The Algorithmic Arms Race: How will it all pan out?

comments 0

Comment

share

Share

0

Rate

Yuan Jun Lim's picture

The Algorithmic Arms Race: How will it all pan out?

‘Algorithmic Trading’ is the new buzz phrase in the trading world. It seems like nobody can get enough of it. Firms of all sizes partake in it, hoping that through it, the performance of their funds can outdo the others and beat the competition. Granted, it is generally perceived that pursuing advancements in this area may prove to be revenue-enhancing. Indeed, there were healthy profits as evidenced by Winton Capital, a London-based algorithmic-driven hedge fund which has returned 14.8% a year over the past decade. Its fund is currently worth $29 billion, one of the largest in the UK. 1 All signs seem to indicate that algorithmic trading will fundamentally alter the trading paradigm; it is destined to be a game-changer. Some may argue that in effect, it has already done so. Consequently, firms are actively upgrading their algorithmic trading capabilities, resulting in an ‘arms race’ of sorts. It is thus worth considering the strategies that will prove to be most viable in the near future and the associated regulatory challenges. Then, what implications can we realistically expect from the arms race; how will it all play out?

Let us first consider some of the different trading strategies, before examining how these can be translated into algorithms. These include trend-following, mean reversion and arbitrage tactics. Trend-following applies for both buy and sell side markets, with differing time frames. Mean reversion incorporates the idea that prices will revert to the average, over a period of time. Arbitrage rides on the concept of potentially identifying risk-free profits by capitalizing on price mismatches of the same product in different markets. This phenomenon is often brought about by the existence of imperfect information in the markets. Arbitrage lends itself nicely to High-Frequency Trading (HFT), a subset of algorithmic trading. Market-making often goes hand in hand with arbitrage in HFT. Traditionally, market-makers termed ‘specialists’ were designated for each product. However, trading firms are now behaving like market-makers with the proliferation of HFT strategies.2 HFT provides a greater amount of liquidity and price discovery to the markets by means of utilizing market-making and arbitrage strategies. Hence, market-making is no longer just confined to the big firms; companies of a smaller scope can assume the role too, with systems supported by algorithmic HFT strategies. In turn, the increased liquidity causes lower spreads, indirectly reducing trading costs for investors.

Then, technological advancements have virtually guaranteed the successful adaptation of any chosen trading strategy. Winton Capital prefers trend-following strategies steeped in quantitative analysis. Practitioners of trend-following prioritize the accuracy and timeframe of their data over other factors. Of course, appropriate algorithms to crunch the dataset are crucial so as to stay ahead of the pack. Algorithmic strategies that center on mean reversion will also probably place the main emphasis on data veracity, like the trend-followers. Thus, the differences between these two algorithmic strategies mostly lie solely on their fundamental differences in trading beliefs. Newer developments include data-mining, which utilizes programming concepts like pattern recognition and artificial intelligence. Data-mining supports the concept of trend anticipation as opposed to only utilizing past data to predict future trends. It involves algorithms that actually screen or ‘mine’ millions of comments on social media sites like Twitter in order to gauge market sentiments. On the other hand, major conglomerates like Citigroup have inclinations to lean towards HFT strategies that further bolster their market-making capabilities. Since 2006, the firm has made a series of acquisitions in the electronic trading sector in order to enhance their share of the total market volume in US exchanges. This seems to evince that big firms tend to primarily reaffirm their market-maker statuses when pursuing algorithmic trading strategies. Then, hedge funds like Winton Capital invariably adopt different outlooks in line with the beliefs of the fund founders. Most firms do not employ a single strategy and their arsenal often includes a combination of preferred algorithms. This illustrates that the strategy choices are often bound to a significant extent by inherent regulations, ‘roles’ and expectations or beliefs, eventually culminating in the proverbial arms race within each segmented market.

Many believe that the dramatic increase in HFT can be mostly attributed to its profitability; in 2010, HFT generated approximately $25 billion in revenue on Wall Street. Algorithmic trading has entrenched itself in the American and European markets and its reach is slowly but surely permeating Asian and emerging markets. Chiefly, the relaxation of regulations in Singapore and Australia allowing high speed trading platforms and off-exchange trading closely followed Japan’s lead in January 2010.4 Interestingly, as the European and American markets become increasingly saturated with competing firms trying to deliver on the same platforms, it may well be that firms who turn their attention to other markets will benefit from first-mover advantage and reap healthy profits. This may potentially yield more than what an obstinate focus on over-saturated markets can bring. However, we can be certain that the arms race will continue to rage when enough firms converge on previously untapped market domains.

