I spent the summer as an intern in the quantitative strategies group at Newfound where I was able to explore a vast array of topics in quantitative asset management, many of which appear on this blog.  Now that I am officially part of the Newfound team, I thought I would highlight some of the important aspects of traditional engineering that are relevant to financial engineering based on observations over the past few months.  After all, engineering has been around for a long time, long before the current, complex financial system.  By incorporating these foundational traits, we are drawing on a large knowledge base that was already gained through many trials, successes, and failures.

One way to define an engineer is anyone who systematically analyzes problems, often using quantitative methods, to devise practical, real-world solutions.  Whether this results in a bridge, oil refinery, microchip, prosthetic limb, solar cell, or search algorithm, an engineer was involved with much of the research and testing that went into it.

Financial engineering is very similar in that new products come about to satisfy investor needs.  Tranched mortgage backed securities, ETFs, exotic options, convertible bonds, and TIPs have all been created by people who noticed a desire for products that met certain specifications.  People are often quick to point out that some of these products have played a part in events like the Financial Crisis, the Flash Crash, and Black Monday; however, the probabilistic nature of our world and the human element behind our designs opens up all forms of engineering to the possibility of failure.  The Tacoma Narrows Bridge, a number of chemical/power plant disasters, and the Y2K panic come to mind.

So here are a few common threads that I see between traditional engineering disciplines and financial engineering.  You may notice some biases from my predominantly chemical engineering training, but as I am sure many of our readers have backgrounds in other quantitative fields, feel free to leave your experiences in the comments.

1.  Be cognizant of precision

When we calculate parameters from data, there is always uncertainty in those estimates.  Quite often, the data set is limited or of questionable quality.   Even with good data, the estimate has uncertainty since the data are drawn from a presumably unknown distribution.

Traditional – All manufacturing processes produce products with inherent uncertainty in their dimensions: bolts, car parts, toys, and building materials, furniture, etc.  If you have ever tried to assemble a new product only to find that two pieces don’t quite line up, you know this well.  Setting tolerances on a process is important for materials management and customer satisfaction.

Financial – Consider estimating the expected returns of assets when doing Mean-Variance Optimization.  Small changes in the estimates can lead to vastly different allocations.  Also, knowing whether the volatility of a stock is 15% ± 1% or 15% ± 5% may not change the numerical answer, but it will sure change the precision of the answer.  Propagating uncertainty from the inputs to the outputs yields valuable information on the viability of a strategy.

2. Optimize stepwise

Once an idea is conceived, engineers naturally look for ways to improve on it.  Having multiple ideas is very helpful, but what is the best way to test and implement them?  By applying the improvements stepwise, we can quantify the marginal improvement attributable to each modification.  This method also allows us to vary the order of application to examine any combination effects.

Traditional – Reducing the cost of power plant by totally redesigning the process is great, but what if a $100M savings is actually a $150M savings coupled with a $50M expense?  Isolating the effects of improvements can result in more savings.

Financial – Including more factors in a model may increase the accuracy at the expense of making the model less robust.  Models should be parsimonious and as intuitive as possible.  Building a model stepwise allows us to quantify marginal improvements.

3.  Know the parts that make up the whole

In conjunction with the previous point, if we start with a complicated system, it is beneficial to divide it into component parts.  This not only aids in understanding the product as a whole but also highlights any areas that can be simplified or eliminated.

Traditional – Chemical plants are comprised of many individual operations that are interconnected to use materials and energy in the most efficient way possible.  When these parts are disentangled during the design phase, they can often be improved individually before being fully integrated into the overall process.

Financial – Oftentimes, a quantitative strategy will have multiple steps that go in to producing the final output.  Understanding the investment universe, signal generation, and risk management portions leads to a better grasp over how the strategy will perform during different market conditions.

4.  Always keep the forest in view

In contrast with the previous point, a deep understanding of the parts cannot replace achieving the ultimate goal, which is a combination of these parts.  This is particularly important when work on a product is divided among people.   It is also relevant when analyzing data or doing research.

Traditional – Time is often a large constraint in engineering.  With unlimited time, improvements would keep happening because there is always something that can be made better, but there comes a point when work must stop so that the goal of producing the product can be realized.  Even large companies release products that are not perfect and follow them up with new versions.

Financial – If enough strategies are tested on a data set, one is bound to work.  Data snooping and overfitting result in views from of trees that won’t work in the general case.   It is essential to test strategies out of sample and not fall victim to confirmation bias.

5.  Be creatively practical

Engineering involves devising real-world solutions.  Thus, those solutions must be relevant, desirable, and cost-effective.

Traditional – My fourth-grade nephew and I were discussing hybrid cars one day, and he was very interested in how they operated.  The next time I saw him, he showed me his designs for improvements to the engine, aerodynamics, regenerative breaking, and battery.  Little did he know, he had devised a perpetual motion machine.  Solutions like this look fantastic on paper but won’t work in real life.   However, thinking creatively leads you to novel, achievable solutions, in the first place!

Financial – A historically backtested price chart may look great, but knowing the assumptions behind it is critical.  Transaction costs, tax implications, and fees can quickly turn a winner into a loser.  These practical constraints are the frictions that prevent financial perpetual motion machines.

6.  Carefully assess risks

Much time and effort is spent in both types of engineering to identify and quantify risks.  Probabilities of scenarios must be calculated or estimated, and systems are stressed to test the limits of the components.

Traditional – Oil refineries have parallel processing lines and spare pieces of critical equipment.  They are designed to withstand elevated temperatures and pressures when appropriate.  Trade-offs of capital costs and lost operating revenue are weighed to determine the most effective design.  Safety systems, which are designed for credible failure scenarios, are not perfect and will not work in every situation.

Financial – Diversification reduces exposure to idiosyncratic risks.  Portfolio insurance provides downside protection.  The extent to which these protections are applied comes from an assessment of possible failure scenarios and likelihoods.  VaR and expected shortfall are two tools that financial engineers use to assess the effectiveness of risk management, and as with chemical plants, risk mitigants are not 100% effective 100% of the time.

An engineer solves real-world problems that are being faced by real people.  Academic curiosities can be interesting and can shed light on areas of new research, but ultimately, the best solutions are the ones that solve common problems.   Just as public demand fuels traditional engineering solutions, investor demand drives financial engineering solutions.  By following these principles, financial engineers can provide investors with practical, risk-managed strategies that fulfill their needs and enable them to meet their investment goals.

Nathan is a Portfolio Manager at Newfound Research, a quantitative asset manager offering a suite of separately managed accounts and mutual funds. At Newfound, Nathan is responsible for investment research, strategy development, and supporting the portfolio management team. Prior to joining Newfound, he was a chemical engineer at URS, a global engineering firm in the oil, natural gas, and biofuels industry where he was responsible for process simulation development, project economic analysis, and the creation of in-house software. Nathan holds a Master of Science in Computational Finance from Carnegie Mellon University and graduated summa cum laude from Case Western Reserve University with a Bachelor of Science in Chemical Engineering and a minor in Mathematics.