Algorithmic Trading Hedge Funds: Past, Present, and Future

CAREER ADVICE   |   8 minutes  |   June 27, 2018

Globally, hedge funds manage 5.3% of all currency in circulation1. Starting at a mere $100,000 in total assets almost 70 years ago, hedge funds have seen dramatic growth2. By 2018 the number of hedge funds globally had increased to well over 8000, holding a total asset value of over $3.2 trillion – an all-time high3. Industry Analysts predict, with the application of technology, the sector will continue to grow in the coming years4.

Within this high growth environment, investment strategies of top-performing hedge funds are increasingly dominated by algorithmic trading5 – as a result, some saw 2017 providing an enormous return of over twice that of their nearest traditional counterparts6. To really come to grips with this growth it’s important to take a step back and understand how we got to this point.

A new approach to investing

Born to a US Diplomat in Australia in 1901, Alfred Winslow Jones read for his doctorate in Sociology at Columbia University in 1941. He went on to work as a Reporter for Fortune Magazine. Despite not having much experience covering financial topics, he was assigned to write an article on “Current Fashions in Investing and Market Forecasting”. It was during his research for this assignment Jones realised he’d uncovered a novel and productive approach to investing. By combining the advantages of long selling, short selling, leverage, incentive fees and shared risk, Jones identified the core tenants of a classic hedge fund7.

His strategy was to hedge long positions (purchasing a share at a future time at a pre-agreed, higher price and set volume) while selling short other stocks (agreeing to a lower price and on a set volume in the future). The hedge fund was able to offset risk on their portfolio by betting on both longs and shorts on the same or related stocks. Debt raised against the fund assets (leverage) combined with this risk offset mechanism, gave him a new perspective on how to use pooled investment resources and paved the way to a new trading philosophy8.

Starting with just $100,000 (40% of which was Jones’s entire net worth) and a stifling 20% commission on performance, the first hedge fund was born in 1949. Jones and his team quietly outperformed the market for 17 years with little interest from other investors. It was not until 1966 when an article by another Fortune Magazine journalist noted the top Dreyfus Fund grew by a staggering 87% over 10 years and was considered the best mutual fund by 44% over five years, that serious competitors started joining the market9.

A hedge fund is a partnership between various investors where the fund pools these assets, attempts to leverage it to borrow further funds and uses various strategies (such as derivatives, futures etc.) to earn active returns off both local and international markets10.

What is algorithmic trading?

Algorithmic trading is the use of algorithms or rules to make purchasing and sales decisions on the behalf of the investor. Surprisingly, by this definition, algorithmic trading is even older than hedge funds11.

While the current application is cutting edge, the laws governing it has been noted as early as 1815 when Nathan Mayer Rothschild used his private postal service and secret couriers knew about the victory of the Duke of Wellington in the Battle of Waterloo before the rest of the market did. Rothschild informed his traders to, on mass, sell as many of their consuls (a government bond) as possible. The market, aware that Rothschild likely knew the outcome of the battle (and was behaving as if it were a loss) quickly followed suit and feverishly sold their consuls – causing their value to crash. Soon after, Rothschild changed his message and immediately asked his staff to buy as many consuls as possible. When the news arrived the next day that the battle had been won, the value of consuls increased dramatically and he earned a 20:1 return in a little under 36 hours12.

Algorithmic trading in the 21st century maximises returns by arbitraging the advantages of enormous amounts of data, tiny processing delays, huge throughput capacity, dynamically balanced portfolios, and intelligent predictive analytics. By exploiting one or more of these systemic improvements, algorithmic trading can take many forms: trading ahead of index fund rebalancing, arbitrage, scalping and mean reversion, pairs trading, and delta-neutral strategies13. Irrespective of the strategy, algorithmic trading almost always means reduced transaction costs for the investor with fees dropping from 20% in 1949 to 0.25% in 2018, for some algorithmic trading platforms.

The amendment of minimum stock value from 1/16th ($0.0625) to 1/100th ($0.01) of a dollar per share, and the increasing digitisation of stock markets throughout the 1980s and 1990s, changed the structure of stock trading. Algorithmic trading exploded and proliferated in the market due to the US Securities and Exchange Commission authorising electronic exchanges in 199814.

Despite the sluggishness of algorithmic trading in 2001 (taking several seconds to execute an order15) it quickly gained 10% of equity orders. Spurred on by technological improvements, algorithmic trade volumes increased by 164% between 2005 and 2009. In 2009 it was calculated 56% of all trades in the US were done algorithmically16.

Algorithmic trading is not without its unique risks and inherent problems – it has been blamed for a number of stock exchange crashes and numerous investors have made monumental losses.

6 May 2010, the US Stock Market went into chaos in what is now dubbed as “The Flash Crash”17. For 36 minutes all shares were inaccessible as all stocks began to rapidly collapse and rebound in quick succession resulting in the loss of 998.5 points in minutes – the largest intraday point swings recorded up until that date. In total nearly $1 trillion was wiped off the market value.

After much investigation Navinder Singh Sarao was eventually convicted for his role in the crash. A prototype algorithmic trader, which he had built for experimentation placed thousands of futures contract orders, refreshed them almost 20 000 times (generating a sale worth $4.1 billion) and eventually cancelled them. This anomaly was accentuated by other algorithmic traders trying to adjust and respond. The resulting regulatory changes which have made it significantly safer to operate algorithmic trading18.

