In your paper on momentum crashes, you mentioned this dynamic strategy to help mitigate such crashes. How has your thinking evolved on that, especially in the light of your recent work?

“To me this was a fascinating paper. I noticed that momentum doesn’t seem highly effective in Asian markets, particularly in Japan. In research with Toby Moskowitz, our crash detection algorithm was tailored for the US. But when we applied it to Japan, it revealed some unexpected results. When our algorithm highlighted certain periods as probable crash periods and we excluded these, we found momentum does actually work in Japan.

The other thing is that momentum is resilient across various asset classes. Once you try and eliminate those potential crash periods, it shows solid performance. So it’s really a robust thing. But a static momentum strategy might not be the way to go. Instead, gauging the current market conditions and determining if it’s the right moment for momentum is crucial. This was the core essence of our research.

Factor timing is always frowned upon, and for good reason because it opens you up to problems with data mining, but from a model-driven Bayesian perspective, a little bit of factor timing can be a good thing.”

Regarding your research on inattention and short-term factors, how do you view the short-term reversal effect, analyst revisions, and flow data? Do these account for some of the short-term dynamics you discussed?

“Historically, it’s striking that a short-term reversal factor, when rooted in sound economic analysis and ignoring transaction costs, going back to the 1980s, yields an annualized Sharpe ratio close to 8. It’s ridiculous. Implementing such a strategy back then, with today’s knowledge, would have been exceptionally profitable. Although I’m not sure that hindsight always offers a valid analysis!

Essentially, with short-term reversal, you’re identifying buy or sell actions that aren’t linked to relevant news. For instance, if a company’s stock price rises significantly on an earnings announcement day, there’s typically no short-term reversal. It continues to rise. The same holds if the company’s industry or other relevant news supports the price movement. To predict a short-term reversal, you have to analyze a price movement against all available information. If there’s still a large residual – after accounting for all known factors – no reversal is expected.

But, in the last few decades, the potential profits from this approach has gone way, way down. Only firms with advanced technology might find this strategy worth it due to trading costs. For others, it’s more of a tool to gauge when to trade – like waiting to buy a stock after it peaks and before its anticipated drop. But on the other hand, this is something that’s very distinct from momentum or post-earnings announcement drifts.”

You mentioned the book-to-price signal possibly becoming less effective recently. What are your thoughts on the current trend, both in academia and practice, of factoring in intangibles and capitalizing R&D expenses for valuations?

“A lot of papers these days on the academic perspective capitalize R&D expenses, take some fraction of SG&A, and try to determine its contribution to brand, knowledge, and organizational capital. By calculating these capitalization measures and making adjustments to the book value, it turns out that metrics such as book-to-price or book-to-market prove to be much more effective.

Alongside colleagues like Tano Santos, Lira Mota from MIT, and Simon [Rottke], we’re on a mission to reconcile what traditional value and fundamental investors are doing. Because when you think about it, what they do is a fancier version of this, right? They assess the real value a firm has built up, including its organizational and knowledge capital, and consider the potential of these assets to generate high returns.

The concept of ‘moats’ – that is, the barriers to entry protecting a firm – is also critical. You really want to adjust the book value not just for the worth of the intangible capital but also for its potential to churn out high returns, because that’s what’s going to really determine a firm’s value. This holistic approach mirrors how value investors think; and as quants, we want to be thinking more along those dimensions as well."

When considering investment decisions and evaluating what’s worked or hasn’t, how do you guard against action bias? For instance, value was considered ineffective before 2021, but then it rebounded. If you had shifted strategies prematurely, you might’ve missed out. How do you navigate such scenarios?

“I think you can be aware of these biases; everything you do has to be scientifically based. So when you see value not working, there are several potential explanations. Maybe the strategy is flawed. Maybe market dynamics have shifted unpredictably. Or maybe technologies took off in a way that we couldn’t have anticipated.

If you evaluate your model and find strong evidence that it’s defective, it would be rash to just discard value investing or flip it around to growth investing. You also want to consider other explanations. It’s crucial to thoroughly understand what happened and incorporate those findings into your strategy.

Looking back, not factoring in intangibles was an oversight we should have addressed earlier. Some firms, showing remarkable resilience, persisted with traditional value investing but adapted their approach. I kind of admire their courage. One firm experienced massive outflows – over 50% – due to consistent underperformance from their bet on value. However, in recent years, they’ve flourished as value investing regained traction. This is something we saw in the late 90s when many firms got clobbered from value strategies but rebounded over time. And I think that what you can’t do is just blindly abandon a strategy.”