Here is Part 2 of our summary (focused on portfolio management tid bits, of course) of an interview with David E. Shaw in Peter J. Tanous’ book Investment Gurus. For additional background and context, please see Part 1.
“The purpose of a portfolio optimizer is to trade off risk and return in some predefined way, and to try to come up with a portfolio that’s close as possible to optimal in terms of those risk/return criteria…The optimizer knows things like transaction costs, the hedging of various risk factors…So it constructs a portfolio that is nearly optimal with respect to some risk/reward criterion, and then it modifies it continuously as new data come in…As an example of what such a program might do, the optimizer might find that one security looks under-priced relative to all these other different instruments, but if I actually bought that security, then I would have to much exposure to automobile stocks, and I would also tend to be short interest rates. I might also be making an implicit bet on economic cyclicality, and if the economy started to go bad, I might lose money that way.”
“…in practice you never really look at an isolated trade. You have to look at the whole universe. We use an optimizer that takes into consideration all the factors we know about…both for predicting profit and also for minimizing various sorts of risk. That doesn’t mean eliminating them. It’s the sort of thing you were describing, where you analyze the influences on a given stock, get information on all of the related stocks, bonds, options and so forth, and then construct a portfolio that tries to get as many of those risk factors as possible to cancel out. But it doesn’t happen in a simple way. Everything relates to everything else.”
The Optimizer is a computer software/algorithm programmed to understand the relationship between risk and return. How those terms and that relationship is defined, remains a mystery. Shaw is very secretive about their process. Nevertheless, the computer constructs the portfolio based on certain risk/reward factors – an idea akin to how fundamental human investors approach portfolio construction.
No (wo)man is an island. Similarly, no risk is an island. In the integrated world of today’s market economy, all risks are related in some way, which makes identification and hedging of risk factors an extremely difficult and delicate task. To “simplify” this process, Shaw talks about getting “as many of those risk factors as possible to cancel out” thus hopefully decreasing the number of transactions necessary to implement a hedging program and thus lowering associated costs as well.
Discount Rate, Opportunity Cost
“…the optimizer also knows about the cost of capital, and it’s not likely to get very excited about something that would tie up a lot of capital for a long period of time.”
What exactly constitutes the discount rate or cost of capital? For companies, it’s the weighted average cost of capital (WACC). One could argue that the calculation of this figure is quite fuzzy, especially with the presence of equity in the capital base. And moving further along the complexity scale, what is the cost of capital for (unlevered) investors?
If we removed from our minds the WACC formula carved deep from years of textbook finance training, the concept of cost of capital becomes quite interesting to think about.
Shaw references how The Optimizer associates the cost of capital with time. Based on traditional finance theory, this is because investors have a time preference of receiving cash today vs. tomorrow, and thus need to be compensated for the time delay. Traditionally, the risk-free-rate is used as the compensation figure. However, if we use the risk-free-rate, how then do you calculate the cost of capital for a portfolio of various international assets (where the risk-free-rate of the underlying holdings varies by country)? Do you take into account the effect of currency fluctuations for each country?
I have also heard some people reference the cost of capital as an opportunity cost figure. If so, does it change based on available returns provided by other opportunities? But if the cost of capital is a relative figure, how then do we calculate the cost of capital for those other investments? It becomes a rather recursive process…
However intangible, and regardless of differing definitions and calculation methodologies for the cost of capital, it is still a very real and necessary consideration in the investment, and portfolio management process. Stay tuned for some juicy bits on discount rate / cost of capital from Seth Klarman.