Demystifying the quant
Quantitative analysis is seen by many as something of a dark art, but it plays an increasingly central role in investment and risk management. (First published on the Dolfin website on 3 April 2017.)
From big data to driverless cars and artificial intelligence, technology is changing the world. At the forefront of this revolution in the investment world is quantitative analysis. Move over Gordon Gekko — the new ‘masters of the universe’ are now likely to be powerful algorithms programmed by ‘quants’ with a background in maths, finance and computer science.
These super-geeks create mathematical models that aim to predict the future of the markets. They may have first been used by hedge funds in their aggressiveness to make money, but banks, asset managers and others have long since caught up with the use of quantitative methods.
Today’s traders predominantly rely on computers to make decisions that were once the preserve of humans. BlackRock, for instance, uses data from satellite images of China’s biggest cities to help work out whether it should buy or sell stocks in Chinese developers by observing the size of the shadows of buildings in the images and drawing a conclusion. Through this type of quantitative analysis, investment managers can remove bias, create consistency and make decisions more effectively.
“Essentially, the job of a quant involves building models to create profitable trading strategies, analyse the risk inherent to a portfolio or company’s book, or find outliers in markets,” says Karl Sawaya, Quantitative Developer at Dolfin.
“A quant builds models to create profitable trading strategies.” — Karl Sawaya, Quantitative Developer, Dolfin
The role of the quant has changed significantly, too. Once just attached to trading desks, today you will also find these maths whizzes sitting within risk management teams carrying out a wide variety of tasks. Improved analytic software has helped quants glean insights from ever larger volumes of information. But, say the quants, there is a subtle difference between big data and a lot of data.
“You can have many problems with the data and end up having gaps in data, dirty data or too much data,” says Geoffrey Boullanger, Senior Quantitative Developer at Dolfin. “For instance, you can have the whole Bloomberg database and still lack information, compared to having access to a dozen of smaller databases with a bigger degree of variability, perspective and angle.”
The human touch
Of course, quantitative analysis should not necessarily be relied upon to come up with the definitive answer. There is still a place for subjective, human-based thinking — even if it is just to sense-check the results of an algorithm.
“Quantitative analysis should still be heeded with care,” says Boullanger. “Many assumptions are made and biases accepted. Lots of models are also sensitive to big shocks in the market. These things cannot be neglected nowadays.”
“Quantitative analysis should still be heeded with care.” — Geoffrey Boullanger, Senior Quantitative Developer, Dolfin
Algorithms, in the main, failed to pick up on the biggest of ‘black swan’ event of recent times — the financial crisis. It could even be argued that the algorithms programmed to trade complex financial instruments like credit default swaps even exacerbated the crash — causing the herd mentality that sent markets into free-fall.
Even one bad algorithm can cost a firm millions of pounds in a matter of minutes. This was the reality for market-making firm Knight Capital in 2012, which lost $440 million in just 30 minutes after faulty code started executing trades erroneously. This huge blow eventually led to Getco buying out Knight Capital, and the formation of KCG.
Despite this, technology and mathematical modelling has moved on. In 2015, eight of the world’s top 10 best-paid hedge fund managers used quant-based methods to make their money.
Many quant strategies today are extremely profitable — with the results gleaned from algorithms generally reliable and objective. Just as importantly, algorithms remove the human emotion of fear from the equation — which can quite easily cloud judgement.
“It makes sense for most people not to accept results coming out from a code, compared to human analysis,” says Sawaya. “Yet, we must not forget that the human element also incorporates fear and psychology in the decision-making process, which is not exactly a good investment approach. Using quant models helps reduce and even eliminate these issues, but we should bear in mind that it is humans who eventually write the code. Hence the logic behind the model.”
For firms to achieve the best results, employing quant strategies can help bridge the gap between an academic view of the markets, technical analysis and profitable trading. Human investment expertise remains the critical factor.
Before it’s here, it’s on the Dolfin website.