Proinsias O'Mahony: Investors must learn the lessons of history

There’s a fund for everything these days. Unprecedented computer power and ever-increasing access to data means more and more apparent market patterns are being discovered. But are these patterns real?

Would you be confident if a fund manager told you their strategy did well over a 10- or 20-year period, but they had no idea how it performed prior to that?

Perhaps not. But would you be reassured if you were told the strategy is a seemingly timeless one that has lasted for more than 200 years? Such strategies exist, according to fascinating new data unearthed by quantitative strategists at Dutch asset management firm Robeco.

In recent years, researchers have claimed various “factors” such as value (broadly speaking, buying cheap assets), momentum (buying outperforming assets) and seasonality (buying at certain times of the year) have managed to outperform the market over long periods. Funds have attempted to profit from these market anomalies by developing smart-beta funds – rules-based exchange-traded funds (ETFs) that aim to outperform markets by focusing on specific factors.

Trust your fund manager to buy stocks but not to sell

---

Past performance no guide to future for chief executives

---

Stock market turbulence: is US recession risk rising?

---

Assets that did relatively well over a particular month in the past were found to be 'likely to do relatively well in the same month going forward'

However, the amount of backtesting going on means the number of so-called factors has soared in recent years. Fluke results are being mistaken for genuine discoveries, according to award-winning research conducted in 2016 by Prof Campbell Harvey from Duke University, who warned that most finance research was false after finding no fewer than 316 factors in the finance literature.

Shoddy statistical testing is not confined to finance. The same problems have been found in medicine (including cancer research) as well as psychology and economics but the scale of the problem means researchers need to become stricter in their methodological approach, the researchers say.

Statistical geekiness aside, the problem with many existing studies is they typically go back “only” 30 or 40 years, says Robeco. In contrast, its new study goes all the way back to 1800, testing the long-term performance of six different factors on four different asset classes – stocks, bonds, currencies and commodities.

The assumption was that many of the factors traditionally associated with market outperformance would disappear once tested over such a long time frame. In fact, the opposite is the case, with the researchers finding “consistent and ubiquitous evidence for the large majority of global return factors”.

Of 24 factors tested – six factors across the aforementioned four asset classes – 19 turned out to be consistently effective.

Best strategy

So what works? Trend following can sound simplistic. Broadly speaking, it refers to investing in assets that have been trending higher over the previous year, an idea reflected in Wall Street adages such as “the trend is your friend”, “cut your losses and let your winners run” and “buy high, sell low”.

The idea you should invest in something simply because it is rising in price is perverse to contrarians like Howard Marks, the billionaire founder of Oaktree Capital, who has said that "superior investors are the exact opposite" to trend followers. However, the data suggests otherwise. Trend following has actually been the best and most reliable strategy over the last two centuries, consistently delivering superior risk-adjusted returns.

Trend following worked in each of the four financial markets – stocks, bonds, currencies and commodities. Researchers and fund managers often emphasise the difference between trend and momentum (in stock markets, momentum refers to equities that perform well relative to others) but the Robeco study suggests they are not distinct sources of returns, with the trend factor tending to “subsume” momentum.

The next best performer was the carry trade – that is, borrowing money at low rates to invest in higher-yielding assets. Today, the trade is most commonly associated with the Japanese yen. Interest rates have been negligible in Japan over the last two decades, resulting in traders borrowing in yen and investing the money in higher-yielding currencies.

The trade is not without risk – critics often compare it picking up pennies in front of a steamroller because while it works most of the time, violent reversals sometimes result in steep losses. However, the new research makes clear that carry trading was taking place long before the opportunities presented in modern Japan and that it is by no means confined to the currency markets, with strong returns also reported in the stock, bond and commodity markets.

Value investing worked in three out of the four asset classes (the one exception was currencies). Investors are unlikely to be surprised by the apparent success of value investing, which has long been extolled by esteemed investors like Warren Buffett and the aforementioned Marks, although eyebrows might be raised at the findings pertaining to seasonality.

Seasonal strategies like “sell in May and go away”, the January effect (the apparent outperformance of stocks in January) and the turn-of-the-month effect (seemingly stronger stock returns at the start and end of the month) have long been laughed off by many fundamentally-minded investors.

However, the results indicate a strong seasonality effect in financial markets; assets that did relatively well over a particular month in the past were found to be “likely to do relatively well in the same month going forward”.

Additionally, the oft-repeated academic assumption that you have to take higher risks to earn higher rewards doesn’t hold up. In stock markets, low-risk stocks deliver better risk-adjusted returns than high-risk stocks, according to the study.

Implications

Certain factors have, it seems then, been a pretty reliable guide to stock market outperformance for some time now. The question is: why?

Many strategies that worked in the past no longer do so

Broadly speaking, there are two competing explanations for market anomalies such as the value effect. Believers in market efficiency, like Nobel economist Eugene Fama, tend to argue that value stocks are cheap because they are riskier.

In contrast, behavioural finance experts like Robert Shiller, who shared the Nobel with Fama in 2013, tend to argue that human emotions like fear and greed drive market mispricings. To assess whether the results "represented compensation for well-known economic or financial risk factors", Robeco analysed returns during both good and bad times. Looking back over two centuries meant there were 74 recessionary years and 143 expansionary years; 43 bear market years versus 174 bull market years.

Importantly, the differences were not significant. The outperformance associated with the various strategies was “consistently present across various macroeconomic states”, suggesting there is “very limited evidence of a link between macroeconomic risk and global return factors”.

Does this mean the same strategies will continue to outperform in the future? Not necessarily. Many strategies that worked in the past no longer do so. Generally, when investors become aware of market mispricings, they devise strategies to take advantage of them, resulting in the mispricing disappearing over time.

Factor investing is big business today and the growing popularity of this approach may well lead to such strategies losing their edge. Indeed, the disappointing performance of many value strategies over the last decade has led to many questioning the future the viability of the approach. At the same time, many factors that delivered the goods for investors two centuries ago have continued to do so in recent decades. It may well be that they continue to work for one simple reason – flaws in human nature that result in emotion-driven investing.