Why this matters: Process is more important than approach

Why this matters is a blog series introducing content topics for the Time Summit. For information on the conference, visit timesummit.org.

There was an opinion piece recently posted on Bloomberg titled “Why the Quants Aren’t Adding Up.” In it, Satyajit Das describes why the recent relative underperformance of quantitative strategies (by his measure) is indicative of fundamental problems inherent to the technique. It’s the best, most vivid example I’ve yet found of confusing process (the steps taken to make a decision) and approach (the tools chosen to implement process).

Systematic or quantitative describe an approach, not a process, and by themselves don’t guarantee some relative outperformance. In fact, I’d argue they have nothing to do with performance. But, for some (including the author), these words have come to promise it, creating expectations that won’t be met on average. I wouldn’t be happy to hear my credit card’s customer service department was now systematic, so what’s different about investing?

The issue is that systematic or quantitative funds are naturally assumed to be scientific. However, that’s not necessarily the case. To be scientific is to use evidence to solve problems. It’s a process descriptor. Within investing, evidence is the tricky part, and it’s exactly where good processes, irrespective of approach, are separated from bad. Just like an artist’s tools don’t always predict the quality of his or her work, we should be careful to separate the methods chosen from the way in which they are applied.

Of the authors criticisms (save for a potential misapplication of hindsight bias), most concern process and are fair, to a degree: quantitative strategies are susceptible to overfitting, can lack transparency, may become crowded, and might be overpriced. But so can discretionary ones. A fund manager with discretion might overapply their experience, participate in crowded trades, charge too much and lack transparency. These are all, again, process descriptors having little to do with an approach, style or implementation. There is no reason to believe that a particular approach would yield sub-par processes on average. The author makes a bold claim about a complicated topic based on very little evidence, which is, ironically, a bad process.

Unfortunately, process often doesn’t receive the same attention as approach, but it’s probably more important. Bucketing a method by the tools chosen is easy: systematic, trend following, relative value, thesis driven etc. But good process, on the other hand, is hard to summarize. You almost know it when you see it. It’s a mix of culture, knowledge and attention to detail resulting from meticulous hard work. It’s not necessarily exciting, which makes it hard to sell. A manager can talk at length about their new machine learning models or a new macro idea, but will the same attention be given to a carefully maintained database that tracks revisions and publication dates as to avoid lookahead bias? Possibly not.

At this year’s Time Summit, in a session we’re calling “What we’re getting wrong with data driven strategies,” we plan to address a selection of vital issues of process:
* The multiple testing problem
* Forward testing and implications for risk management
* The reality that we have much less useful data think we think

Knowing such topics demand a careful and experienced delivery, we’ve worked hard to secure a group of some the industry’s best. We’re incredibly excited and look forward to sharing details soon.

While we eagerly await every Time Summit, 2019’s event will likely deliver our most important content yet, making good on our promise to leave attendees with actionable ideas that make a difference.

Despite its weak moments, I think there’s a lot to learn from this article, namely that allocators should prioritize sound, scientific process over particular approaches. While there are areas of the investment universe where the breadth enabled by the systematic approach is, probably, the most efficient implementation, other areas, especially where data is scarce or hard to model, are likely more suited to discretionary ones.

Additionally, managers of systematic product should be aware of the implicit promise and natural assumptions sometimes conveyed by their method. Not every quantitative fund is Renaissance, nor should it try to be. Relative performance and relative cost exist within a continuous space with opportunities for many kinds of winners. However, fostering realistic, honest expectations is always a good idea and key to finding long term roles within allocator portfolios.