Finally, the continued advance into markets less populated by quants will prove a rich seam-corporate bonds stand out to me but are by no means the only opportunity. Most of the data is probably useless, some of it is gold. Unsurprisingly, the enormous amount of new data becoming available gives us many alpha opportunities. These new strategies-many of them non-linear and employing machine learning-will be difficult for most market participants to develop and will certainly not be immediately apparent. So, paradoxically, for the foreseeable future, progress in machine learning may depend even more on people-people with an interest in financial applications and great skill in computer science, statistics and modeling.Īt one level the future for quant investing is easy to predict: The alpha produced by simpler, well-understood, linear strategies will erode while newer, more sophisticated strategies will generate higher alpha. Systems talent and a systematic approach are required to develop the complex infrastructure that enables data aggregation, analysis and computation reliably and at scale.Įqually important are the highly experienced, intelligent, and creative researchers who can select from the plethora of machine learning techniques available and develop approaches that generate true value. Second, as more financial services firms embrace artificial intelligence techniques like machine learning, it will become apparent that human talent and effective teaming are key to unlocking their promise. Making this shift will require private market investors to approach their work through a very different lens, but we believe that the rich opportunities in certain private markets more than justify the effort. Others are applying sophisticated data science methods to increase the growth and profitability of their portfolio companies. Some firms are already exploring ways to source and analyze deals quantitatively. A next logical step is to extend this methodology to data-rich areas like private equity and venture capital. We’ve all seen how these approaches can be effectively applied to investing in publicly traded securities. In other words, the future of quantitative investing should include a continued energetic search for the new but also include a lot more of the past than many realize.įirst, I think we’ll see quantitative approaches-meaning those grounded in advanced data science and systematic techniques-play a growing role in private market investing. Well-known factors are well-known for a reason, and we should not toss them out lightly. New factors are more susceptible to data mining compared to well-known ones, like value, that have strong long-term evidence across many asset classes and geographies. Trying to improve our factors and find new ones does not mean we should discard the old ones. But the opposite danger, of overreacting to recent performance, is at least equally as important. It is always important to keep an open mind to the possibility that the world might have changed. While we have generally cautioned against too much “factor timing,” today we recommend a modest tilt toward value. In other words, value appears currently quite cheap compared to history-more like a shunned out-of-favor factor than one too many people are chasing. Instead, we find that the current value spread is among the widest in history. If value has been “arbitraged away,” we’d expect to see the value spread-the price difference between expensive and cheap stocks-narrow to smaller than historical levels. Real things, like becoming too expensive-yes, perhaps because of this widespread knowledge-can. But simple knowledge doesn’t kill a strategy. Many types of quantitative factor-based investing strategies, particularly the value factor, have had recent tough times, leading some to question if it is “broken.” A common critique argues that value no longer works as too darn many people know about it.
0 Comments
Leave a Reply. |