• C++ Programming for Financial Engineering
    Highly recommended by thousands of MFE students. Covers essential C++ topics with applications to financial engineering. Learn more Join!
    Python for Finance with Intro to Data Science
    Gain practical understanding of Python to read, understand, and write professional Python code for your first day on the job. Learn more Join!
    An Intuition-Based Options Primer for FE
    Ideal for entry level positions interviews and graduate studies, specializing in options trading arbitrage and options valuation models. Learn more Join!

Financial Times Article on QF.

<H2>Does not compute: How misfiring quant funds are distorting the markets

By Anuj Gangahar
Published: December 9 2007 19:59 | Last updated: December 9 2007 19:59

George Pelham-Box, one of the most influential statisticians of the 20th century, once remarked that "essentially all models are wrong, but some are useful".
During the global capital markets turbulence of August, several celebrated quantitative hedge fund managers were forced to admit that their long-held investment models were indeed wrong and, at least that month, not especially useful.

Marble Bar managers in $400m bonanza - Dec-03

Low-risk trades put all others in the shade - Nov-25

1000% hedge fund wins subprime bet - Nov-25

One hedge fund in 10 to go bust, says Man - Nov-20

View from the Top: Peter Clarke - Nov-20

Hedge funds on new ground - Nov-20

The performance of these so-called quantitative funds, which trade using statistical models designed to identify patterns in financial markets, is increasingly important because they account for such huge trading volumes: Tabb Group, the US consultancy, predicts that by 2010 algorithmic trading, one aspect of "quant"- based investing, will account for half of all US equity trading.
Since August, when many quant funds lost badly, the specialists who developed these models have been fine-tuning them to try to iron out the bugs. So far this has been unsuccessful. According to some estimates, the amount of money managed by quant funds has dropped by up to 40 per cent in the past six months as the drawbacks of the strategy become apparent.
Furthermore, concerns that the models have been playing havoc with markets persist. The tell-tale signs are sudden movements in US equity indices, most frequently as a result of a sell-off in the last hour of trading sessions, which are accompanied by steep increases in volume as the computer-based trading programmes sell huge numbers of shares quickly. These late-day swings have occurred more recently as well, suggesting that the August pattern was not a one-off.
Traders themselves are undecided about what causes such spikes and how exactly they are connected to quantitative trading. Andrew Wilkinson of Interactive Brokers, the direct-access broker, said the swings could be related back to the "Vix", Wall Street's so-called "fear gauge", which is used by many as the primary benchmark for equity market volatility. It is derived from the price of buying protection on the options market against swings in the S&P index.

