Information Processing

Just another weblog

Archive for November 2007

Jim Simons is my hero

with 5 comments

This is long, but worth the read! I just quoted the first part below. The whole thing can be found here.

Simons at Renaissance Cracks Code, Doubling Hedge Fund Assets 2007-11-27 00:13 (New York)

By Richard Teitelbaum
Nov. 27 (Bloomberg) — On a hot afternoon in September, Renaissance Technologies LLC founder Jim Simons is too busy to take a phone call. It is, he says, from Cumrun Vafa, a preeminent Harvard University professor and expert on string theory, which describes the building blocks of the universe as extended one-dimensional filaments.

“Get another time when I can talk to him,” Simons tells his assistant.

Then he mentions that the next day, he’ll be meeting with Thomas Insel, director of the National Institute of Mental Health, to discuss autism research. And he’s slated that Saturday to host a gala honoring Math for America, or MFA, a four-year-old nonprofit he started that provides stipends to New York City math teachers.

“I’m undoubtedly involved in too many things at the same time,” Simons says in his 35th-floor office in midtown Manhattan. “But you make your life interesting.”

String theory, autism, math education: It’s fair to ask how Simons, 69, manages his day job overseeing the world’s biggest hedge fund firm. The answer, judging from the numbers, is very well.

Renaissance is on fire: Its Medallion Fund — which uses computers and trading algorithms to invest in world markets — returned more than 50 percent in the first three quarters of 2007. It had about $6 billion in assets as of July 1.

Simons registered that performance as subprime and related markets were collapsing, sending two mortgage-related hedge funds run by Bear Stearns Cos. into bankruptcy. The turmoil pummeled the Goldman Sachs Global Alpha Fund, a rival to Renaissance’s funds, which fell more than 25 percent during the same time. Morgan Stanley’s computer jockeys lost $390 million in a single day in early August.

Life Story

Medallion’s returns are no anomaly. The fund, which trades everything from soybean futures to French government bonds in rapid fire, hasn’t had a negative quarter since early 1999. From the end of 1989 through 2006, it returned 38.5 percent annualized, net of fees.

More surprising than those returns is Simons’s life story. At an age when hedge fund pioneers such as Michael Steinhardt have long since stopped managing other people’s money, Simons is building on Medallion’s success. He’s adding funds and strategies and accumulating assets, which totaled $35.4 billion as of Sept. 28.

In August 2005, Simons started Renaissance Institutional Equities Fund, or RIEF, which invests in U.S. stocks. Through Sept. 30, it has returned 12.8 percent annualized. Unlike Medallion, which turns over its holdings dozens of times each year, RIEF keeps its positions for months or longer. Simons said at the time of the fund’s inception RIEF could theoretically manage as much as $100 billion.

‘New Possibilities’

In December 2006, he limited new investments in the fund to $1.5 billion a month. As of Sept. 30, 2007, it had $25.6 billion in assets.

In October, Simons started Renaissance Institutional Futures Fund, or RIFF, to invest in commodities. It’s up 5.2 percent for the month. He says Renaissance’s research shows the new fund can manage as much as $50 billion. Along with RIEF, it will promote cross-fertilization of ideas inside Renaissance, Simons says.

“Challenge is good,” he says. “It opens one’s eyes to new possibilities.”

When not in Manhattan, Simons runs his empire from a 15-foot (4.6-meter) by 20-foot office in Renaissance’s gated and guarded campus off Route 25A in East Setauket on New York’s Long Island, some 50 miles (80 kilometers) east of the Empire State Building. With most of the trading automated, there’s little of the hurly-burly of a typical hedge fund firm.

Doubling Assets

Along with routine personnel and marketing tasks, Simons makes time for the researchers and programmers who stop by his office to discuss mathematical and statistical issues they’ve encountered as they work on new trading strategies.

More than 200 employees, of whom about a third have Ph.D.s, work in East Setauket. Another 100 are based in Manhattan, San Francisco, London and Milan. “He creates an environment where it’s easy to be creative and works hard to keep the bullshit level to a minimum,” says former managing director Robert Frey, who worked at Renaissance from 1992 to 2004.

