Rafet Eriskin of AP4: Investor Bias [4 of 5]
To eliminate bias, investors must face the emotional pain of confronting their mistakes, and embrace technology as a partner in the journey.
Bias is an inevitable part of investing. However, better data helps us see through bias and generate more alpha.
Rafet Eriskin of AP4 has long quantified investment biases among the managers he’s invested into using Novus. In some cases, he even shares insights and conclusions with managers, who in turn implement changes in their investment process, yielding positive results for both themselves and investors.
In the third installment of our series on Rafet Eriskin, Senior Portfolio Manager at AP4, we saw how embracing digitization led to higher returns, lower resource utilization, and a more pleasurable workflow. We called it adopting a Portfolio Intelligence credo. In this installment, we’ll look at how Portfolio Intelligence can help generate higher returns by making us aware of—and helping us counter—behavioral biases.
Overconfidence Bias in Investing
Are you a better driver than average? If you’re like 93% of American drivers, your answer to that question is “Yes.” Of course, it’s impossible for 93% of people to be better than average at anything—most people believe they’re better drivers than they actually are.
The same phenomenon occurs with investing. A study by James Montier found that 74% of fund managers believed they are above average at their job. Of the remaining 26%, most thought they were average. Almost no managers Montier surveyed believed they were below average.
When investors are overconfident, they miss out on opportunities to add additional alpha to their portfolios. This is because they are unwilling to face up to their limitations, which leads to sub-optimal patterns of decision-making.
During our work with portfolio managers, we often ask them to recall the top three contributors to last year’s performance. Their eyes light up, and they throw themselves at illuminating us with the story behind their top contributor. When we ask them about the top detractors, a veil of oblivion descends on their faces. Most can’t even recall more than one. Why are we treating equally important information in such a drastically different way?
In my recent conversation with Rafet, he noted, “We’re mentally wired to protect ourselves from the pain of looking at our own mistakes. But if we can’t see our own mistakes, we can’t learn from them and improve.” His solution to this natural human aversion has been to embrace data and analytics that reveal not only his own biases but those of the fund managers he works with. And he’s been quite vocal about sharing those with fund managers. Read on to see how the story unfolds.
Nobel Prize-winning economist Daniel Kahneman has argued that our confidence in the decisions we make is largely determined by a story we are telling about ourselves. “The bias toward coherence favors overconfidence,” he writes. “An individual who expresses high confidence probably has a good story, which may or may not be true.”
Many investors make decisions and judge performance based on the story of returns. If an asset or manager has shown good returns for the past few years, the story tells us to keep investing. Moreover, if the returns are good, the decisions must have been right, and therefore the investor is successful. It’s a very simple storyline.
As many investors realize by now, returns don’t tell a complete story, only a (somewhat arbitrary) snapshot of a portfolio’s status. They don’t tell us the impact of specific decisions, or whether a skill set was consistent over time. It’s akin to measuring the success of a general in the middle of a war by the number of territories occupied, without taking into account the number of troops lost, the isolated effectiveness of various tactics, or the overall strategy at work in the general’s campaign. (Incidentally, let it be clear herewith that we despise war, conflict and violence as a way of resolving disputes, or as a way of life in general. It just so happens that war and military analogies offer very visual and poignant way of expressing things.)
A Richer Narrative
Novus has enabled allocators like Rafet to move beyond returns and consider data that sheds light on the decision-making process. Metrics like batting average (the number of times you were right divided by the number of times you tried) and win/loss ratio (the amount of money you gained from winning positions versus the amount of money you lost from losing positions) give investors insight into actual process of building a portfolio, and whether biases exist. As Ari Kiev once said, batting averages are indicative of how good you are a “selecting what to manage,” while win/loss ratios tell you about “managing what you selected.” Further, these figures can reveal missteps and inconsistencies that otherwise would remain buried.
Among many frameworks we use in our daily work with portfolio managers, the five-element Novus framework provides a very powerful, generalizable way to quantify investment skill-sets and investment biases. It quantifies manager skill along five degrees of freedom:
- Exposure management
- Category rotation
- Security selection
- Position sizing skill
- Trading skill
To quantify, for example, whether the manager suffers from endowment bias (i.e., hanging on to profitable positions beyond necessary, until they turn sour), we compare the manager performance to a simulated portfolio where positions are equally weighted at the beginning of each month, as in Figure 1 below.
If the manager’s portfolio consistently beats its equally weighted version, the manager shows good position sizing skill, and there is little or no endowment bias at play. There is another variation of this methodology that it determines how much earlier should you sell positions, depending on the presence of an endowment bias. This methodology also allows us to quantify how many bps the manager added (or detracted) through skill.
Another tool we use at Novus, Event Analysis, reveals how managers react to certain events—say, an earnings announcement or a merger—and how the market behaves after the manager’s reaction. By looking at events such as price increases of a certain size, and measuring whether the manager is correspondingly increasing position size, we can quantify if managers tend to chase returns, another significant bias often at play. See Figure 2, below, for an example.
In this example, we are looking at a position’s traces right before and right after it was exited. This way we can see if there is a trend around the event’s “exiting a position.” This allows us to both quantify manager skill around exiting positions and test whether the manager is affected by behavioral biases when making investment decisions. Our proprietary methodologies (illustrated on the right of Figure 2) guide users to distinguish signals from noise.
Incremental Alpha Through Behavioral Analysis
Rafet once showed a manager a Novus analysis demonstrating that the number of small positions in their portfolio was excessive and detracting (a classical “distract and detract” phenomenon resulting from overconfidence bias). The manager proceeded to re-jig their investment profile to capitalize on the insights—and the manager has added 300 bps of alpha to their historical average ever since.
Investors like Rafet who are willing to take a hard look in the mirror and perform this deeper analysis are exceptionally better off. The hardest part is usually making the decision to face up to your biases. As Rafet observes, “Investing in technology that reveals harmful biases is a net positive—it’s the emotional cost that keeps some people from adopting new systems.”
Our independent analysis confirms this conclusion. We estimate that institutional investors could add up to eight percentage points of alpha per year, on average, if all elements of their portfolio were run optimally, and all existing biases removed. And while optimal performance may be difficult to sustain, a significant movement in this direction is possible, as the success of AP4 illustrates.
View Part 1 of this series, Selecting Managers
View Part 2 of this series, The New Transparency
View Part 3 of this series, The Digital Revolution in Asset Management
View Part 5 of this series, Simples vs. Complex Models