Long gone are the halcyon days of multi-year big-budget waterfall-style projects serviced by traditional relational databases, overnight batch processing and monthly reports.
Today’s world is about immediate and global access to vast oceans of information, insights derived from the analysis of floods of data and the ability to quickly and effectively move to market and deliver value to our customers.
Financial services operate in this increasingly complex and demanding environment with escalating demands on security, performance and scalability. Regulatory requirements abound from a proliferation of financial regulatory authorities like the Financial Conduct Authority (FCA), Prudential Regulatory Authority (PRA) and Bank of England in the UK, the Security and Exchange Commission (SEC) and Federal Reserve (Fed) in the US and dozens more in the countries that we operate in around the world, spawning regulatory jargon like SOX, MAS, SCAP, and BASEL III.
In this talk we’ll look at the evolution of In-Memory Computing in Financial services and how it adds to and helps to address the challenges faced by large scale banking enterprises. We’ll also look a little bit ahead at emerging technologies and discuss the opportunities and challenges that they present.
16. MARKET PRESSURES: CHALLENGER BANKS
Fidor Bank, Atom Bank, Metro Bank, Starling
Digital First, Mobile First
No legacy technology
No physical presence
17. MARKET PRESSURES : FINTECH STARTUPS
Processing of Payments
- Paypal
- Apple Pay
- Samsung Pay
Investing and Financing
- Kickstarter, IndieGogo
18. REGULATORY PRESSURES: UK REGULATORS
Prudential Regulation Authority
Bank of England
Financial Conduct Authority
20. REGULATORY PRESSURES : THE WORLD
Monetary Authority of Singapore
European Securities and Market Authority
Hellenic Capital Market Commission
Fifty+ other countries
21. REGULATIONS
Sarbanes – Oxley
Markets in Financial Instruments Directive - MiFID
Dodd Frank
Protection of Client Assets and Money CASS
BASEL III
CFDC
Fundamental Review of Trading Book - FRTB
EMIR
Volker
FFIEC
DSS
PCI
22. GOALS OF REGULATION
Market Confidence
Financial Stability
Consumer Protection
Reduction of Financial Crime
23. TECH IMPACT
Large scale regulatory reporting
Massive risk and analytics processing
Consolidation of platforms, simplification
Faster deployment cycles
31. ROUND UP
Financial Services will never get simpler.
The amount of data being processed will never get smaller.
The speed at which data is gathered, analysed and reported
will never decrease.
BIG Data. FAST. In Memory.
Editor's Notes
Hi everyone, its great to be here in San Francisco.
Today we’re going to be talking about In Memory Computing for Financial Services.
I’m hoping to
show you a little of how financial services like banks use in memory computing,
to look at whats driving our world and the impact that’s having on the tech we’re using,
3. and to maybe peek ahead at what might be happening in the future
Lets start with a quick look at what we mean by Financial Services
There are a number of companies that are generally considered to belong under the umbrella of Financial Services, like
Accounting firms
Insurance companies
Stock brokerages
Investment Funds
Credit Card Companies
The one that springs to mind for most people, and the area I work in, is Banking. Lets take a closer look at that.
When most people think about banking, they’re probably thinking about Retail or Consumer banking.
This is the type of bank that deals directly with individuals and offers services like Savings and Current Accounts
Retail banks lend directly to the consumer with products like Mortgages and Personal Loans to help you buy high value items such as houses or vehicles or boats.
Retail banks typically also provide services like Credit and Debit card facilities.
In more recent times there’s been a growth in online capabilities, with banks expanding their banking facilities and offering you the ability to do all your banking on the go.
These types of banks are have a lot of customer data at their disposal, and are pretty advanced users of in memory computing. Simple use cases are things like simple data caching or web session sharing.
One more advanced application I’ve worked with uses an in-memory system as a collation point for all their consumer data, having several systems flow all their data into one single point to provide a holistic view of their customer.
Investment Banks are a different kettle of fish entirely.
Generally they’re split into three major sections: Front office, Middle office and Back Office
The Front Office is the “face” of the investment bank and usually has the following roles:
- Investment services
- help clients raise money through activities like floating on the stock market
- mergers and acquisitions
Sales and Trading
This is the area in which traders buy and sell products on behalf of both the bank and it customers.
We trade in products like:
- Equities, which are stocks traded either on an exchange or over the counter
Fixed income instruments like bonds
commodities like gold, silver, sugar, oil
money market instruments like mortgages, treasury bills and other forms of cash
We also trade in derivates of those products. These derive their value from an underlying instrument.
For example,
we have swaps - Give me your oil and I’ll give you some of my gold
options - the option to buy or sell at a given date
futures - buying and selling tomorrows products at todays prices
and lots of variations
On the research side, our analysts review companies and provide advice on whether a particular company is likely to do well or not and whether to buy or sell your holdings
The front office is again where we see fairly complex use of In memory computing. Trading systems feeding information to traders and updating them in real time as positions change as the market fluctuates require some pretty hefty computing power.
