Friday, April 26, 2013

"Big Data" in Banking

Image courtesy of Blogging4Jobs

The notion of "Big Data" has been circulating for quite sometime now. Nowadays, commercial and investment banks are more interested in acquiring, managing, and analyzing data so they can be better equip themselves for a more competitive and risky environment. Big data analytics require high-skilled scientists and personnel, as well as very powerful large machines, to manage large-sized data sets. Many think that big data management is one of the major trends in the financial services industry (they are absolutely right!). To gain competitive edge as well as to mitigate risk, banks need to allocate time, money, and energy to this area.  A lot of talk has been surfacing in the news about how banks are managing risk. Some banks would like to see a convergence between audit and risk management  others still prefer to have autonomous departments. But the truth is, data management can be the key solution for risk management going forward. Managing data through a pure scientific approach (what is called Data Science) , could provide a better way for creating different scenarios for different group of customers and asset classes.



The attached video might sound a little bit technical and doesn't deliver a clean message to people not familiar with the subject matter, so I will try to summarize it in few sentences.  The main takeaways are (1) banks have to take advatnage of open-source data by creating data centers or through outsourcing (2) working with regulators and government officials will be key in acting, and possibly commoditizing, collected data sets  (3) embracing cloud sourcing (4) paying more attention to data as a service that can be acquired from third-party providers.

With all that being said, banks are among those privileged to have easy access to data through various routes. Commercial banks have enormous data from their customers (mortgages, credit cards, loans, deposits, salaries, etc), treasury operations with the interbank, wealthy individuals and corporate customers. So these banks are sitting on a big sets of data mines to the extent they don't know what do with this influx of information. According to the attached graph below, financial and insurance companies score HIGH on the easy of acquiring data and HIGH on the value derived from big data management. According to a study by "The Digital Universe", available unprotected data comprise 80% of all data circling the internet and other networks  but less than 1% of the world's data is analyzed and managed. The report also highlights the increasing fear of data insecurity because the rate of data creation is unproportionately related to data protected; the study suggested that 35% of the analyzed data is seriously endangered and unprotected.

Photo acquired courtesy of Chuck's Blog

The traditional approach of using data to mitigate risk entailed the following:
  • Creation of large warehouses that imported data using a specific format  
  • Constant recalculation of data to reach a good estimation
  • Employment of different risk systems and different formats for each
  • Optimization is needed for each format 
  • Lack of flexibility inability to accomodate different risk management systems, which means there is alway a need to scale up and create new warehouses. 

Problems that the proper usage of Big Data can help fix includes: 
  • Storing, analyzing, and managing data in one large system rather than multiple silos. 
  • The occurrence of data analytics without the need for optimizations (system and formate incompatibly).
  • Using the data to analyze customers/investors trends and potential risks. Traditionally, systems offered breadth but not depth of analysis. Big data management can tackle both areas more effectively. 
  • Hiring high skilled data scientists, and people with a knack for analytics, is a prerequisite to risk management and data management in general. 
  • Leveraging data to drive innovation.
According to the article Big Data: The Management Revolution, there are five main challenges management executive (including those in the financial services) will face to make a successful switch to big data management and make the best use out of it: (1) Top executive need to adapt and adopt the data-driven decision making mentality, (2) Talent acquisition, (3) Having the necessary technology, (4) Decision making should encompass multi-functions and cross departments, and (5) Company-culture adoption.  Banks should widely embrace and accept Big Data management and analytics as a source of innovation and an essential part of running their business. The traditional model of storing and using information isn't practical anymore, rather banker need to have simpler "cleaner" solutions at their disposal to make decisions.

-FM

Wednesday, April 24, 2013

Trading in the Era of High-Speed Algorithms (Part II)



In light of the recent incident that rattled the S&P 500 and the Dow Jones Industrial Average that involved the Associated Press (AP), I decided to write a sequel to the previous post. I would like to highlight how algorithmic trading (algo trading) computers can exacerbating outcomes. Beyond the many obvious advantages of algo trading, trading platforms (and machines in general) might fail sometimes (duh, obviously!). Due to the high interconnectedness of trading platforms, such as the one used by the, the results can be devastating.

Tweet #1

Tweet #2

On April 23 2013, a false tweet (check tweet #1) by AP viraled out and rattled the S&P 500 index, where AP's stock is traded. On the outset, the whole incident can be attributed to cyber security but because we live in a ubiquitous and seamless digital environment, instantaneously, high-frequency traders jumped on and acted on the "false" tweet. The algorithms was triggered by unusual volatitlity that took place in a span of 2 minutes. The false news about the explosions in the white house created a panic in major stock markets in the United States. In this seemingly short amount of time (2 minutes), the S&P 500 lost $136 billion. Although the market had recovered its loses in the next 3 minutes, analysts think that many orderes might have gotten stuck or even that some foul play might have occurred.  Algo trading wasen't the cause of this ripple effect, but it made it worse by executing automatic major sellouts (orders that were previously placed in the systems).  One of the options that algo trading is famous for is called "stop-loss" in which the system will automatically start selling your stocks if the price falls below a certain pre-indicated level. As you would imagine,  after AP's tweet, programmed computers started executing thousands of stop loses in a short span of time. What is even more scary is that the algo machines acted on the news with very minimal (or even no) human supervision, at least not during those 2 minutes. 




