I am a big fan of quantitative investing, although that term is often misapplied in my judgment, so let me define what I mean by that term. When I say quantitative investing, I am simply referring to a process driven approach to evaluating securities and making decisions.
‘Quant’ investors typically rely on computers to organize and evaluate data that serves as input in the decision making process. The financial press often overstates quant investing by making broad statements about ‘using computers with sophisticated models to identify predictable patterns.’
A statement like that makes it sound like the computer is running the show and that if you just have enough processing power, you can find and trade foreseeable trends that others can’t see. I would say it might happen on occasion, but that’s too narrow a definition.
A broader definition would include a lot of the things that we do for clients. For example, we use something called ‘mean-variance optimization,’ or MVO, to set our strategic asset allocation targets. We make return, volatility and correlation estimates and use software to spit out an ‘optimal’ portfolio.
That’s considered quantitative, but it doesn’t require a lot of computing power – after all, it was developed in the 1950s before computers were really on the scene. It’s true now that the models are more powerful and can evaluate more securities and asset classes, but it’s just a model where the old ‘garbage in, garbage out’ concept still applies.
When I was first getting going in this business, I thought that the math behind MVO was the key to understanding. I’m glad that I got my fingernails dirty figuring it out, but what I came to realize is that it relies on our estimates of future returns, volatility and correlation.
Who says that our estimates, or anyone else’s for that matter, are worth a darn? I can make the model output anything I want by adjusting the inputs and assumptions.
You could also say our approach to value is quantitative. We don’t hire actively managed mutual fund managers like Marty Whitman at Third Avenue to manually go through SEC filings, tear up the financial statements and make judgments about what stocks are cheap.
Instead, the firm that we use simply buys the 20 percent of stocks that are cheapest based on classic fundamental measures like price-to-book and owns them by their market capitalization (bigger companies get more of a share than smaller companies).
There are small tweaks of course and I am oversimplifying, but the basic idea is pretty simple. And, despite (or perhaps because) of the simplicity – it works.
But there’s no predictable pattern, at least in the short run. Over long periods, we can say that value stocks have outperformed, but we can’t say that about the future. We have a theory for why value stocks should outperform, but we can’t use computers to prove anything.
What I like about our quantitative strategy is that it’s transparent. I own a fund that I would never recommend to clients, mostly because it’s not transparent at all. It’s an opaque black box – I have no idea what’s going on inside.
Based on information that the fund company provides, I do know that the fund will likely lose as much as 20 percent in any given 12-month period. But because the fund is designed to be uncorrelated with stocks and bonds, when it falls, there won’t be a good explanation for why it’s falling. I won’t be able to tell whether it’s a normal, expected decline or whether the black box is broken.
I also won’t have a theory for why it should do well. When stocks fall, I can say that a stock is a business and as long as businesses have a way to make money, stocks will recover. I can’t say the same thing about mystery meat.
Why do I own it? To learn. Just like I spent a lot of time working out the details of an MVO model only to find out that the math isn’t the important part of the model, I am learning a lot with this goofy, but intriguing fund. For me, the education is worth the potential loss, but I’m only willing to do that with my money, not yours.
When we recommend something to clients, we have to be willing to invest in it ourselves. Strike that – we have to want to own it for ourselves. We eat our own cooking and when we want to make a change, we will lead with our own money.
For that to happen, we have to have good data, a sound theory and the ability to understand what is happening so that we can make a judgment about whether a fund or strategy deserves money. And as much as I like process driven strategies, no computer can replace human judgment.