Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Joe Mezrich is the founder of Metafoura, and the former head of quantitative equity research at Nomura Securities International.
The “quant winter” that enveloped quantitative mutual funds in 2018 turned into spring in late 2020 — when Treasury yields bottomed – and has blossomed into summer as they headed upwards. That’s probably no coincidence.
That three-year spell of terrible, no-good, very bad performance still haunts quants. But data hints that the quant winter was only a season in what could be called a quant climate change — an ongoing trend lasting decades.
We can infer a rough long-term trend for performance by analysing the foundation of quant investor portfolios — namely factors like value, size or quality. Rather than primarily focusing on stock selection, most quants focus on owning a portfolio of factors, constantly refining them and experimenting on combinations to find the optimal mix for returns.
For example, an investor could own a long-short value factor that would be long a portfolio of cheap stocks by some value measure while being short a portfolio of expensive stocks. Similarly, the investor could also own a profitability factor, a momentum factor, and so on. The quant investor’s task is to define and combine factors to perform well as a portfolio.
There are many equity factors, but most investors and academics agree that besides the market (beta) between four and six factors are enough to capture what drives equity returns. My approach is to use the widely accepted factors in the Fama-French model — value, profitability, size, investment — plus momentum.
Ken French provides returns for these factors in his data library with history back to 1963. If you calculate the average of these five-factor returns for every month of the history and equal-weight them then you can think of this as a passive benchmark factor portfolio that factor investors can either own or try to outperform.
NB, this factor portfolio is not the benchmark against which quant investors actually measure their performance. But it’s a helpful construct for analysis. If there is a trend in the five-factor portfolio performance, that should produce a corresponding trend in quant performance.
Below you can see the 10-year rolling average of the five-factor monthly average. That is the green line in the chart. The blue line is the yield on a constant maturity 10-year US Treasury bond.
The trend of the factor average is clear, as is the curious relationship between factor returns and Treasury yields.
After rising in the 70s and the early 1980s, the average return to the portfolio of five equally weighted factors has steadily declined for over three decades (The jump in factor returns after 2000 and the collapse of the line a decade later is somewhat of an artefact of the rolling 10-year window, as the impact of the tech crash entered the rolling window and left a decade later.)
Notably, the trend in factor portfolio returns has co-varied with the path of Treasury yields. You can see the quant winter and the recovery in 2020 in the chart’s green line, which connects to the path of interest rates in the blue line.
In the mid-1980s and the first decade of this century, the passive long-short factor portfolio return averaged about 7 per cent yearly. But by the beginning of this decade, that return had dropped below zero before recovering — the quant winter and more recent thaw.
This suggests that the connection between factor returns and Treasury yields is over half a century old. Why?
It’s hard to say for sure, but last year Niels Gormsen and Eben Lazarus published an intriguing paper titled Duration-Driven Returns, which argues that there is a cash flow duration factor at the heart of many other factors — a kind of One Ring To Rule Them All.
Or as Gormsen and Lazarus put it:
Across a broad global sample of 23 countries, risk factors based on value, profit, investment, low risk and payout invest in firms with low growth rates. This common feature is sufficiently pronounced that the risk factors can be summarised by a single factor that invests in low growth firms. We refer to our new factor as a duration factor. We do so because the firms in the long-leg of the factor not only have low growth rates but also a short cash flow duration.
The good news for quants is that the long slide of rates to zero has ended, and rates have reverted to levels last seen before the 2008 financial crisis. A replay of the quant winter is unlikely — as long as rates don’t collapse again.
