While I’m more of an Inkings guy (like the Dean of Oxford Business School, Peter Tufano), but in homage to BYU’s Dean Brigette Madrian and her Harry Potter obsession, I’m going to start a series of blog entries under the rubric of “Defense Against the DARQ Arts.” My goal is to be somewhat of an online course for those looking to understand Artificial Intelligence and Machine Learning. Hopefully, by the end of this series, you’ll be able to understand AI/ML more and suss out those that do know what they’re talking about, and more importantly, those that will succeed. Hopefully, it will even help you, the reader, succeed. In this particular series, I will concentrate on the DARQ Art of Optimization.
I will begin by relating several experiences over the years that I’ve had with investment management groups related to optimization.
CEO of One of the Largest Asset Management Firms in the World
Several years ago, at the beginning of the current AI/ML Boom, I was asked to meet with the CEO-designate of one of the world’s largest asset managers. I was there not as a consultant but as a courtesy to friends in the bank who provided me data for the ULISSES Project. The CEO-designate was smart, so much so that she asked to meet with me with her transition group and a Ph.D. in engineering from Berkeley (there to evaluate me). After an exhaustive search, in frustration, she turned to one of the biggest investment banks in the world to find her the top expert in AI/ML in finance, and they came back with me.
The CEO-designate started the meeting by saying that they had met with over 200 so-called AI experts and that fruitless search had brought her to me. Her advisor then said, “Tell us about Artificial Intelligence and Machine Learning?” I responded by saying that “I don’t feel comfortable using those general terms, so why don’t I speak about the underlying technologies?” The first technology I spoke about was optimization. At the end of our discussion, her advisor turned to her and said, “you can trust this man,” she said thank you and that she’d proceed in the way I suggested, looking to domain-specific experts with hard skills. She’s since ascended to run that firm, and I have great hopes for them under her leadership. Incidentally, it was because of that trip and that courtesy that we acquired our early Dell sponsorship of the ULISSES Project.
Head of Quant for one of the Largest Hedge Funds in the World
Some years ago, another one of our ULISSES Project sponsors asked me to meet with one of the largest hedge funds in the world. I’d advised this group on AI before and they hadn’t taken my advice, so I was not so keen to meet with them. The meeting was in the wake of very public problems that the hedge fund had with technology that resulted in considerable losses, including but not limited to significant losses in their “AI Investments.” The head of “quantitative strategies” had already engaged McKinsey to help them navigate their problems, and not surprisingly, in my meeting with him, he had a former McKinsey guy with him (who had consulted with them before joining them). Rather than asking questions as the CEO-designate had, the two men began by bragging to me that they had just spent $20 million implementing a new optimizer. I was suitably impressed until I asked some basic questions.
After our Q&A, it was clear that they knew nothing about optimization and had entered into the project thinking it was a “magic bullet.” I turned down the offer to work with them. In discussion with my ULISSES Project sponsor, he pled with me to work with the group, saying, “sure, they don’t know what they’re doing, but they are only leading the effort.” I responded that they not only didn’t have the right external consultant but then they’ve compounded the error by bringing one of them on full time. I saw echoes of the AOL-Time Warner merger all over again and investment banker who thinks Dan Bricklin is a “loser.” No, it would not work, and I didn’t want to be a part of it. It should not be a surprise that the group has not turned around.
CEO of a FinTech Company Purchased by a Major Financial Company
The last thing that I will cite was a discussion I had with the CEO of a “FinTech” company that a major financial company bought recently. I knew the guy well; at university, he was a mediocre student and couldn’t even program. He then started a FinTech company because he had a close friend that could give him backing. He called me on the phone not that long ago, and we got into a discussion. I get these calls all the time, basically people seeking free consulting. When he told me about his company, he used the word “optimize” and “optimizer” no less than 30 times—I know because I counted.