                                                 AP Twitter Hack Highlights Vulnerability of Markets

According to the attached video from Bloomberg TV(via You Tube), Stephen Ehrlich of Lightspeed Financial says that the main problem with these algo trading machines was that they picked up the tweet feeds and acted on their own. This shows how vulnerable and frigle are these systems to any bogus information thrown at them. During this incident, thankfully it went by with minimal damage (or at least, this remains to be seen) but what would happen if things would go out of control and started causing irreversible damages!

This incident shows how interwoven and thin are social networks, their affect is felt even beyond their direct platforms due to over-connectivity and dissemination of information. This was a clear example of how powerful is the impact of ICT  on not just media and social networks but also on the behavior of programmed machines that is capable of bringing a whole financial industry to halt. Another issue is clearly cyber security which many high-frequency traders and hedge fund managers aren't big fans of because they thrive on the high-volatility in markets. Having tight security systems would restrict how algo trading operates and might weakens its whole purpose.  

-FM

Monday, April 22, 2013

Trading in the Era of High-Speed Algorithms (Part I)


http://davidbrin.wordpress.com/2011/12/10/gingrich-asimov-and-the-computer-trading-monster/
Automated Trading 
http://davidbrin.wordpress.com/2011/12/10/gingrich-asimov-and-the-computer-trading-monster/

In the previous post, I gave the example of scientists and statisticians looking at “Big Data” from Twitter to predict trading and investment opportunities. Well, this convergance between tech and finance is not new and can be manifested in different forms. To continue our discussion on the impact of ICT on financial services, let us look at algorithmic trading (algo trading). In this day and age of Hi-Tech it has never been easier to trade securities. Note that I haven't I said that trading has become more efficient or safer, simply, it has become easier and, probably, cost-effective. With algo trading, an investor can specify her trading preferences and punch then into a trading platform (check photo). 

            My own trading page with NCB Capital
Down to the nitty gritty details, traders can add and modify their orders (trades) according to their needs and risk tolerance, including execution date and time, buy/sell price, quantity, price range, etc. This might sound commonsensical for you, but what if you have to place, trade, and monitor hundred of thousands of deals a day! Try to imagine how it would be like if you were a stock or currency trader getting orders willy nilly, to the extent that you are unable to organize your order books, let alone concentrate on your main task, that is trading.

Let us look at high-frequency trading (HFT) which is said to comprise the majority of equity and futures tradings. In 2009, HFT firms accounted for only 2% of all of the trading firms in the U.S but they dominated the equity market with 73% of the total trading volume. HFT is the result of a proprietary computer programming that enables firms to trade large-volume of equity (as well as a myriad of other securities) in a blink of an eye, literrally. High-speed trading is the name of the game and it is reported that about 19,000 deals can be performed in 1 second. Although people are still involved in the input/output of the decision making process, the impact of ICT (i.e automated trading) has changed the whole trading ecosystem. Thanks to the automated nature of trading, common investors as well as savvy hedge fund investors nowadays would rather speculate rather than invest long term. The trading behavior is shifting and the tech-finance convergence I mentioned earlier is fully manifested. AYan Ohayon in one of the TED Talks said (check the attached YouTube video) "The average holding time period at Wall street has shrunk to a mere 22 seconds". 


TED Talks: Algorithmic Trading and its Impact on Markets

With HFT, latency has been reduced greatly while computational power is rapidly increasing. Two obvious advantages of HFT passed to small investors are (1) added liquidity to the markets and (2) reduced trading costs. Another important aspect of trading algorithms, such as HFT, is their impact on how investors choose which stocks to buy and how to value firms. Hedge fund managers, CEOs, analysts, as well as main street investors started relying on computers to determine the "price" of a firm's stock. Finding the true value of a security has become relatively easy given the power of computational machines,  yet it is way more complex than that. Because automated platforms can run all sorts of trading averages, closing/opening prices, and financial ratios yet, as we all know, that market prices are subject to much more than the fundamental and technical analyses.

Automated trading has had a great impact on opening up the financial markets to small investors. Nowadays, basically anyone with a bank account can trade currencies (forex) or buy/sell stocks on an over-the-counter market or on a stock exchange, such as NASDAQ. According to the Man vs Machine infographic, the ease of use, anonymity, and reduced market layers, are among the top reasons behind algo trading increasing popularity. 


Man vs Machine Infographic
The future of trading will continue to change with the evolvement of high-speed computational platforms and machines. Which in turn is having a tremendous impact on the role of regulators, banks, financial brokerage houses, insurance companies, indexes, and technology providers. Reduced commission costs, higher traders productivity, low latency, information sharing, execusion consistency and customization are among many impacts information and communication technology (ICT) has had on the financial trading business.  

-FM