Even without the new commodities fund, Renaissance’s assets have more than doubled in a year from about $16 billion on Sept. 30, 2006. That growth has catapulted Renaissance past such titans as Daniel Och’s Och-Ziff Capital Management Group LLC, Ray Dalio’s Bridgewater Associates Inc. and David Shaw’s D.E. Shaw & Co. to become the world’s largest hedge fund manager, according to data compiled by Hedge Fund Research Inc. and Bloomberg.

Code Cracker

Medallion’s 3.9 percent return during August, though that fund too was whipsawed by volatility, bolstered Simons’s reputation as the silver-bearded wizard of quantitative investing.

In quant funds, mathematicians and computer scientists mine enormous amounts of data from financial markets looking for correlations among stocks, bonds, derivatives and other instruments. They search for predictive signals that will foretell whether, say, a palladium futures contract is likely to rise or fall.

“There are just a few individuals who have truly changed how we view the markets,” says Theodore Aronson, principal of Aronson + Johnson + Ortiz LP, a quantitative money management firm in Philadelphia with $29.3 billion in assets. “John Maynard Keynes is one of the few. Warren Buffett is one of the few. So is Jim Simons.”

Aronson credits Renaissance with validating the entire field of quantitative investing and proving that the freedom accorded to hedge fund managers to short stocks, borrow money and invest in myriad instruments can produce results that far outstrip typical market returns.

`Role Model’

Simons, standing just under 5 feet 10 inches tall and weighing 185 pounds (84 kilograms), has trod an unlikely path. A former code cracker for the U.S. National Security Agency, in 1968 he became chairman of the mathematics department at Stony Brook University, part of the New York state university system. He built the department into what David Eisenbud, former director of the Mathematical Sciences Research Institute in Berkeley, California, calls one of the world’s top centers for geometry.

In 1977, frustrated with a math problem and eager for change, he abandoned academia to start what would become Renaissance, hiring professors, code breakers and statistically minded scientists and engineers who’d worked in astrophysics, language recognition theory and computer programming.

“All the quants in the world are trying to follow in Jim’s footsteps because what he’s built at Renaissance is truly extraordinary,” says Andrew Lo, director of the Massachusetts Institute of Technology Laboratory for Financial Engineering and chief scientific officer of quant hedge fund firm AlphaSimplex Group LLC. “I and many others look up to him as a tremendous role model.” …

Written by infoproc

November 27, 2007 at 11:44 pm

Singularity summit

with 4 comments

Someone once said that the Singularity is like the Rapture for geeks 🙂

Nevertheless, some of the talks from the recent Singularity Summit are interesting.

So far I’ve enjoyed Omohundro, Jurvetson and Thiel. (Respectively, a physicist turned computer scientist, an engineer turned venture capitalist, and a derivatives trader turned PayPal CEO turned hedge fund manager.) Omohundro argues that AI beings will be much more rational than we are, using game-theoretic ideas of Von Neumann. Jurvetson thinks that we’ll use evolutionary algorithms to create AI. But the resulting beings will be so complex that, while we understand the process of their creation, we won’t understand how they really work — any more than we understand evolved biological beings. Thiel’s talk about investing for the singularity is goofy, but the comments near the end about Buffet’s positions are quite interesting. The talk by robotics pioneer Rodney Brooks is surprisingly lame. There are quite a few talks I haven’t yet had time to listen to…

Written by infoproc

November 27, 2007 at 11:19 pm

Posted in ai, podcasts, singularity

Monster minds

with 6 comments

There’s a story in Surely You’re Joking, Mr. Feynman, about the first seminar Feynman gives at Princeton. He’s just a graduate student, working on a formulation of electromagnetism in terms of advanced and retarded potentials with his advisor Wheeler:

…So it was to be my first technical talk, and Wheeler made arrangements with Eugene Wigner to put it on the regular seminar schedule.

A day or two before the talk I saw Wigner in the hall. “Feynman,” he said, “I think that work you’re doing with Wheeler is very interesting, so I’ve invited Russell to the seminar.” Henry Norris Russell, the famous, great astronomer of the day, was coming to the lecture!

