Archive for April, 2008

Quantative Hedge Funds – Non-Anecdotal Evidence Required

April 7, 2008 at 7:15 am · Filed under Uncategorized

I think I should take a different tack, as the response to my last post has been somewhat underwhelming, ranging from Nicole’s “well you’re obviously some kind of communist” (I’m paraphrasing), to Amir’s perfectly reasonable “hmm… I think there might be something beneficial to this stuff”.

What can’t be disagreed on is that quantitive trading strategies have a significant impact on the operation of the financial markets, and that the operation of the financial markets has a signficant impact on all of us. My focus in all this comes from my 3 main interests:

1) Accelerated Computing. I work in this field, and I see a lot of interest in orienting the industry towards serving the needs of quantatitive finance.

2) FInance and Economics. I’m not a qualified expert in this field, but I’ve had professional exposure. I’ve implemented some quantitive finance algorithms, and would say that I understand them and their weaknesses fairly well (e.g. assumptions based on historical volatilities in options pricing). In my spare time I read a lot on economics and finance.

3) Pseudoscience. I’m interested in how strongly otherwise rational people can hold beliefs that defy logic and scientific evidence. Examples include homeopathy. Homeopathy is implausible to the point of impossiblity in its proposed mechanism of action. More importantly, scientific evidence does not exist to suggest that it works beyond placebo. Proponents of homeopathy provide anecdotal accounts accompanied by logical fallacies (I took homeopathic remedies, and got better, therefore homeopathy works).

To prove that quantitative hedge funds have merit, they need to be investigated. This investigation should take place on two levels:

Level 1: Do these funds on average actually make money for the people that invest in them? Saying “people wouldn’t invest in them if they didn’t” is not a valid argument to say that they do.

Level 2: Is the effect on the wider economy of unrestricted quantitative trading beneficial, neutral or negative? This one’s a lot harder to quantify, but it doesn’t mean we shouldn’t try.

A good article on the state of play in quantitative finance can be found here.

So now I’ll pick out some things from the article and comment on them, asking some questions I would sorely love to know the answer to:

The article discusses how quantatitive hedge funds have losing lots of money since the credit markets blew up. Their models didn’t see it coming. This on its own doesn’t prove that there is no value in the quant strategies of course, if it’s balanced out by the good years, then all is fine.

In discussing the turmoil, Marek Fludzinski says:

“We had layers of option strategies on top of our stat arb strategies to protect them from this catastrophic risk.”

Question: Do these strategies take into account the ability of the counterparties to pay out on these options in the event of catastrophe? If the “catastrophe” you’re insuring makes your counterparty insolvent, then you’re not insured against the risk.

The article suggested suggested a reason for the decline in quant hedge fund performance:


Too much money trying to exploit the same market inefficiencies accounts for some of the problem. HFR estimates that investors had $225 billion in quant hedge funds at the end of last year, more than the entire hedge fund industry had in early 1996. Overcrowding can be especially trying for quants, whose portfolios typically hold thousands of positions, making overlap among managers inevitable.

Question: When devising their quant strategies, to what extent do algorithm developers consider the role that quantitative trading activity plays in market behavior?


Thales’s 35-person research team, which includes ten Ph.D.s, is looking for ways to forecast investor behavior on different timescales. (In total, the firm employs about 50 people.) Although Thales focuses most of its effort on improving current trading systems and strategies, Fludzinski says the firm devotes as much as 20 percent of its resources to finding a unifying theory of markets. Like physics, he says, finance lends itself to mathematical models based on observations. Through experiments that replicate what they see in the natural world, physicists have described its underlying principles.

Question: Is there any rational basis for supposing that the markets (made up almost entirely of entities attempting to predict the behavior of the markets) should have behavior that can be described by theories like those in the physical world?


“Maybe we need to build a computer simulation that has 50 million people, with complicated rules for each,” Fludzinski says. “It’s very difficult to explain why people behave irrationally.”

