One of the early quants at former Wall Street heavyweight Salomon Brothers, Dr Lance Smith has a rich mathematical background, but believes there is much more to risk management than the mathematical models.
“A tsunami strikes Japan, or a Brexit strikes the UK," says Smith. “Things happen that are not in the tea leaves of historical data. On the other hand, physics is repeatable. Given the same starting point and conditions, you can predict the outcome with well-defined and known probabilities.”
“Not so with the markets,” he adds. “You don’t get do overs. Traders make bets that are tilted in their favor, and the key is to make sure that you can survive the negative surprises and reap the benefits of making such bets.”
How did you get from academic mathematics to selling risk management technology?
I was a math professor at Columbia University when I left the ivory tower for Salomon Brothers on Wall Street, where I served as the chief quant for the equities prop trading desk from 1986 to 1992. I was one of several academics at that time, they called us “rocket scientists,” who were making the move to Wall Street.
As you can imagine, it was a huge transition, and a great way to learn the rough and tumble of the markets (as well as the corporate culture) my own way. I had the opportunity to build models from the ground up – bringing a sophisticated mathematical approach, but also accommodating the realities of the marketplace.
That required transparency into the models so that the traders could understand very quickly what implicit assumptions were being made, and what could happen if market conditions changed and the assumptions turned out to be incorrect. As I tell my quants at Imagine, it’s not physics. It just acts like it is, until it doesn’t, and that’s when the large sums get made or lost.
In 1992 I left Salomon with a close associate of mine at Solly – and a year later we had set up our own company (along with two others with complementary skill sets) to develop risk management software for prop desks and the fledgling hedge fund industry. That was 25 years ago.
In 2000, Imagine launched our ASP cloud-type service – not called the cloud back then – and we began offering software + services, including data. This dovetailed very nicely with the rapid expansion of the hedge fund market, as it was pretty much turnkey.
What strengths does your mathematical approach bring to risk management software?
I’m a pure mathematician. That is, I have a broad background in multiple subfields, some of which are not ordinarily associated with problem solving in this industry. But this means I can take a fresh approach to a problem, and this sometimes results in more accurate modeling with higher performance. The underlying assumptions may be similar, but the resulting algorithm is more efficient. A simple but good case in point is the binomial option pricing model. It’s a wonderful teaching paradigm, but actually not the most efficient or accurate way to price an option.
So I’m always looking for approaches that are not beholden to assumptions of an earlier time, when the industry requirements and the technological capabilities were much simpler.
Indeed, what we see in the industry today is that everything is moving towards more and faster real-time risk and even real-time VaR. In response, Imagine has redesigned our architecture and even the models themselves, so that we can not only meet the current performance requirements but also the more stringent ones yet to come.
We also recognized that in order to be successful across a wide range of players, we would need to provide flexibility and extensibility. No two firms look at risk management exactly the same way. A hedge fund is quite different from a bank or an asset manager. And tightly- regulated entities need to respond quickly to new regulatory requirements.
For a locally-hosted system, most vendors provide API’s, which we do as well for such clients. But what do you do for a cloud-based service for which there is a single version being deployed across all users?
Our first response back in the early 2000’s was to provide the ability for our clients themselves (or our support team) to create “custom columns” in which they could combine existing calculations using basic logical and arithmetic functions to create new outputs. This is much like writing a formula in Excel. This ability, by itself, was unique to our offering for probably a decade, and it enabled our clients to create on-the-fly their own elaborate stress tests within a few hours. The new calculations existed only in their private database, and was visible only to them.
Then, a few years ago, we took our capabilities to entirely new level with the introduction of the IFP – the Imagine Financial Platform.
Now our clients can actually write programs (apps) on top of our system. They have full access to the database and analytical functions, and now more than 95% of them use the IFP to extend our offering in ways that we hadn’t imagined – and not just for risk management.
That is, not only can they add their own models, but they can also write reports that project upcoming cash flows, reports for the regulators, analyze executions and so on and so on. We use the IFP ourselves to do such work on their behalf, as well as to add on new capabilities – truly agile development.
Is cloud adoption the way forward for the risk industry, particularly if it’s providing this agility?
I think so. There are enormous cost-cutting pressures on the financial industry, as tightened regulatory reporting and capital requirements have raised the cost of doing business. So how does one cut costs while meeting the higher requirements? Well, for one thing, there is a lot of data that is common to everyone, yet still maintained separately by many institutions, for no financial benefit.
And the industry has changed. In the old days, each firm had its own models for derivative securities. That seems kind of quaint now, because today the regulators do not want to see firms using “proprietary” models. In the place of secret sauce, they require transparent and well-documented analytics. These realities support a cloud offering.
Firms will still have to allow for different clients making different assumptions about their inputs to the models, such as underlying volatility, and so a cloud offering needs to come with a lot of flexibility as each user customizes for their own strategies – and indeed their current book of positions. In other words, even though there is a level of consensus as to the underlying risk models, the cloud offerings cannot be “one size fits all.”
And, of course, internet-based offerings themselves are also evolving. While DLT is increasingly being adopted, there will be enormous savings in back office operations and potentially new and innovative ways to view risk. At Imagine, innovation has always been a central part of our corporate culture, and we plan to be ready to service and take advantage of these new emerging capabilities.