Two years later, Knight Capital provided the world with another diabolical algorithmic trading example. Immediately after taking their new algorithmic trading platform live in 2012, they realised something was wrong with their high-frequency trading strategies. Their algorithm wasn’t making intelligent buys or sales, resulting in a $10 million loss per minute. It took their team almost 45 minutes to shut the platform down19.

As the market observed a number of algorithmic trading examples similar to the two mentioned, investors and traders questioned the true predictive capacity of this mechanism – to borrow from Gartner’s Hype Cycle, algorithmic trading entered the “Valley of Despair”, waiting for software techniques and hardware capacities to mature to support its ambition20.

Now, six years later, algorithmic trading has transitioned into a market leading solution. Algorithmic traders have reduced their latency (the time that it takes for them to get the information to and from the market) down to just 20 nanoseconds21 – an extraordinary achievement, and empowering for high-frequency trading strategies. This is about two million times faster than the normal internet users are served, even with the fastest web content. This is achieved despite the tremendous volumes of data needing to be processed to keep up with, store, and process a live feed of all market activities, along with a plethora of other related data sources. The supporting technical infrastructure is cutting edge – allowing upward of four billion algorithmic trades per day (vs 2.2 billion from Active Funds).

Everything is poised for enormous algorithmic trading success.

2017 algorithmic trading Leaders

The regulatory changes and technological advances of the last few decades have enabled algorithmic trading hedge funds to regularly outperform traditional funds.

Take 2017 as an example for comparison.

S&P 500 Index reached a three year high – up 19.42%22 – and the top performing traditional fund (SH Capital Partners) posted 234.09% returns. Over the same period Silver8 Partners and Global Advisors Bitcoin Investment Fund achieved 770.75% and 330.08% returns respectively23.  Both of these algorithmic trading examples are automatically traded but differ on specific strategies – however, both attribute their success to their algorithmic trading winning strategies and their rationale on digital assets.

In the case of Silver8, they explore ways of investing in innovation within financial technology itself – further allowing them to leverage the growth in the sector on a many-year time horizon, while Global Advisors focuses exclusively on cryptocurrency trading24.

While hedge funds present a suitable solution to the business-to-business algorithmic trading market, the recent introduction of roboadvisors has made algorithmic trading accessible to individual investors with self-managed portfolios. These automated trading solutions make individual stock selections based on personal risk profiles. Currently the top performing roboadvisor platforms, such as Betterment, are achieving upwards of 11% returns – on par with traditional indexes, but at a fraction of the commissions and other management costs25.

Considering the macro-future

By volume algorithmic trading is already responsible for between 60-73% of all US equity trading. With roboadvisors and hedge funds wielding algorithms like magic wands, there is increased pressure on the financial professionals to stay relevant26.

Competing with the sheer processing capacity and 20 nanosecond latency of huge data centres is humanly impossible. Innovative application of these technologies and creative problem solving will be required from corporates and startups alike – should they wish to forge their own path in this space and ensure market relevance in the coming years. In particular, proactive and forward-looking businesses will be paying attention to the types of data they can apply, by means of collecting, then processing it (such as machine learning trading) and how to then leverage this data for intelligent (human-assisted) investment decisions.

Following initial efforts by the stock market to make their systems open via (an early form) API in 1998, there are examples of this thinking permuting other components of the financial sector. Within the EU for example, Payment Services Directive 2 (PSD2)27 is a piece of regulation which forces financial institutions to provide secure, privileged, and consenting access to banking details by third parties via an Application Program Interface. It’s fundamental purpose is to reduce systemic friction and empower a new wave of products and services to be built upon existing banking infrastructure. This sort of thinking is expected to start to extend into all areas of the financial market – further exposing opportunities for agile and proactive financial market players.

The accessibility of the algorithms and tools driving algorithmic trading are increasingly in the hands of the amateur investor, further accentuating pressure on industry leaders to stay relevant.

For example, at a total setup up cost of less than $100 including new hardware, Gekko – a cryptocurrency algorithmic trading platform which connects to popular exchanges – enables and assists individual investors to design their own trading strategies28. The platform creates the algorithmic equivalent of this strategy and can then either test it on historical data or implement it live.

The success trifecta which characterises algorithmic trading – strong performance, broad adoption, and relatively low workforce needs – also presents an interesting situation for the individual financial market employees. In 2003, Oxford University concluded “47% of all jobs were at ‘high risk’ of being automated within the next 20 years, with 54% of these losses coming in the financial sector”29. To illustrate this prediction, India is already reporting financial sector headcount reductions of 7% per quarter (Q3 and Q4, 2017)30.

Despite this seeming negative, there is a key positive truth: even with a “fully automated” system, humans are still critical to the processes of relationship development, system development, and various other key functions within the business. By leveraging these tools to do the heavy lifting, your focus on present-day execution can decrease allowing for an increased focus on building truly innovative and disruptive solutions to future problems.

Adapting to a future of change

The 21st century can be categorised by an increasing rate of change, an increasing level of uncertainty, and an increasing reliance on and integration of technology into societies and cultures31.

To become or remain relevant, businesses will need to clearly identify their purpose, understand their new role, keep up-to-date with developments, and remain agile to environmental and market forces.

Driven by innovation and disruption, this shifting environment will present many groundbreaking opportunities for rewarding careers in trading. Preparation and adaptability will determine who thrives.