Question: Is it reasonable to expect to be able to construct a model of the markets that is anything other than full-scale (i.e. containing a perfect model of every single actor in the market?) If such a model existed, would it be possible to intervene in the markets without changing the course of the market evolution predicted by the model? How can such models deal with unpredicted events with the ability to affect markets? e.g., earthquakes, terrorist attacks, conflict, and other such occurences.


As he seeks to right his faltering hedge fund, Thales’s Fludzinski could stand to reap such rewards sooner rather than later. But he’s not losing sight of his quest to explain why markets do what they do — a goal that may require a leap of imagination. Just as quantum mechanics and the theory of relativity shook up the world of physics, Fludzinski says, “finance needs a similar out-of-the-box insight — something that changes the way we look at things.”

Question: Is it a logical absurdity to suppose the existence of laws that predict price movement in markets?

Lo says he’s not surprised that statistical arbitrage and other quant strategies got slammed last summer. “Quant funds by their nature are going to be invested in mostly liquid strategies, and so when we have massive liquidations, they’re going to be first in line to take their licks,” he explains.

To protect themselves, Lo says, quant funds must understand their exposure to future sell-offs. That means paying close attention to the relationship between liquidity and their strategies’ expected returns. “We can’t forecast with any degree of accuracy when the next LTCM is going to hit,” Lo says. “But we can tell when certain market conditions are ripe for a dislocation and gradually try to adjust our risk exposures to take that into account.”

Lo believes that forecasting such disruptions will remain a problem — at least until hedge fund firms surrender more-detailed information. “The hope is that if we can get more data for both investors and managers to use, we will be able to avoid this kind of mad rush to the exits in the future,” he says. — N.R.

Question: If hedge funds could detect an upcoming sell-off, and therefore sell off before the crash, would this not simply move the sell off further forward in time?

One thing I’m in no doubt of is that there are lots of “clever” people in this field, but what these guys are looking for seems implausible and illogical, and I’m still waiting to see concrete data that shows the benefit of quantitative hedge funds, firstly to the investors, and secondly to the economy as a whole. Given the 2% off the top and 20% of profits that is taken from investors money and given to the quants, I am in no doubt that it’s economically beneficial to the quants.

Comments (4)

Acceleration of Algorithmic Trading – The Emperor’s New Clothes?

April 2, 2008 at 9:45 am · Filed under finance

Acceleration of automated trading algorithms. Much effort is being expended in this area of computing to accelerate trading algorithms that buy and sell financial securities. These financial securities include common stock, futures, options, bonds and other more esoteric derivatives and structured finance products. Large amounts of research capacity and investment capital in the field of accelerated computing are targeted at such trading algorithms. Initially, trading algorithms were limited to replacing repetitive trading operations that were previously carried out by human traders. An example is “market making”, where a trader in a particular security or class of securities would buy and sell on their own account, either to facilitate a liquid market, dampen market volatility or to profit on bid-ask spreads. Other algorithmic trades include transaction cost reduction and arbitrage. These examples of algorithmic trading can be considered as labor-saving uses of computing, and are not what I want to talk about.

Instead I want to focus on information-based algorithmic trading, where computers process information relevant to the market of a particular security or class of securities, and then buy or sell as a result.

Given the extent to which the electronics and computing industries are positioning themselves to serve this market it is important for us to ask questions about its viability.

It is important for us to cogitate on what has just happened in the financial world. The sub-prime financial crisis originated in a single logical fallacy. Financiers observed that the US median house prices had not decreased on an annual basis in modern history. Fallacious logical deduction led to the axiom: Median House Prices in the United States Do Not Decrease on an Annual Basis. Upon this axiom an entire industry was built. The mathematical models, and the rigorous logic deployed in their development, did nothing to correct the base axiom of the industry. No matter how rationally the people that implemented and obeyed the models behaved, the end result could never overcome the original fallacy of logic. The result was a lot of companies and individuals profiting short-term, and the economy and wider populace suffering in the long term. I fully expect that the net result to the world will be seen to be negative when the dust has finally settled.

A lack of critical thinking and a belief in easy money got us into that mess, and it’s a pattern we are likely to repeat, unless we have a serious think about what is and isn’t possible in the financial markets.

