You are on page 1of 6

JULY 2021

Real-Time Risk
Management in Practice
The Experts’ Views
REAL-TIME RISK MANAGEMENT IN PRACTICE: THE EXPERTS’ VIEWS JULY 2021

Reliable, accurate and scalable risk management systems are the


lifeblood for financial institutions and traders alike. The need for
ever-more cutting-edge solutions is being driven by an inescapable triple This paper was derived from
whammy of regulation, cost cutting and market volatility. Not only is this the webinar, Real-Time Risk
driving up automation through increased use of AI, machine learning and Management: Uses and
robotic process automation, but it is also leading to widespread adoption
Benefits for Risk Managers.
of real time or near real-time capabilities for risk managers. Historically
the preserve of traders and desk-side risk managers, enterprise level If you missed the live
real-time risk management is now both possible and affordable thanks broadcast, you can access
to technology innovations that are pushing the boundaries in terms of
the webinar here. You will
functionality and cost.
learn the views and insights of
But it is one thing to talk about real-time risk management in a theoretical
market experts as they discuss
sense, it is quite another to talk about it in practice.
real-time risk management in
Recently, a virtual roundtable discussion between a group of market
actual practice.
experts, titled Real-Time Risk Management: Uses and Benefits for Risk
Managers, which was hosted by CubeLogic in partnership with Numerix
and PRMIA, examined various applications of risk management in
real time. Here, we share an abridged version of that discussion. As you
consider the key takeaways, we encourage you to reflect on your own
real-time risk management experiences and needs.

Defining Real Time: It Depends


Real time can mean different things in the context of business needs. Can you explain how?

It depends on your objectives and the time scale you are targeting. For example, you can have near real-time VaR
calculations performed initially on your enterprise portfolio at market opening and then, as those positions are updated
throughout the day, the P&L vectors can be reused to refresh your VaR calculations almost instantaneously. When new
positions or instruments are added that require new P&L vectors to be calculated, this can be done near real time in
a matter of minutes. This provides a powerful framework for continuous refreshes of your portfolio VaR throughout the
trading day rather than relying on stale end-of-day reports. It also enables real-time “what-if” analysis to be performed
before a position is added and the implementation of effective intraday limit controls.

The selection of the algorithm and the appropriate methodology for VaR becomes very important in a near real-time
context. Let’s say you are trying to do a full 20,000 path Monte Carlo simulation, then doing incremental intraday
is possible but is likely to be too slow for near real-time application. On the other hand, historical or parametric VaR
are both very effective for this purpose, allowing incremental and even full updates in minutes.

There are also a number of practical things that can be done to pre-process key inputs. For example, basic position
information and even Greeks can be pre-calculated in seconds or less, after which you can get down to things that
are really time critical, such as real-time credit checks as part of the flow of order placement or execution, which needs
to happen in milliseconds.

One thing that needs mentioning is that once you step outside of technology considerations, there is another set of
challenges to achieving real-time risk calculations. In some market segments, such as structured products and the
commodities sector, determining what trades should be in your portfolio at any one time can become quite a challenge

NUMERIX 2
REAL-TIME RISK MANAGEMENT IN PRACTICE: THE EXPERTS’ VIEWS JULY 2021

as negotiating deals can take days. At what point are you comfortable enough to include those trades in your portfolio
for the purposes of “what-if” analysis? Those are not decisions that can be made at a sub-second level, but in the order
of minutes or hours. Therefore, the goal may not be the sub-second metrics, it might be an update every half an hour.
In other words (and, unsurprisingly), the functional requirements of the business unit will dictate the risk management
related technology choices.

What is clear is that we are seeing an increasing trend towards firms deploying intraday risk, and if near real time is
not possible for practical reasons, then something that is every hour, half an hour or fifteen minutes is seen as being
superior to the traditional end-of-day approach. In conclusion, real time doesn’t necessarily need to be real real time as
needed on a trading desk, but something that has been updated in the last hour or two can make all the difference in
fast moving markets. The move from traditional end-of-day reporting enables firms to react in quicker fashion to market
moves, better monitor the related risks and could provide a leading edge vis-à-vis their peers.

Real-Time Risk:
Enterprise Level Versus Trader’s Perspective
Enterprise risk management aims to identify, assess, and control any threats and dangers
to an organization—with the ultimate goal of supporting business strategy and driving
value and success at the enterprise level.

But there is a big difference between real-time enterprise risk management and real-time
decision making at the trading level. What is real time to a risk manager in an enterprise
context and how does that differ from a trader’s view of real time, for whom real time is
a minimum standard requirement?

Traders are constantly monitoring and adjusting the risk in their books and, specifically, for the products and asset
classes in their portfolios. On the other hand, from an enterprise risk management perspective, risk managers are
concerned with risks and threats on a more holistic basis across all businesses, clients, counterparties, geographies,
legal entities, and so forth.