I was impressed because he even used the words in the right context. I mean, this guy couldn’t even have passed a grad business school OR class, let alone understand how to implement an optimizer, but keeping an open mind, I continued. Thinking perhaps he’d learned something over the intervening years, I asked him what optimizer he uses in his company. My query was met by dead silence. Upon further examination, it was clear that he didn’t know what he was talking about and that “optimization” and “optimizer” had become management buzzwords that he’d learned to master to sell his concept and, in turn, his company. I doubt the company that bought him will be able to recoup their investment. In the biz, we call this “vaporware.”
My Own Story – The Early Days
I was one of the early members of the Columbia Business School Internet Club. As strange as it may seem, it was something that was required by my dissertation advisor, with me acting as technical advisor for the MBAs. I remember advising people who came back to visit me after a few years and had made fortunes with their start-ups. As an advisor to these start-ups, I was asked to join many of them, but I turned them all down, pursuing my passion—investment technology, and looking to complete my Ph.D. This said it did not take too long until I caught the bug and I began building my own start-up. In my apartment’s second room, I ran a server room and became obsessed with building next-generation investment systems. Little by little, I tested my technologies, funding my work with consulting revenue.
One of my clients was the world’s largest investment manager. The Chair of my department, who had a deep CS background, was contacted by their head of research because they needed to implement “next-generation technologies.” Knowing that I planned on launching my start-up upon graduation, my Department Chair introduced me to the firm. Within days, they flew out to New York from San Francisco, and soon after that, I started working with them. They were desperate as they were going bankrupt and needed to work fast.
After about six months, I flew out to San Francisco to meet with the team in person, and I took it as an excellent opportunity to finally close VC funding for my start-up. You see, in that interim time of meeting my first big client, the Associate Dean of Columbia Business School had already introduced me to top VCs like Tim Draper of Draper Fisher Jurvetson, who’d met me at Columbia and offered to invest in my start-up. It was viewed as such a hot idea that pretty much everyone I met with offered to invest in the company, but I put them all on hold, telling them that I needed to graduate first.
Knowing my plans to start my company upon graduation, the head of research asked me to stop by San Francisco before my Palo Alto meetings, which I agreed to. In that meeting, we talked about optimization, and as he had an operations research Ph.D. from Berkeley and was the former head of the finance department, he knew what he didn’t know. As I described the proper use of new optimization technologies, he stopped me and said, “We need you. Please join us full-time and help us. You will be given carte blanche to implement your changes and will report directly to me.”
He continued that I would not only have all roadblocks removed but that I would also be the first person other than him to have complete access to the entire codebase, and I’d have a budget to implement my plans. I was shocked, but his plea won me over, and I agreed on the spot. I then had the difficult experience telling my potential backers that I was going a different route, pursuing a once-in-a-lifetime opportunity. I remember well one VC telling me the company I was going to join were “losers” and would soon be out of business and it was a foolish choice.
As hard as the discussion with the VCs was, it was more difficult when I got back to New York. I had already started building my team in anticipation of the launch, but it was clear that they wouldn’t be able to join me in the investment firm. As I was the “founder,” the company would never get funding without me, so after I explained my decision to them, I immediately went about finding them good jobs, something that was not difficult in the blinding hot tech market. Two of them even went on to be internet company CEOs themselves.
When I graduated and joined the company in San Francisco full time, one of the first things we did was implement an “artificial intelligence” technology I’d been following for years, an optimizer called CPLEX. I’d wanted to use the CPLEX optimizer for years, but had instead used a much more limited open source program developed at Stanford called MINOS. For context, before my arrival, the company had also been using MINOS. I argued to the head of research that he was getting what he paid for and that he should instead get the “best in the business,” one developed by one of the smartest mathematical programmers in the world, Bob Bixby. That decision to use CPLEX helped turn the “losers” into one of the world’s top-performing investment groups in the world and helped them start an entirely new and much more lucrative business, hedge funds.
I am starting this set of blog entries to help as many of you as possible understand optimization. Keep in mind that the CEO-designate and Head of Research had different backgrounds, but both knew what they needed and how to find proper advisors. These entries are not to help you develop the right buzzwords, but rather something to help you effect real and lasting change in your organizations. I look forward to the journey that lies ahead of us.