Wigner went on. “I think Professor von Neumann would also be interested.” Johnny von Neumann was the greatest mathematician around. “And Professor Pauli is visiting from Switzerland, it so happens, so I’ve invited Professor Pauli to come” – Pauli was a very famous physicist- and by that time, I’m turning yellow. Finally, Wigner said, “Professor Einstein only rarely comes to our weekly seminars, but your work is so interesting that I’ve invited him specially, so he’s coming too.”

By this time I must have turned green… Then came the time for the talk and here are these monster minds in front of me waiting!

Last Friday Sean Carroll emailed me to ask if I’d give a short blackboard talk at an informal cosmology meeting they have on Monday morning at Caltech. My real talk was in the afternoon, so I said, sure, no problem. I thought I’d mainly be talking to grad students and postdocs, so I didn’t prepare anything. My plan was to give some background on entropy, information, black holes, etc. so that they could better follow the afternoon talk.

To my surprise, rather than a bunch of grad students I found monster minds arrayed around the big oak table in 469 Lauritsen: Carroll, Kamionkowski, Wise, Preskill, Politzer (Nobel laureate), Ooguri and Stanley Deser! No need for elementary background. I was kind of nervous at first, but we ended up having a lively discussion that lasted over 90 minutes — I pretty much covered my whole talk using the blackboard, and ended up giving it again using slides later in the day. We had a funny moment when I first started discussing the ADM energy of the monsters. I looked over at Deser (the “D” in ADM) and smiled; he smiled back and nodded slightly 🙂 Having Stanley in the audience helped a lot because the entropy packing I described depends on using negative gravitational binding energy to nearly cancel the energy of the constituent matter. He was quite familiar with these constructions and helped convince the audience that I wasn’t nuts.

Ten years ago I wrote a paper (unpublished) showing how to obtain a zero energy configuration in GR out of massive constituents. Particle theorists I discussed it with all thought I was crazy, but the referee was a very erudite relativist, who pointed out that a similar result (using different constructions) had been obtained by ADM, Novikov and Zeldovich, and others long ago. (So my result wasn’t really new, but at least it wasn’t wrong…) I had suspected Deser of being the referee but he said it wasn’t him. He thought it might have been Wald… 🙂

I had a wonderful visit, clouded only by the news (received by email on my cellphone while chatting with Sean) that Sidney Coleman had passed away on Sunday.

Written by infoproc

November 21, 2007 at 4:35 pm

Posted in brainpower, physics

Goldman OK?

with 9 comments

According to the Times, Goldman unloaded their subprime exposure:

NYTimes: … Late last year, as the markets roared along, David A. Viniar, Goldman’s chief financial officer, called a “mortgage risk” meeting in his meticulous 30th-floor office in Lower Manhattan.

At that point, the holdings of Goldman’s mortgage desk were down somewhat, but the notoriously nervous Mr. Viniar was worried about bigger problems. After reviewing the full portfolio with other executives, his message was clear: the bank should reduce its stockpile of mortgages and mortgage-related securities and buy expensive insurance as protection against further losses, a person briefed on the meeting said.

With its mix of swagger and contrary thinking, it was just the kind of bet that has long defined Goldman’s hard-nosed, go-it-alone style.

Most of the firm’s competitors, meanwhile, with the exception of the more specialized Lehman Brothers, appeared to barrel headlong into the mortgage markets. They kept packaging and trading complex securities for high fees without protecting themselves against the positions they were buying.

Even Goldman, which saw the problems coming, continued to package risky mortgages to sell to investors. Some of those investors took losses on those securities, while Goldman’s hedges were profitable.

When the credit markets seized up in late July, Goldman was in the enviable position of having offloaded the toxic products that Merrill Lynch, Citigroup, UBS, Bear Stearns and Morgan Stanley, among others, had kept buying.

“If you look at their profitability through a period of intense credit and mortgage market turmoil,” said Guy Moszkowski, an analyst at Merrill Lynch who covers the investment banks, “you’d have to give them an A-plus.”