NUMERIX 3
REAL-TIME RISK MANAGEMENT IN PRACTICE: THE EXPERTS’ VIEWS JULY 2021

Let’s take an example in the prime brokerage world and a holistic risk appetite framework setting out a matrix of credit,
market and liquidity risk limits. For that appetite to be utilized and controlled, risk managers need to have visibility
on each client’s multi-asset portfolio of exposures, collateral and haircuts in real time in order to monitor risk levels,
including default-contingent risks, and to manage intraday margin calls. As part of this framework, liquidity also needs
to be monitored in real time as this will be a critical assumption embedded within the enterprise’s risk methodology.

In summary, these are the key differences. From an enterprise risk perspective, risk is being looked at on a holistic basis,
whereas a trader is primarily looking at market risks in their specific books.

Finally, intraday risk management is now becoming far more important because of trading on electronic platforms, where
you need to have real-time risk management tools, metrics, and triggers. As markets move much faster, both the trading
and risk management functions need to react more quickly, and this requires a capable technology to support them in
this quest.

This technology must include the ability to determine in real time if a client has sufficient collateral or credit limits before
accepting a trade. This requires both visibility of the amount of collateral in client accounts as well as measuring how
much more collateral clients may have to post depending on the trade or strategy they are executing. Where there is
a shortfall in collateral, there is a need to go to the counterparty and say, “I need more collateral in order to continue
taking your trades”—all in real time.

Real-time Metrics: Requirements Differ


Depending on the type of trading business an institution is engaged in, the specific real-
time risk metrics that are considered most important can vary from business to business.
Let’s use the examples of three distinct sectors—prime brokerage, crypto and commodities.
What are the top metrics that are important in a real-time context for these segments?

From a prime brokerage perspective, it’s about


intraday collateral health and maintaining an adequate
margin buffer given the volatility and liquidity of the
portfolio. Accordingly, real-time risk management
and aggregation is key. Within that, different risk
categories need their own tools. For counterparty
risk, the key real-time risk management tools focus
on exposures, stresses, collateral and liquidity to
ensure the risk to the client remains within approved
risk appetite. This is not simple and requires multiple
system linkages and even real-time visibility on
operational issues such as disputed margin calls to
ensure these feed into your risk limit utilization.

For the cryptocurrency market, not only is real-time


risk critical but, as the markets trade literally 24/7,
risk system downtime cannot be tolerated. This has
a huge impact on IT architecture design, as well
as on performance and resilience requirements—
a world away from traditional end-of-day or even
intraday risk management tools.

NUMERIX 4
REAL-TIME RISK MANAGEMENT IN PRACTICE: THE EXPERTS’ VIEWS JULY 2021

The most important real-time metrics are monitoring the exposure against collateral and the crypto price. Crypto
prices can move significantly in a matter of minutes or even seconds. This requires additional attention to metrics such
as volatility and volatility pick ups to ensure sufficient anticipation of large intraday price swings, especially before the
weekends or during the time of the day when liquidity is in shorter availability.

Liquidity metrics are also critical, as liquidity itself is highly volatile in the cryptocurrency markets and comes with a big
spread in this asset class. It can be extremely challenging in times of stress to find sufficient liquidity to execute large
close-outs, especially at night and over weekends. This requires continuous real-time testing of liquidity on your trading
venues, which highlights another challenge in this asset class. Cryptocurrencies are traded in a decentralized way
across a large number of venues. Consequently, liquidity is highly fragmented and venue specific, requiring a focus on
the major venues to ensure you have enough liquidity in each marketplace. Also, this fragmented nature may mean that
you need to carefully manage large orders by splitting them up across different venues.

From a commodity risk manager’s perspective, the most important real-time metrics are the simple ones. In the physical
commodity world, keeping track of your positions, where your inventory is stored and whether your shipments slipped
to a different month are all critical. Ensuring that these inputs are accurate forms a critical basis for your hedging
activities and is therefore important for VaR and any other risk metrics. They are, in fact, quite simple requirements but
still hard to achieve—and can mean the difference between trades becoming a real problem or being managed well.

Data and Other Challenges


The importance of data analytics has grown exponentially over the past several years, and
there is no question that risk management is increasingly dependent on data. What are your
views regarding the relationship between data and real time?

Inevitably, there is a heavy lifting process for marshalling data to understand the firm’s positions and to run the
real-time risk calculations. However, from a risk manager’s perspective, the main challenge is information overload.
For that reason, it is important to keep things simple, such as having a dashboard with, say, five key metrics and other
indicators you are managing on a real-time basis. No one has time to look at multiple reports, especially when they
could contain a number of false triggers. That requires clarity of thought on what you are trying to achieve in using real-
time metrics and data. For example, what are the ones you should act upon? How do you separate the noise from the
true information without that in itself becoming a time-consuming activity? Here, AI and machine learning can be vital
tools, such as using machine learning cluster analysis models that learn to identify anomalies in large and ever-changing
data sets.

Another type of challenge is data quality. You need some very robust processes intraday to make sure that the
information you are seeing is coherent. Your real-time risk is only as good as your worst data source/data point. Again,
this is where the trend towards machine learning and AI models for identification of data anomalies in vast datasets is
helping firms achieve their real-time risk ambitions.

NUMERIX 5
www.numerix.com www.cubelogic.com

You might also like