From Color Science to Wall Street

You may remember from the 1980s a vector model of color by Guth, Massof, and Benzschawel. The last author is Terry Benzschawel, a noted color-vision psychophysicist at Indiana University, Berkeley, and Johns Hopkins. But Terry has spent the past two decades as a Wall-Street “quant.” Here Terry writes of his journey:

For much of my life, I have wondered how and why I perceive myself as separate from my environment and other people. Studying the human visual system provided a perfect opportunity to think deeply about the relationship between mind and body.

My post-doc trail took me through psychology, optometry, ophthalmology and engineering. But when I failed to secure a faculty position by my third post-doc, I became so despondent that I quit my last position and remained unemployed for nearly a year. Finally, I answered a New York Times ad, “Scientists – Earn Big $$$ on Wall Street.” Upon meeting me, the recruiter told me that I was “totally unsuited for a career in finance.” To her surprise, a mathematician consulting for a prominent bank picked my resume out of a stack, interviewed me, and offered me a job. Thus, my career in finance was launched.

I was unprepared for the financial world. Although I was expected to master the financial literature and terminology, my environment didn’t support that effort. I had to compete with people ten years my junior who had been preparing for finance for their entire career. Most of my immediate superiors had less education than I did. I had to overcome my Ph.D. arrogance and acknowledge that there are many very intelligent people in the world without Ph.D.s.[1] Also, I had to give up publishing my research. Models similar to the proprietary ones I developed were published independently by academics several years after mine were in use.

My first job in finance was on the ill-fated 78th floor of the World Trade Center. My boss had a genetic algorithm to predict the likelihood of corporate bankruptcy. I was given information about the model only on a “need-to-know” basis. This was frustrating to me, but guarding information is common in the business world.

After about a year, my boss’s contract was terminated and I faced unemployment. However, my original recruiter quickly found me a position building neural networks to detect fraud on credit card transactions. Having built non-linear models of the visual system, I was able to build a successful network model that was used in the company’s fraud early-warning call center. Still, salaries and promotions were frozen, so I was dissatisfied. While on vacation in 1992, I met a managing director at a bond trading house. He passed my resume to their Fixed Income Arbitrage Group, featured in Michael Lewis’s book Liar’s Poker. I interviewed, was offered a job, and gleefully accepted.

With prospects of wealth and glamour in the famous “Arb group,” I began the Associate Training program. The Arb group was engaged in “proprietary trading”, risking the firm’s money, in contrast to their larger broker/dealer “sell side” business. In my second year, my direct supervisor resigned and my job suddenly worsened. Things got better after I built several successful models for pricing risky debt in emerging markets and we traded on those models. In 1998, after a corporate takeover, the Arb group was disbanded and we were all fired. The firm found me a job as a trader/strategist. I built neural network models and traded U.S. Treasury securities and the Mexican Peso while applying my credit models to help our customers manage their credit portfolios.

By 2002 I had gained some notoriety and began to travel the world visiting clients while building a research group. The great liquidity boom of the new century was on and I was riding high, helping clients manage their risk. Unfortunately, my firm didn’t apply my methods to manage our own risk, but instead offered my wares to induce clients to buy our products. One advantage of working on “the customer side” is that I was encouraged to publish my work for clients and, at last, in journals and at conferences.

During the past decade, I have coordinated the recruitment and training of Ph.D.s for the firm’s “quant” groups. In that role I travel to major universities and give talks about our firm. I speak with hundreds of talented young prospects each year and review resumes of several times that. Supervising young staff, both interns and full-time hires, has been a satisfying aspect of my job. Having temporary help, such as interns, has allowed me to do more speculative work I do not have to justify to the trading desks. I also coordinate a weekly seminar series featuring speakers from our firm and faculty at major universities.

In late 2008 I become a partner, called a “managing director”---no small achievement for someone of my temperament. There were challenges. With the financial crisis, much credit business was lost or curtailed. During this period, I’ve made myself useful by applying my methods to help manage risk within the firm. Only recently, as market activity returns, I’m back helping clients manage their credit portfolios.

The markets are relentless. They open every business day and proceed regardless of one’s mood or personal problems. Workdays are consistently long. Personally, I have had a lot to learn both about finance and life. I still do.

Even after 20 years, I sometimes view myself as an academic “spy,” probably because my ambitions are atypical for this business. My interests are not always on the direct track to short-term corporate revenue, so the road to partner was longer than typical. But because I established a pipeline of speculative projects that have come to fruition, I have bought the freedom to explore issues not directly related to our trading business.