This contrast in performance has been hard for competitors to swallow. The bank that seems to have a hand in so many deals and products and regions made more money in the boom and, at least so far, has managed to keep making money through the bust.

In turn, Goldman’s stock has significantly outperformed its peers. At the end of last week it was up about 13 percent for the year, compared with a drop of almost 14 percent for the XBD, the broker-dealer index that includes the leading Wall Street banks. Merrill Lynch, Bear Stearns and Citigroup are down almost 40 percent this year.

…At Goldman, the controller’s office — the group responsible for valuing the firm’s huge positions — has 1,100 people, including 20 Ph.D.’s. If there is a dispute, the controller is always deemed right unless the trading desk can make a convincing case for an alternate valuation. The bank says risk managers swap jobs with traders and bankers over a career and can be paid the same multimillion-dollar salaries as investment bankers.

“The risk controllers are taken very seriously,” Mr. Moszkowski said. “They have a level of authority and power that is, on balance, equivalent to the people running the cash registers. It’s not as clear that that happens everywhere.” …

Written by infoproc

November 19, 2007 at 4:27 am

A Caltech wedding

leave a comment »

I’m at Caltech this weekend, for the wedding of one of my college roommates and to give a seminar (Monday) on monsters and black hole entropy.

I’m glad one of us had the good taste to get married at the Athenaeum, the beautiful faculty club. Wonderful to be back on campus with old friends on a mild Pasadena evening. 20 years ago seemed like yesterday!

As the reception was winding down, we took a stroll down the olive walk to peek in on Interhouse, the annual 7 house bash that takes place in the fall term. During my time in Page house we engineered ambitious rides each year for the party, including a roller coaster, a sky tram over a lagoon in our courtyard, and a jungle raft ride along a flooded main corridor. I didn’t see anything quite as impressive last night, but the kids seemed to be having a great time. Someone told me the male to female ratio is nearly equal in the freshman class — quite a change from the old days 🙂

Earlier yesterday, coming out of the gym, I overheard the following conversation:

Undergrad 1: Yeah, they emailed me the test. I guess it’s honor code.

Undergrad 2: But it’s a hedge fund, right? Is there an honor code?

Undergrad 1: Hmmm…. good point.

Written by infoproc

November 18, 2007 at 6:35 pm

Posted in caltech

CDOh no!

with one comment

This Economist article summarizes the current situation nicely. Notice the doomsday scenario in which bond insurers suffer a ratings reduction, which has spillover effects in the muni and state bond markets. This was mentioned previously in a comment here but I hadn’t appreciated the seriousness of it before.

I suppose there is some muni-treasury spread index I should be betting on?

Over the years I’ve noticed a lot of my super-rich friends in fixed income keep a disproportionate (at least according to naive portfolio theory) fraction of their net worth in munis — probably because of the tax benefit and because fixed income people always think equities are airy fairy overpriced and can never bear to place a big bet there. I’m going to see if any of them are worried about spillover into their personal portfolios if the bond insurers take a downgrade.

CDOh no!

Nov 8th 2007 | From The Economist print edition

With trades scarce and losses mounting, it is going to be a harsh winter.

IT WAS not a good omen. This week Lewis Ranieri, a pioneer of mortgage securitisation in his “Liar’s Poker” days at Salomon Brothers, sold his property-financing firm because the subprime crisis had cut it off from fresh debt. If the industry’s godfather can’t navigate the storm-tossed markets, what hope its greedy children?

Banks that a few months ago were falling over each other to underwrite mortgage-backed securities and the labyrinthine pooling structures, known as collateralised-debt obligations (CDOs), that sit atop them, have admitted to more than $30 billion in losses. That figure is set to rise sharply as mortgage defaults in America climb. Citigroup estimates that big banks may be facing $64 billion in write-downs, excluding its own figures—and it was one of the top two underwriters of CDOs. Banks will be dealing with the pain for a lot longer than anyone imagined only a couple of months ago.