The middle office is the enabler for the Front Office activities and is where the big boys of In Memory computing play.
The major users being Risk calculations and Regulatory reporting.
This is where a seriously large amount of computing power is dedicated to the calculation of RISK
Risk comes in two main flavours: Market Risk and Credit Risk
Market Risk is the potential losses arising from movements in market prices.
This covers things like :
-- Equity risk: - what happens when the stock price changes?
-- Interest rate risk: what happens when interest rates change?
-- currency risk: what happens when foreign exchange rates change?
commodity risk: risk that the commodity price will change
Credit Risk is the risk that one of more of the parties might default on their debts
Calculations include the standard set of Greeks
– Vega, the sensitivity to volatility
- Theta, time decay, the amount of money each share of the underlying that the option loses per day
- Rho, the sensitivity to the interest rate
We also calculate Value at Risk or VaR using Monte Carlo Simulations, which is the worst case scenario on an investment over a given time period and with a given degree of confidence, and Stress testing which looks at possible losses with worsening market conditions – an increase in volatility, a sudden change in interest rates, unexpected dips in foreign exchange rates
We also model Partial Differential Equations such as Black Scholes for analysing the value of an option over time.
What we’re trying to achieve with all these calculations, aside from reporting to regulators is to decide how to adjust our market positions in line with our risk appetite.
Middle office is also where we’re looking at trade reporting, which we’ll touch on later
Back Office is where the plumbing of the company is.
This is where Accounting happens, where the Operations and Technology guys make sure the databases are working and the networks are in place and where functions like settlements, clearances and record maintenance happen.
Back Office systems use a surprising amount of in-memory tech. Finance and treasury are two areas that may not have the scale of the risk systems, but can certainly match them in terms of complexity.
Now that the crash course in Financial Services is over, lets look at how things worked in the past, what’s happening now, what’s coming down the pipeline and where technologies like In Memory Computing will come into play.
In the not too distant past, banking systems were a lot like systems in every other sector.
We saw a lot of monolithic systems that held their own silos of data close and didn’t communicate with each other.
This led to major inefficiencies like duplication of data.
In one example, trade data could be duplicated up to 12 times:
- through trade capture onto the legal department and Know your Customer checking
- then through front office risk management, the back office risk processing, then onto finance and accounting, then through payments and onto regulatory reporting.
We saw a growth in more and more complex products being offered, with a lot of exotic derivatives on the market
In terms of projects, long waterfall style projects spanning a year or two was the norm.
Things have moved on a little since then..
Now, let’s take a look at the forces shaping our world
Aside from our usual competition of other institutions, we’re seeing competition from two new areas.
One area that’s on the increase is in the form of Challenger Banks.
With changes in regulations and a lowering in capital requirements, the door was opened to allow new banks to set up. Before Metro Bank in 2013, the last banking license in the UK was issued 150 years ago.
Challenger banks don’t have the legacy of past technology and frequently don’t even have a physical presence.
These banks are born on the internet, they’re digital first, Mobile First or sometimes Mobile ONLY.
They move fast and iterate quickly.
Fin Tech startups look at ways to make financial services more efficient.
While not being fully fledged banks themselves, they look for inefficiencies and look to disrupt the industry with new technology based solutions.
Areas they look at include
processing of payments, for example using mobile payments like Paypal, Apple Pay, Samsung Pay, Android Pay
Investing and financing which is driving the growth of crowdfunding and new players like Kickstarter and Indiegogo are emerging
We’re also starting to see the likes of Google, Amazon and Virgin move into the financial industry. They can leverage their technology roots and their huge in-built customer base
They also move fast and iterate quickly.
These new smaller players who are agile and innovative are driving the same behaviours in the bigger players.
We now have to innovate, we now have to constantly look for ways tech can make us faster to market and give us that competitive advantage.
The other driving force we have is regulatory pressures. To make a massive understatement, banking is pretty tightly regulated
In the UK for example, we have the Prudential Regulation Authority (PRA), the Bank of England (BofE), the Financial Conduct Authority (FCA)
While in the US we have the SEC and the FED
Those are just two countries in which a major bank might operate.
We’re required to comply with regional regulations in all the countries in which we operate, so there are a number of other bodies such as MAS, the Monetary Authority of Singapore,
ESMA the European Securities and Market Authority ,
the Egyptian Financal Supervisory Authority in Egypt,
the Hellenic Capital Market Commission in Greece
and around fifty other countries each with their own regulatory requirements.
Obviously, not all regulators have the same requirements and a lot of times they may conflict, or may not be feasible to implement.
So we see a lot of negotiation with the regulators to try to find a happy medium.
These regulators produce an amazing array of regulatory requirements.
