I’ve received several comments from readers defending the usefulness of quantitative techniques. They are definitely useful and they have been widely adopted. Which is why its important to understand their limits.
Understanding their limits only improves their utility. For example, imagine that you’re at the hardware store to buy a blade for a circular saw. There are many different kinds of blades on display. Some blades are designed to work at higher speed and some are suitable only for softer materials. Other blades have more teeth and are suitable for finer work.
Imagine if there were no labels or packaging explaining what each blade was designed for and how to use it safely. Not only would it make using a circular saw harder, it would also make it much more dangerous.
In a similar way, my objective is not to argue that quantitative techniques are bad or wrong, but rather to make them easier and safer to use by considering their limitations.
Other comments have pointed out that “have a hunch, buy a bunch and go to lunch” (also known as fundamental) investors also have their limitations and weaknesses. Fair point. But it’s a topic for another time.
One reader opined that “all investors are quants, its just that we have different investment horizons.” Both quantitative and fundamental investors look for investments featuring attributes that history has shown to be effective way to improve returns/reduce risk.
The most interesting part of this opinion is the reference to different time horizons. This is what we’ll explore in greater detail in this post.
We’ve probably all heard or read the following quote from Benjamin Graham:
“In the short run, the market is a voting machine but in the long run, it is a weighing machine.”
What Graham is alluding to is the shift between the market paying attention to the signal (i.e. intrinsic value) and noise (i.e. emotion, short-termism, etc.) that happens over time.
I believe that quantitative techniques have the upper hand when noise dominates. In other words, the shorter the investment horizon, the more investors should rely on quantitative techniques.
This isn’t because quantitative tools are better, it’s because they’re more consistent. One of the biggest weaknesses of human decision-making is inconsistency.
For example, we know that valuation has a big impact on future investment returns. But cheap investments are often cheap for a reason. We need to also understand what is causing the investment to be cheap and why. Valuation is one variable among many, some of which matter more than others. In other words, investors need to weigh up the pros (i.e. cheapness) and cons (e.g. potential for bankruptcy).
A quantitative process will usually assign a constant weight to each variable, resolving investment trade-offs in a consistent way. The danger for humans is that the weight that they assign to each variable is not constant. Instead, it varies with their emotional and mental state, is influenced by their circumstances and the overall mood of the market.
To use Graham’s vernacular, we’re more likely to “vote” in the short-term.
This problem only gets worse when there’s more noise. Noise adds the additional risk that investors will mistakenly the noise for the signal.
In other words, quantitative techniques have the advantage over shorter periods not because they perform well, but because humans are likely to perform worse.
You’re probably thinking, aren’t professional investors smart enough to learn from their mistakes? Yes, they are smart. But they’re faced with what researchers Robin M. Hogarth, Tomás Lejarraga and Emre Soyer refer to as a “wicked environment”. Hoggarth explains that wicked environments are:
situations in which feedback in the form of outcomes of actions or observations is poor, misleading, or even missing. In contrast, in kind learning environments, feedback links outcomes directly to the appropriate actions or judgments and is both accurate and plentiful.
Daniel Kahneman builds on these ideas in his book Thinking Fast and Slow. He observes that a person can only really become an expert when two conditions are satisfied:
- An environment that is sufficiently regular to be predictable
- An opportunity to learn these regularities through prolonged practice
In the short-term the market is far too wicked for this sort of learning to take place. Here are a few examples:
- Bad economic news often triggers a stock market sell-off. Yet on another occasion, poor economic news triggers a stock market rally as investors anticipate central bank stimulus.
- Falling oil prices usually boost investor optimism for cyclical companies, as it is a major cost for many of these companies. Yet on another occasion a falling oil price is seen as a sign that global economic growth is stalling, causing the performance of cyclical stocks to suffer.
- The stock market reaches an extreme valuation-level, only to continue becoming more expensive for several more years.
You get the idea. There are few, if any, iron-clad cause and effect relationships in investing. But as Graham reminds us, the few relationships that do exist only really emerge over the medium-to-long term.
Consequently, the advantage starts to shift towards fundamental investing as the investment horizon lengthens and the ratio of signal to noise improves.
The longer the time horizon, the smaller the relative disadvantage suffered by fundamental investors. Casual relationships start to emerge. For example, valuation, which tells an investor next to nothing in the short-term, starts to matter.
This may be what our reader mentioned at the outset was referring to. Both quants and fundamental investors aim to take advantage of historical relationships. it’s just that the types of relationships that they focus on are different.
For example, high-frequency trading is a field that’s dominated by quants. Even discretionary day traders rely very heavily on quantitative tools to back test strategies and to identify potential trades.
On the other hand, quantitative techniques are largely absent from investing in unlisted assets. I’ve yet to hear of a quantitative private equity strategy or an unlisted real estate strategy. And as we’ve already considered, quantitative techniques aren’t much help with strategies that involve a changing future, or low-frequency, high impact opportunities.
In summary, quantitative strategies require data, breadth and are better than humans at filtering out short-term noise and evaluating investment opportunities consistently.
What does this all mean for fund managers? It provides a framework to assess future viability of their business.
For example, value investing relies on historical (assets, normalised earnings) and current information (prices). There are thousands of stocks listed on stock exchanges around the world. And the investment horizon of most clients is less than 3 years. That’s 3/3: 1) historical information, 2) breadth and 3) a short-to-medium term time horizon.
The implication is clear: a large number of fundamental value investing strategies are at risk of being replaced by simple, rules-based quantitative strategies.
Fundamental managers need to find a niche. If you’re a fundamental value investor, it’s probably a good idea to focus on special situations, bankruptcies, spin-offs, busted IPOs, activist investing or some other niche.
Relying on screens (which is ironically a simple, rules-based quantitative thing to do) to perform most of the heavy lifting makes about as much sense as a guerrilla army choosing to fight a set-piece battle on open ground. Leave the benchmark plus or minus x% at the stock, sector and country level, tracking error of 3-5% stuff to the quants.
A similar principle applies to growth managers. They need to move beyond simply using obvious proxies for growth that quantitative strategies can easily access, such as revenue and earnings growth, profitability and price momentum.
How about investors? It provides a framework for thinking about how and when to pay for active management. Simple, rules-based quantitative strategies are usually cheaper. It makes sense to use them in situations where its hard to argue that humans have a clear advantage.
One reader remarked that the quant vs fundamental debate is nuanced. I couldn’t agree more. Its why developing a frame work to assess the relative attractiveness of each approach is so important. The best results are often achieved when you use the right tool for the job.