Most CDOs were engineered to provide both yield and safety, with a thick band of each rated AAA or even better, “super-senior”. Lower-rated tranches have been in trouble for months. But the prospect of a collapse in the value of the supposedly safe portions terrifies the banks—not surprisingly, since there is at least $350 billion-worth of such CDOs outstanding.

This looks all too possible now that rating agencies have started to downgrade AAA-rated CDOs, some of them by several notches (14 in the case of one notorious tranche). The agencies have given warning in the past month that they might downgrade another $50 billion-worth of top-rated CDOs, and that looks like the tip of the iceberg. One fear is that this leads to a wave of hurried sales, because many institutional investors are allowed to hold only AAA-rated paper. In addition, default notices have been issued on more than $5 billion-worth of CDOs, as senior investors try to grab what they can.

The uncertainty is compounded by the difficulty of finding a “fair value” for these complex instruments. The fall-back method recommended in a recent paper by the Centre for Audit Quality, an industry research body, is to employ “assumptions that market participants would use”, a technique known as “Level 3”, which becomes subject to strict accounting regulations in America on November 15th. But “Level 3 is not that useful,” confesses a risk controller at a big European bank. Banks have tended to use it as a bucket into which they throw any securities they find hard to value and then make an educated guess at the price. Among Wall Street firms, the soaring amounts of Level 3 securities now exceed their shareholder equity.

Finding a better indicator of market prices is no easy task, however. One measure, though an imperfect one, especially for CDOs, is the ABX family of indices. These relate to derivatives linked to subprime, which are traded even when the underlying bonds are not. The ABX indices are near record lows, having fallen precipitously in October. Even the top tranches are well below par value (see chart). According to Citi, some AAA-rated CDO tranches are faring even worse—at a mere 10 cents on the dollar.

Most banks are probably reluctant to mark down their assets that far. Citi and Merrill Lynch lead the list of shame, with combined write-downs of more than $22 billion. But others may just be slower in coming clean—even the teflon traders at Goldman Sachs. CreditSights, a research firm, estimates Goldman’s potential CDO-related charges at $5.1 billion, for instance. On November 7th Morgan Stanley said it would write down its assets linked to subprime by $3.7 billion. For the first time, there is serious talk of banking giants running short of capital.

European banks can expect more grief, too. UBS, a Swiss bank, has reportedly been criticised for booking its mid-quality paper at twice the level implied by the ABX index. Marcel Ospel, its chairman, faces mounting pressure to resign after the bank reported big losses on fixed-income securities in the third quarter.

Banks are not the only ones who need to worry. Hedge funds hold more than 45% of all CDO assets, according to the IMF. Insurers are exposed, too; American International Group, the world’s largest insurer, this week fell far short of earnings targets because of mortgage-related problems. In addition, one obscure but important corner of the industry faces a fight for survival over its subprime exposure: the specialist bond insurers.

In return for a premium, bond insurers guarantee repayment of interest on a variety of debt securities in case of default. Their mainstay used to be municipal bonds, but over the past decade they moved aggressively into structured finance. Before October, it was thought that the two biggest, MBIA and Ambac, would get away with losses in the low hundreds of millions. But the rating agencies’ assault on high-grade CDOs, the bread and butter of the insurers’ structured business, raises the prospect that they could run low on capital. Analysts at Morgan Stanley forecast combined losses for the two firms of up to $18.7 billion. Even the minimum expected loss, a much lower $3.3 billion, would be a huge blow for companies with combined equity capital of just $12 billion.

Some think the rating agencies will eventually have to strip the bond insurers of their cherished AAA ratings. They are loth to do this because it would “wreak havoc”, not only in structured products but across financial markets, says Andre Cappon, a consultant. New issues of municipal bonds could slow dramatically, since many borrowers rely on the insurers’ top rating to enhance their own creditworthiness. Over $1 trillion of debt issued by American cities and states—much of it held by retired people through funds—might have to be downgraded. Public-private partnerships in Britain, which are also customers, would also be affected.

