A risk management system (RMS), in the trading sense, is the software layer that enforces risk limits on orders and positions. Every order passes through it before reaching the market, and it continuously recalculates exposure, margin and profit and loss as trades occur. If an order or a position would break a limit, the system rejects the order, restricts further trading or closes positions automatically.
Such systems are run by brokers, exchanges, clearing members and trading firms. A broker's RMS protects the broker from losses caused by its clients, while a trading firm's own RMS protects it from its traders and its algorithms. The term is especially common among brokers in India, where it also refers to the team that operates the system, and it should be distinguished from enterprise risk management in the wider corporate sense.
Pre-trade controls
Pre-trade checks are applied to each order in the instant before it is released. Typical controls include the following.
- Order size and value limits. A maximum quantity and notional value per order catches a fat finger error, such as an extra zero typed into a quantity.
- Price collars. Orders priced too far from the current market are rejected.
- Position and exposure limits. The order is refused if it would take the account beyond its permitted position in an instrument, sector or the whole portfolio.
- Margin and credit checks. The system confirms that the account holds enough funds or collateral to cover the initial margin on the resulting position.
- Message throttles. Limits on orders per second and on repeated identical orders stop a malfunctioning algorithm from flooding the market.
- Restricted lists. Trading is blocked in instruments that are banned, suspended or subject to regulatory limits such as a market-wide position limit.
Post-trade and real-time monitoring
Once trades are executed the system tracks the consequences. It marks positions to current prices, computes realized and unrealized profit and loss, and compares the result with daily loss limits. For derivatives portfolios it aggregates the option Greeks and may calculate value at risk or run stress scenarios.
When an account's losses erode its margin, the system issues a margin call or, at many retail brokers, automatically squares off positions once a threshold is crossed. Most systems include a kill switch, a single control that cancels all open orders and blocks new ones for a trader, a strategy or the whole firm.
Regulatory requirements
Automated controls are a legal obligation in the main markets. In the United States, the Securities and Exchange Commission adopted Rule 15c3-5, the Market Access Rule, in 2010. It requires brokers that provide access to exchanges to maintain pre-trade financial and regulatory risk controls under their direct and exclusive control, which ended the practice of unfiltered sponsored access.
In the European Union, MiFID II and its technical standards require firms engaged in algorithmic trading to have pre-trade controls, real-time monitoring, kill functionality and annual self-assessments, and they place similar duties on firms offering direct market access. In India, the Securities and Exchange Board of India and the exchanges prescribe margin collection, position limits and order-level checks that brokers must build into their systems. Exchanges add their own safeguards, such as price bands and a circuit breaker mechanism, as a further line of defence. These measures are collectively described as algorithmic trading controls.
Speed and design trade-offs
Every check adds delay, and for firms competing on low latency the cost is significant. Risk checks are therefore engineered to run in microseconds or less, sometimes on specialised hardware such as a field-programmable gate array. Designers must also decide how conservative to be: limits that are too tight reject legitimate orders at critical moments, and limits that are too loose defeat the purpose.
A second design question is independence. A control that sits inside the trading strategy can fail along with it, so regulators and good practice call for limits to be set and enforced by a function separate from the traders whose activity they constrain.
Failures and limits
The standard cautionary case is Knight Capital, a United States market maker that lost about $440 million in under an hour in August 2012 when faulty software sent a stream of unintended orders. The regulator's findings cited inadequate controls and procedures. Such incidents are a form of operational risk.
An RMS only enforces the limits it has been given. It cannot judge whether the limits are sensible, and it measures risk with models that may understate losses in fast or illiquid markets, where positions cannot be closed at the prices assumed. Automatic liquidation can also sell into a falling market and deepen the loss it was meant to contain.