It follows that the more the Vix rallies, the more fearful traders are, and the less they are prepared to bet on its direction. Traders simply stop trading the Vix because they get anxious. Quantitative models programmed to sell in huge volumes when certain activity levels in the Vix are reached then kick in, leading to the sell-offs and volume spikes. "It is as though the market is playing some game of chicken."
Other critics offer a more fundamental explanation for the recent problems endured by quant models - lack of innovation and unrealistic assumptions. Nassim Nicholas Taleb, author of the best-selling books The Black Swan and Fooled by Randomness: The Hidden Role of Chance in the Markets and Life, go so far as to say that the very principle of using quantitative models based on history is bound to fail and should be abandoned.
Dimitri Sogoloff, president and chief executive of Horton Point, a US-based quant hedge fund, believes the failure to apply cutting-edge scientific principles to quantitative investment has contributed to its downfall. He says the lack of recent crossover of scientific theory to economics and finance has damaged the ability of quant models to maximise their predictive power. Mr Sogoloff calls for the development of so-called "algorithmic alpha generation" - a product of a truly innovative, uncorrelated and dynamically adaptable investment process.
But Paul Alapat of Amba Research, a Bangalore-based quant research house that serves many of Wall street's largest banks and hedge funds, argues that the mathematical, engineering and physical science backgrounds of many quant managers and their teams has contributed to their strategic dilemma. "Engineers and mathematicians are programmed to think in a very precise and rules-based way. I think they need to temper the physics with psychology."
He draws parallels with the dilemma faced by investors in property. "Everyone knows there is a housing bubble, but the problem is that if you take your money out even a day before it bursts, you lose out. With quant funds, when market moves are sudden you can be forced to liquidate valuable positions, and this is intrinsically difficult to do."
August was by no means the first time that quantitative trading strategies have caused trouble. Long-Term Capital Management, the hedge fund that famously collapsed in 1998, boasted some of the founders of the field among its senior executives. But the explosion of over-the-counter derivatives products, including those at the heart of the current credit crisis, has made such strategies more accessible to a far wider range of investors.
The rise of trading based on mathematical models has led to criticism by other investors that these giant automatons were distorting market behaviour and making it harder for less sophisticated investors to react to opportunities in time. Such criticism has died down. But a deeper question remains. Can they outsmart human beings - particularly when markets move unexpectedly?
The lessons of August are stark. Quant managers learnt that the models - many of which had been years in development - were flawed. Renaissance Technologies, perhaps the most celebrated quantitative hedge fund, and Goldman Sachs, the investment bank, reported steep losses.
One quantitative analyst says: "August demonstrates what we already suspected - that quant models cannot, whatever their complexity or relevance, adapt to brutal changes in market conditions. Quant funds hire physicists, mathematicians, astronomers and computer scientists and they typically know nothing about finance."
Companies also learnt that whatever parameters they had been using to make their predictions, a host of their peers and rivals had been barking up precisely the same tree, meaning models they thought were exotic and cutting-edge were in fact mundane. This was partly because the pool of quantitative investment professionals who are truly at the cutting edge is still very small. Mr Alapat says: "If a quant fund has a proprietary model that is successful, it gets mimicked very quickly."
Quant funds go to great lengths to hang on to their specialists, knowing that the second they depart they will use the same strategies at a competing firm, effectively diluting returns by chasing the same opportunities to make money.
Whatever its flaws, however, quantitative trading still has many advantages over its chief competitor, the human brain. From the individual who holds on to losing stocks for too long to overconfident money managers who mistakenly think they can predict financial trends, human nature is capable of placing bad bets time and time again. Psychology and raw emotion often rules the stock market.
Quantitative model makers point out that their products have had notable successes. Renaissance Technologies, for example, made average annual returns of 38 per cent between 1989 and last year.
But even when a trader is using a mathematical model rather than gut instinct, the question of when to use a mathematical model still comes down to judgment. It is this decision, particularly when markets are volatile, that can be the most fraught with difficulty and potentially most damaging to quantitative strategies.
The use of computer programs based on proprietary algorithms - to which even their investors are not privy - to make day-to-day investment decisions is often known as black-box investment. A "black box" contains formulas and calculations that the user does not see nor need to know or even understand in order to use the system.
"Black box models have a limited shelf life," says William Strazzullo, chief market strategist at Bell Curve Trading. "When they work it's good but over time relationships [between the factors included in the model] inevitably break down."
"Markets are in a state of constant flux and there is a danger that when the model breaks down, it's too late," says Mr Strazzullo.
Some participants are trying to adapt black box models to take into account factors that are not strictly financial - such as management behaviour and sometimes even press coverage and its impact - in an effort to remain a step ahead of the quant crowd.
One trader says that in his view, August was a special situation that could not be used to draw the pessimistic conclusion that quant models are fatally flawed. "In August you had a bunch of hedge funds that were essentially market-neutral and which had huge positions, which then developed funding problems due to the situation in credit."

Of the most recent spikes in volatility, Gary Ardell, head of financial engineering at BNY ConvergEx Group, says: "It looks like August again. People are selling the crown jewels. All the things that are supposed to move higher are falling. We are generally approaching the open of the trading day from a neutral position and by 9:45 decide which kind of trading model should be used."
In other words, many quant funds have been forced to sell some of their most successful positions in order to raise money, as they are unable to borrow in credit markets. For other investors who run their models over a longer time horizon, trading since August has been fraught with pitfalls. Models need constant fine-tuning in order to account for shifts in the investment landscape.
Daniel Stroock, a professor at MIT, said in a recent study that the role of quants in the market was analogous to the role batfish play in keeping coral reefs tidy.
Just as batfish do not construct the reef but are essential to its health, quants do not create the structure financial markets depend on but do preserve the conditions that make markets function. So, he argues, it would be misleading to suggest in any way that quants were responsible for this summer's meltdown in the subprime-mortgage market or for the broader troubles that followed.
The functioning of financial markets, he says, relies on the general acceptance of certain assumptions. One of the most important is the so-called "no arbitrage" assumption: that if there is a "free lunch" to be had, someone will eat it. If a trader can make risk-free profits by buying one security and selling another, he will do it. If a computer programme can do this faster, then it will get the lunch.
Thus, he says, it is essential that the assumption be correct, and an important role of the quant is to make sure that it is. By scrutinising huge volumes of financial data, he says, quantitative funds spot arbitrage opportunities and alert their fund managers before others have a chance to act.
Whether quants have a future, however, depends on how far investors burnt by the losses of August and the subsequent flight to less risky-looking assets can share Mr Stroock's confidence.