I am grateful for my past and present opportunities. Much of my time now is spent on innovation and my mental life is as stimulating as it was when I was a vision scientist. I am certain that this is rare for someone in finance. I continue to have a passion for learning and enjoy collaborating with talented younger people.

Terry L. Benzschawel

[1] See Emmanuel Derman’s book My Life as a Quant. My experience resonates with what is written there and I found the book entertaining.

The lipstick smudge that betrays color infidelity

Have you ever been a subject in a color-matching experiment? If so, you may have encountered…

by Michael H. Brill, Datacolor

The Maxwell spot is an entoptic image of the eye's macula, a yellow-pigmented retinal area extending 3 or so degrees about the center of fixation. Until this year I regarded the Maxwell spot as an arcane effect that I would never see. Reportedly the spot is inconspicuous because it is fixed to the retina and hence the retinal receptors adapt to it. But even with rapid fading of the spot, I still should have seen it transiently in moving my gaze, say, from a blue sky to a white sheet of paper. But that didn’t happen. The paper showed me yellow journalism, but never a yellow spot. Ethan Montag [1] gave a demo (alternating blue and yellow field) to show the Maxwell spot---but no guarantees. (Evidently Montag also found it hard to see.) Also, Montag's demo shows the spot as a dark smudge on the blue field or a light smudge on the yellow field. It's still not yellow.

Then, twice in the past year I saw the Maxwell spot, both times in the context of a white light created by three narrowband LEDs. In neither case was the spot yellow. It was rather like a pink lipstick smudge on a white collar---betraying color infidelity by interfering with my ability to match colors. What a nuisance!

I first saw it when looking at a broad white surface in a light box that simulated daylight by mixing LED illumination. Several light mixtures flashed on and off in sequence, and curiously the “three-band lamp” always revealed a pink smudge for a few seconds. Could it be spatial inhomogeneity of the three-band lamp? No, the smudge covered less area when I got closer, and it always was centered about the direction of my gaze.

I saw it again at the latest IS&T/SID Color Imaging Conference. Abhijit Sarkar (a PhD student at Technicolor Research in Rennes, France and University of Nantes) gave what was judged to be the best student paper at the conference, on devising observer categories to reduce observer metamerism. He performed abbreviated color-matching experiments on multiple observers, using two 3-primary displays powered by different primaries. The observer categories he found did not agree well with the age dependency found by earlier investigators. As an on-site demonstration, Sarkar brought a 10-degree matching setup powered by a pair of LED triads, with wavelength peaks (452, 508, 642) nm and (462, 522, 592) nm. I was amazed how difficult it was for me to make the match, because the left-hand semicircle always had a fuzzy pink spot that faded away when I attended to the right-hand semicircle. When I backed away from the apparatus, the left-hand side of the match appeared uniformly purplish-pink. This latter effect had been noted by Sarkar. I thought we were seeing the Maxwell spot, and Mark Fairchild agreed.

Why is the spot called yellow and yet looks pink? Because the macular pigment absorbs strongly in a broad band about 450 nm [1], it would appear yellow when transilluminated by a full-spectrum daylight. When there are gaps in the light spectrum (as with 3-band lamps), attenuation of the green band can enhance the relative weight of the red, hence we see pink.

Not all three-band lamps show the effect, but Sarkar’s left-side green wavelength (508 nm) is low enough to be highly absorbed by the macula, leaving the 642-nm red primary to predominate. Because the G primary carries a lot of luminance, lack of that luminance in the Maxwell spot makes the pink darker and enhances my perception of it (relative to the yellow I'd managed to escape all the rest of my life).

Jack Moreland [2] describes a related way to reveal the Maxwell spot: “A large bipartite field (14 deg square) is presented. The two half-fields are approximately matched in colour: the appearance being a near-white. The mixtures are cyan and reddish-orange (490 + 610 nm) on the left, and blue and yellowish-green (460 + 470 nm) on the right […] An observer sees [a] patch about 3 or 4 deg in diameter [that] changes from ‘pink on green’ (left) to ‘green on pink’ (right) on switching gaze between the two half-fields.” So the Maxwell spot has shown itself to be pink to other eyes before mine.

Together with the best-paper prize, Sarkar now has a new factor to consider in selecting LED primaries. Also, I begin to understand how color-matching subjects must feel when told to "ignore the Maxwell spot." When the spot is lipstick-pink, that task is hard enough to make one consider “cosmetic” surgery.

[1] Ethan Montag, JIMG 774: Vision & Psychophysics, Chapter 8, Part 3: Parts of the eye.

[2] Jack D. Morehead, Entoptic visualization of macular pigment, J. Physiol. 485, 4P-5P (1995).