We have seen how HFT can actually increase liquidity and reduce trading costs for retail investors by fulfilling classical market-making roles. Are there any associated dangers then? Cue the flash drop of 6th May 2010 on the NYSE, which saw a 600-point drop over 5 minutes. Fairly or unfairly, HFT has often been attributed as one of the main contributors of that fiasco. Till this day, there are differing viewpoints as to the root causes and it did not help that the public viewed the review committee which was responsible for investigating the incident as inefficient. The investigation details were only revealed on 30th September, nearly five months after the crash. Certainly, HFT exacerbated the decline of prices when the algorithms worked to eliminate their positions and withdraw from the market. This affected liquidity and further exerted downward pressures on prices. In essence, it was thought that although HFT does contribute to significant amounts of volume, there is no actual accumulation of large positions amongst firms unlike traditional market-makers.5 Specifically, trading patterns are aggressively driven towards the direction of price changes and positions are ‘flipped’ from one firm to another. Consequently, there is no actual liquidity; what is in place is the illusion of liquidity. This is most dangerous for individual investors and smaller funds. Worse is to come when firms attempt to rebalance their positions and compete for liquidity, amplifying price volatility. HFT has the potential to augment speculation such that it is too fast, too furious to reasonably control or regulate, at least not with current regulatory mechanisms.

Regulatory bodies are thus under public pressures to enhance regulatory controls and compliance for HFT firms. Trading curbs in the form of circuit breakers to prevent re-occurrence of flash drops have subsequently been introduced in the stock exchanges. The percentage of HFT has dropped slightly as a result, with focus shifting to the futures, forex and commodities markets. The trend remains: Whichever market that accommodates HFT will entail dealing with increased volatility levels. However, correlation does not necessarily equate causation, and it would be too simplistic to attribute HFT as the sole factor of volatility. More importantly, regulators must decide on the appropriate levels of volatility and enforce monitoring measures with tough penalties in order to ensure the markets are operating desirably. It must be said that one HFT firm on its own with its preferred set of algorithms cannot do much; potential havoc is wreaked only when interactions amongst many firms occur in the markets simultaneously, taking the same trading decisions because of the inherent similarities in their trading algorithms. It is the prerogative of HFT firms to adopt any lawful trading strategy, though their choices will most probably center on the algorithms that can potentially generate the most profits. It is not their concern if the by-product of these choices results in increased volatility for other investors. Thus, it is unlikely any self-regulation will take place and the onus is on the regulators to police and enforce whatever trading ideals they hold.

The digitization of trading exchanges throughout the world has inevitably led to the proliferation of algorithmic trading strategies. There is an ongoing ‘herd mentality’ that is embraced by firms and no one wants to lose out. The algorithmic trading stratosphere is ever evolving, with changes occurring at a frenetic pace and firms continually ‘sharpening their armaments’. Truth be told, practitioners of algorithmic trading will stop at nothing in order to stay ahead of the pack. This is especially so when they recognize or believe that there are still profits to be made in their respective markets. Unscrupulous firms may even exploit regulatory loopholes all in the name of profits.4 It does not matter which strategy one chooses; the race is truly on and the competition is intense. There are no signs that it will abet anytime soon and it is up to the regulators to ensure that the markets are of a relatively level playing field, in the interest of fair trading.

References: 
  1. Hedge Funds Review http://www.hedgefundsreview.com/hedge-funds-review/interview/1648716/winton-futures-fund-winton-capital-management (accessed May 29 2012)
  2. Michael K, Tanmoy Chakraborty, Market Making and Mean Reversion http://www.cis.upenn.edu/~mkearns/papers/marketmaking.pdf (accessed May 29 2012)
  3. Michael J. McGowan, The Rise of Computerized High Frequency Trading: Use and Controversy, 9 Duke Law & Technology Review 1-25 (2010)
  4.  Snider L, Criminalizing the Algorithm? Stock Market Crime in the 21st Century
  5. http://www.deakin.edu.au/arts-ed/shss/events/anzsoc2011/keynote-snider-plenary.pdf (accessed May 29 2012)
  6. Kirilenko, Andrei A., Kyle, Albert S., Samadi, Mehrdad and Tuzun, Tugkan, The Flash Crash: The Impact of High Frequency Trading on an Electronic Market, Social Science Research Network
  7. http://ssrn.com/abstract=1686004 (accessed May 29 2012)