MiFID for example, covers pre- and post trade transparency and requires us to publish the price, volume and time of all trades in listed shares and to take reasonable steps to obtain the best possible result in the execution of an order
Basel III focuses on the risk of a run on the bank, requiring different levels of reserves for different bank deposits and borrowings
Fundamental Review of the Trading Book (aka Basel IV) is a series of proposals by the Basel Committee and is intended to harmonise the way market risk is treated and will probably result in higher global capital requirements and a major increase in the amount of data we gather and the analysis and reporting we do on that data
Ultimately the goals of regulators are:
Market Confidence
Financial Stability -
Consumer Protection – securing an appropriate degree of protection for consumers
Reduction of financial Crime such as Money Laundering, Corruption, the financing of terrorism and rogue trading practices
These usually take the form of an increase in the amount of operating capital required,
changes to the way the organisation is structured and the way it operates,
and almost ALWAYS an increase in the amount of data gathered, analysed and reported.
These two forces drive the way the business works and ultimately the technology that powers it.
So, we’re seeing a major increase in the amount of regulatory reporting and the speed at which the data needs to be processed.
This in turn is driving a consolidation of platforms and a tearing down of silos.
Data MUST be shared, both for efficiency and to ensure that we’re reporting consistently.
This is coupled with a shift towards fast iterations and agile projects.
One major computing platform reporting an increase from 40 deployments a week to 400 deployments a week over a period of about 5 years.
In terms of architecture, we still see some applications following classic n-tier architectures.
A number of other applications are moving to scale-out architectures to deal with their Big Data issues. For example, our High Performance Computing Grid processes around 100TB of data and performs around 290 years worth of calculations per day. They have a tremendous BIG and FAST Data challenge.
A lot of applications are meeting their performance needs by adopting tech like In Memory Data Grids, usually first by using it as a simple cache but then using more sophisticated features like parallel querying and aggregation of data in memory.
Some more advanced applications have moved to a pure in-memory model, dropping the database entirely and relying on horizontal scalability and resilience to handle their persistence and non functional requirements.
There are still a lot of physical servers being used, but we’re seeing a major shift towards virtualisation to make better use of the hardware.
So, that’s the current state of play. Peering a little into the crystal ball, what might be happening in the future?
Cloud will be everywhere. We’re already seeing this
Financial services aren’t in the infrastruture or data center business, so it makes sense to outsource to companies that are.
Infrastructure as a service allows companies to scale out to meet compute demands only when they need to, which cuts down on inefficient hardware usage and minimises the need to own and maintain our own firewalls, networks, servers and data centres
Platform as a service in turn allows application dev teams to concern themselves with adding value to the business and to focus less on running their own hosts, web servers, application servers, their own databases.
As a nice bonus, private cloud, public cloud or hybrid cloud models gives us the power to track actual usage of resources and to think about internal charging models
Cloud comes a with a host of its own challenges :
regulators are at different levels of sophistication when it comes to cloud so their regulations differ dramatically
there are security implications of hosting confidential information like credit card details
some code like the way we do pricing are proprietary and are considered confidential.
and there’s always the performance implications of processing data on a virtualised environment outside of your direct control
Next up : Data Science and Machine Learning.
With the advent of Big Data and the availability of cheap compute we’re seeing a growth spurt in these kinds of tools for financial services
Techniques like Supervised learning, unsupervised learning and reinforcement learning are things that might allow us to improve areas like
Fraud Detection
Stock prediction. Will a particular stock do well or poorly?
We can combine whats happing on the markets with sentiment analysis from social media like Twitter to get a broader insights into how stocks might perform
Market trend analysis. We now have a large amount of data around market crash conditions. Being able to replay those conditions might allow us to predict and avoid the next crash?
Block Chain is a cryptographic ledger technology probably best known as the basis for BitCoin.
Blockchain technology holds the promise of removing the middleman in a lot of industries – including Finance.
There’s a lot of interest in this both inside banking and in the Fintech startups. New Cryptocurrencies are popping up all the time – DogeCoin, RoboCoin, Coinye, Ethereum, Ripple to name a few.
With BlockChain, in the future we might see transactions times drop dramatically
Non Volatile Memory or NVDIMM is something we’re hearing a lot about recently.
Simply put, memory retains state when power goes down.
We’re already seeing a convergence of Relational databases, NoSQL, Big Data and In Memory Data Grids. As traditional spinning disk becomes less relevant for persistence, this convergence can only accelerate.
Artificial Intelligence – using AI like IBM Watson to augment our intelligence, and be able to process more contextual data; the use of Genetic Algorithms and Evolutionary Computation to attempt to model and learn from the stock market.
No matter the future tech, be it Cloud, Machine Learning, Blockchain or AI, they all have one thing in common: MORE
More computing power, More Data, More speed. More and more in memory.
We’ve seen a glimpse of how financial institutions work and how they use in memory computing,
what the market and regulatory forces are which drive their behaviour, and a peek at what might be happening in the future.