For those holding CDOs, things could get worse before they get better. Tim Bond of Barclays Capital points out that defaults on subprime loans are still accelerating in America, particularly on mortgages made since 2006. This will take time to feed through to CDOs, via mortgage-backed bonds, but feed through it will. A consumer-credit slump, which looks increasingly likely, would clobber securities backed by credit-card and car loans, which are also pooled in CDOs. That would be all the beleaguered banks need.

Written by infoproc

November 14, 2007 at 4:43 pm

Gender differences in "extreme" mathematical ability

with 7 comments

Since the Larry Summers debacle I’ve kept my eye out for relevant data on gender differences in mathematical ability. Finally I’ve found some analysis of results from a nationally representative study of elementary school children (K-5). Interestingly, the larger variance in male math performance is already observed at the beginning of kindergarten — yes, before formal schooling has begun. By 3rd grade males are outperforming throughout the distribution, but the advantage at the high end is roughly unchanged. Note the authors consider 95 percentile to be “extreme” ability, which is kind of funny. You have to go quite a bit further out on the tail to find the talent pool from which professors of math, computer science, physical science and engineering are drawn.

Taking a quick look at their numbers, it appears that at the beginning of kindergarten the male distribution has standard deviation about 8 percent greater than the female distribution (larger variance — both tails are overpopulated by males), although means and medians are pretty much the same. This implies that, already at age 5, at the 1 in 1000 talent level there will be roughly 2.5 times as many boys as girls. This ratio becomes larger and larger as one looks at more elite groups — for 1 in 10k talents the ratio is something like 4 to 1 male to female. (I am extrapolating the normal distribution here, which might be a source of error.)

If subsequent societal effects were exactly gender neutral after age 5, one still might expect to find a strong asymmetry in gender representation in certain fields. Therefore, gender asymmetry in outcomes is not by itself evidence of discrimination at higher levels of the selection process. Removing gender bias at all levels, starting from kindergarten and continuing through grade school, high school, undergraduate, graduate and postdoctoral training, and, finally, faculty hiring, will not correct for the effect which is already present at age 5!

Note, I’m not claiming that the male advantage at age 5 is necessarily biological in origin — it might be due to environmental causes. If one believes the causes are entirely environmental, and if one wants to equalize the numbers of male and female math geniuses, then intervention had better begin quite early — extending to how mommies and daddies raise their infants.

In some other research by the same authors (I don’t have a web link), international scores on the TIMMS examinations show that at the 90th percentile in math ability among seniors in high school, the ratio of males to females varies between roughly 2-3. This is a much larger discrepancy than the kindergarten numbers (strongly apparent already at only the 90th percentile), although it would be hard to know whether it is due to biological causes such as hormones and differences in male/female development, or to societal causes. The fact that there is some variation between countries does suggest at least a significant societal component.

If you read this post carefully, you will see that I have done little more than interpret the results of the nationwide testing examined in the paper below. Nevertheless, I anticipate I might get into trouble for having the temerity to perform this simple analysis. Let me therefore state, for the record, that I do believe that societal effects tend to discourage women from achievement in math and science, and that we can do much better than we currently are in promoting female representation in math-heavy fields. However, I do not think that there is any data supporting a complete absence of gender differences in the distribution of cognitive ability.

Gender Differences in Kindergartners Mathematics Achievement! Evidence from a Nationally Representative Sample

Paper presented at the annual meeting of the American Sociological Association (to appear in Social Science Research)

Paret, M. and Penner, A., Dept. of Sociology, UC Berkeley (2006, Aug)

Abstract: Gender differences in mathematics achievement are typically thought to emerge at the end of middle school and beginning of high school, yet some studies have found differences among younger children. Until recently the data available to examine gender differences among young children consisted of small non-nationally representative samples. This study utilizes data from the Early Childhood Longitudinal Study, Kindergarten Class of 1998-99 to analyze differences in a nationally representative sample of kindergarteners as they progress from kindergarten to third grade. Using quantile regression techniques to examine gender differences across the distribution, differences are found among students as early as kindergarten. Initially boys are found to do better at the top of the distribution and worse at the bottom, but by third grade boys do as well or better throughout the distribution.

Written by infoproc

November 8, 2007 at 6:20 pm

Posted in brainpower, gender, iq