positions arqa.ru Case study: Retail 150 000 clients’ accounts 400 000 positions in cash and equities 40 000 positions in derivatives 0.9 - 5 ms pre-trade check latency 0.4 ms position adjustment after trade ~10 mln processed transactions per day ~1.5 mln processed trades per day arqa.ru Case study: Genuine for HFT 15 mln transactions per day 30% of all transactions at Stock market of Moscow exchange 450 transactions per second - average speed 800 transactions per second - peak speed ≤5.6 ms average transaction latency 2-3 ms risk server overhead arqa.ru Case study: Buy-side Broker A Buy-side Risk Server Risk Server Broker B Trading Venues Risk Server arqa.ru Case study: Fast pre-trade for HFT Client Trading Venue FIX2Market Exchange protocol FIX yes / no HFT Engine Market data and client operations Co-location Risk Server ≤30 mcs checks per transaction arqa.ru Case study: Collaboration with exchange Moscow Exchange Workstation Trading Venue MICEX bridge Exchange API Market data and client operations Exchange co-location yes / no 3 mcs checks per transaction HFT Engine Risk Server arqa.ru Three actionable takeaways 1. Effective risk control starts with a trading pattern, employs mathematic models and appropriate infrastructure options while taking into account market peculiarities. 2. A fast and scalable risk server must be controlled by the broker. The server applies various risk control models simultaneously to cope with multiple instruments and portfolios with different settlement conditions at many trading venues. 3. The choice between full pre-trade and fast pre-trade is dictated by the latency requirement of a particular trading technique (algo, OMS, retail, buy side, HFT). There are options to implement risk controls for maximum benefits. arqa.ru Thank You! Questions?