Our greed and hero worship mean we look to skypilots like Ray Kurzweil to find magical ways of turning advanced technology into money.

“Artificial intelligence is becoming so deeply integrated into our economic ecostructure that some day computers will exceed human intelligence,” Kurzweil tells a room of investors who oversee enormous pools of capital. “Machines can observe billions of market transactions to see patterns we could never see.

The idea is:

“A market theory” + clever people + computers = money

“These ideas are the future,” said David Atkinson, a private investor who attended another lecture later that day by Kurzweil. “I’m not really sure I understand them, but they’re making some folks rich.”

However, if the model is fundamentally flawed, then over time all information-based algorithmic trading can hope to achieve is a steady erosion of capital through the costs of trading: that is to say the per-trade commission, the costs of the hardware-software infrastructure and the human resource costs as well as other costs. Those receiving commissions, salaries and purchase orders from the information-based algorithmic trading industry are incentivized to promote such trading regardless of its efficacy. However, unless it’s directing people to better invest capital in more productive industries, then it is a negative-sum game for humanity, a cancer that will eat at our economies. Look at the example of Long-Term Capital Management, and then look at how John Meriwether has fared in his latest hedge fund, which is going through the hoop as I write. We have turned our economies over to witchdoctors, convinced that because these people are intelligent and influential that they are correct. Well, they’re not. There is no easy answer. An economy sinks or swims on how wisely it allocates its capital. There is no reason to think that running billions of routines per second to try to exploit trends in security and derivative pricing would aggregate to the intelligent allocation of capital.

Robert Shiller’s plot of the S&P Composite Real Price Index, Earnings, Dividends, and Interest Rates, from Irrational Exuberance, 2d ed.[1] In the preface to this edition, Shiller warns that “[t]he stock market has not come down to historical levels: the price-earnings ratio as I define it in this book is still, at this writing [2005], in the mid-20s, far higher than the historical average. … People still place too much confidence in the markets and have too strong a belief that paying attention to the gyrations in their investments will someday make them rich, and so they do not make conservative preparations for possible bad outcomes.”

 

Shiller points to the increasing disassociation between the price people are paying for securities and the underlying fundamentals on which price should depend in a rational market. How else can this be viewed but as a Ponzi lunacy?

But, in the new world, who cares if a company is paying dividends or not? All you need do is predict trends!

Last year, funds co-managed by Karaali returned in excess of 20 percent by using nonlinear techniques, according to his company. Whereas older methods of stock analysis rely on certain assumptions – for instance, that market volatility always reverts to the mean – Karaali’s models calculate probabilities and generates [sic] assumptions on the fly, and might predict that during a panic, investors will sell Microsoft but, for seemingly irrational reasons, hold onto IBM.

“Only an elite group of people are using these ideas, but a lot of people are thinking about them,” said Stacy Williams, director of quantitative strategies at HSBC Global Markets. HSBC is working with Cambridge University in using models based on how viruses spread to forecast foreign currency markets.

“The downside with these systems is their black box-ness,” Williams said. “Traders have intuitive senses of how the world works. But with these systems you pour in a bunch of numbers, and something comes out the other end, and it’s not always intuitive or clear why the black box latched onto certain data or relationships.”

This is self-evidently absurd! The idea that automated trading algorithms can predict the behavior of a system and act accordingly when their actions will change the very nature of that system. Together with the hundreds of other machines that are attempting to do the very same thing.

Automated trading algorithms should be relegated to the realm of pseudoscience where they belong, with homeopathy and astrology. Besides, can you imagine a caveman society where the cavemen didn’t go out on the hunt, but instead stayed in the caves, gambling on the result of the hunt? Do you think such a society would ever have made it out of the caves?

We should not allow automated trading algorithms to shape our destinies. The computer and electronics industry should be putting its human and financial capital to bring humanity forward. The grand challenge problems of energy, environment and healthcare need to be solved. It is plain immoral to use our vast computing resources to destroy our capital markets with lunatic financial alchemy. We will need both to solve our real world problems.

Comments (4)