I was speaking with the risk manager at a new shop when he stopped me mid-sentence and asked, “why wouldn’t I just build this myself in Excel?” Now I’m used to the question, as most of us in this profession are quantitatively savvy and have at least one person in the firm that is a power Excel user, but this was different because I could see that he was ready to engage Alpha Theory immediately or go back to his desk and start building his own version of Alpha Theory. My first reaction was to explain that when I was an analyst at a hedge fund, I built my first version of Alpha Theory in Excel. As I developed Alpha Theory I kept coming across hurdles created by the Excel limitations for which I knew a true software solution was the only answer. Below are some highlights of why Excel does not work, the complexities of building a full solution, and most importantly, the positive ROI of using Alpha Theory off the shelf:
Cost and Time to Build – Alpha Theory is the culmination of over 20 man-years and millions of dollars in design and development. For many funds, the money variable in the equation gets a small weighting, but the time is another matter. A few dedicated resources will be required to steward the process but do not forget that the portfolio manager and analysts will continuously be involved with design and testing. Their time is precious and their mental capital is better allocated on analysis, not software building.
It’s Complicated! – We all have smart people on our teams so yes you can figure out many of the challenges but there is also a possibility that some of these hurdles may render the system ineffective. Let’s go through a few challenges:
+Time Horizon. How do you deal with short dated returns versus long dated returns? You can’t use text book annualization because they produce wildly inaccurate return profiles over very short timeframes. Additionally, how do you handle losses in short time periods? Let’s ask a question, would you rather lose 20% in 2 days or 2 years. The gut reaction is 2 years, but that’s incorrect because you would rather get the loss behind you. I’ll ask it another way, would you rather have $0.80 two days from now or $0.80 two years from now. Dealing with these challenges in determining returns is complicated and a challenge that Alpha Theory has solved for you.
+Portfolio and Sector Exposure. Do you care about total portfolio gross and net exposure? How about sector exposures? If so, then let’s take a portfolio with constraints of 200% gross/40% net, global region exposure maximums of 50% gross/30% net, and industry exposure maximums of 40% gross / 20% net. Assume you have lots of good ideas and your portfolio of research exceeds many of these constraints. How do you construct a portfolio that maximizes risk-adjusted return while paying heed to each of these constraints? The answer requires an optimization function. To do this in Excel you either need to buy incremental software or create kludge Solver functions. The optimization function is inherent in Alpha Theory.
+Extreme Loss Constraints. Not all returns are created equal. The width of the distribution has a dramatic impact on the long-term geometric return of the portfolio for two assets with the same arithmetic risk-adjusted return. For example, assume you could only make one bet over and over for the rest of your career but they both had a 20% risk-adjusted return. Bet #1 has a 100% chance of going up 20% each year. Bet #2 has a 50% chance of going up 90% or 50% chance of losing 50% each year. Each bet has a 20% expected return but which one do you prefer and how should it change your position size? Well if we assume that we have $1 today and we make each bet sequentially 10 times in a row we would end up with $6.20 from choosing Bet #1 and $0.77 from choosing Bet #2. This explains the potentially damaging effects of wide-distribution returns and why Kelly Criterion is the optimal method of choosing bet size. Alpha Theory includes this dynamic in position sizing and continues to research new ways to improve the long-term geometric return of the portfolio.
+Other Complications. Alpha Theory has spent exhaustive time investigating improvements to representing research and constraints in the form of position size, including market correlation, liquidity, loss constraints, analysis confidence, market-implied expectations, differences between longs and shorts, and many more factors that go into position sizing. These are challenges that every homegrown solution will have to traverse. Any cost-benefit analysis of growing your own solution must include these intricacies.
Collaborative Environment – To foster an effective solution the system must encourage multiple users. Excel is a closed environment that generally allows one user at a time and in not conducive to personalized views. An analyst, a portfolio manager, and a risk manager will all look at the system differently and be tasked with different elements of maintenance. They need their own customized views that highlight the variables to which they need to manage. Alpha Theory is an open architecture which allows multiple users simultaneous access to the system at any time and any geography. Excel is not built to allow this level of collaboration and synchronization and will falter as the organization tries to create a holistic solution.
Three Dimensions – Excel is two-dimensional which makes it difficult to have much variance in research structure. For instance, one analyst wants to describe the distribution of returns in a simple Bull and Bear case, but another has a more complex representation which has a Bull and Bear case but also a possibility of Black Swan and takeout. These are real scenarios that should be encouraged in the research process. Alpha Theory’s multi-dimensional platform allows analysts the flexibility to describe their research in ways that best represent the true byproduct of their research.
History – To maintain a history of research in Excel, firms will keep snapshots of the system going back in time. This is kludge way to retrace the firm’s investment process. Alpha Theory keeps a record of change made by analysts and portfolio managers and can show a consolidated history of changes.
Maintenance – Once the first version is complete the second version should be under way. A system is a constantly evolving organism that requires constant feeding and training to keep up with the demands and challenges of the firm. There will need to be dedicated resources for the system and the portfolio manager and analysts’ time will be sucked into continuous testing and design, just as with the initial build.
Return on Investment – Alpha Theory points out inefficiencies in the portfolio where the firm’s research and constraints do not match their position sizing. For most firms this can represent the largest source of lost return and unnecessary risk. If Alpha Theory points out a couple of inefficiencies per month it will return hundreds of basis points of incremental alpha over the course of the year versus the 1 or 2 basis points of cost for a medium sized fund (even lower cost for larger funds). Even without pointing out a single mis-sized position, Alpha Theory improves the investment process by allowing analyst to describe research in the form of distributions that the portfolio manager can use to make portfolio decisions. No matter how you measure it, the ROI is wildly positive.
Continuous Innovation – Alpha Theory releases updates to the product on a quarterly basis with improvements that continually enhance a firm’s ability to manage the investment process. These enhancements come from internal development and input from partners and clients. Alpha Theory spends 100% of its time dedicated to this concept. An internally developed product would have to focus dramatic energy to keep pace with Alpha Theory’s development and industry-driven ideas.
By the way, the risk manager at the beginning of this post decided to go and build his own version of Alpha Theory (probably because I did not have this article in my hip pocket). However, after a few months, he hired us as a consultant to help him get over a few of the challenges above. Finally, after a year he decided to scrap the whole process and use Alpha Theory. Now granted, we added several features between our initial conversation and when they finally implemented Alpha Theory, but the moral is to proceed with caution if you plan to build your own Alpha Theory.
