The crash in bitcoin and other cryptocurrencies Tuesday was a stark reminder of the dangers of overconfidence.
As podcasts, blogs and videos were filled with can’t-miss predictions of bitcoin imminently hitting $100,000 or even $1 million, traders took on tremendous amounts of leverage to buy, buy, buy, convinced they could pick the next big winner.
When the crash came, some investors got wiped out, their heavily leveraged portfolios unable to bear a downswing that had seemed unimaginable days earlier.
Overconfidence is endemic to financial markets. Overconfident investors trade more frequently, and the more they trade, the more they underperform the market. They are also more likely to commit common investment errors such as under-diversification and overconcentration on familiar stocks, and investor overconfidence is often a contributing factor to market bubbles and crashes, like the 2008 financial crisis.
While there is a lot of academic research documenting overconfidence, pinpointing which types of investors are more overconfident (men, young people), and exploring the devastating consequences overconfidence can have on individuals and society, much less attention has been paid to why it is so prevalent.
In one sense, it is surprising that investors tend to be overconfident, because trading provides so much feedback and opportunity for learning. If traders are not consistently beating the market, how and why do they maintain the belief that they will do so in the future?
Overconfidence is usually explained by appealing to information-processing errors such as confirmation bias. For instance, a trader might take credit for a winning trade but blame bad luck for a losing trade to maintain the belief in his or her trading prowess.
In new research published in the Proceedings of the National Academy of Sciences (PNAS), we highlight a different explanation: biased memory for past performance.
The link between traders’ faulty memories and overconfidence has been suspected for some time. Erik Davidson, a former chief investment officer at Wells Fargo, once said: “Much like our human predisposition toward nostalgia about the past, where we only remember the good times and gloss over the bad, investors likewise tend to take a nostalgic view of their past winners but forget about their past losing investments.”
It turns out he was right, and our work is the first to scientifically document this phenomenon.
Two kinds of memory bias
In our research we asked investors to recall the performance of their most consequential trades — both winners and losers — over some time horizon, such as the past year. An investor might tell us that they bought and sold 100 shares of Apple for a 75% return, for instance.
Next, we asked them a series of questions to figure out their level of overconfidence and how frequently they trade. Overconfident investors were sure they would beat the market by a huge margin going forward and were likely to report making very frequent trades. Finally, we asked them to access their financial statements and tell us the true performance of those most impactful trades.
We then compared the figures they reported from memory and the true figures from their financial statements. We found two kinds of memory biases which we call “distortion” and “selective forgetting.” Distortion means that people’s reported returns from memory were positively biased on average. Winners were remembered as having a more positive return than reality, and losers as having a less negative return. Selective forgetting means that people were more likely to remember winners than losers.
Critically, we also found that participants with larger memory biases were more overconfident and traded more frequently. This result suggests that biased memory likely contributes to overconfidence. It also may explain why investors can maintain an inflated sense of their own abilities even in the face of evidence to the contrary: Their self-image is determined more by warped memory than actual results.
Our research also identified a possible remedy, a simple intervention that mitigates memory bias and thereby reduces overconfidence. In this study, half the investors simply looked up their actual returns at the beginning of the experiment instead of toward the end, providing no opportunity for biased memory.
We assessed participants’ overconfidence and gave some of them an opportunity to take part in a trading experiment with $500 to invest. They were asked to pre-commit to how many trades they planned to make in the experiment and had to pay $10 for each trade.
Participants who looked at their returns first were less overconfident and spent less of their $500 stake to pre-buy trades, indicating that they planned to trade less frequently. So just looking at past performance before making decisions appears to reduce overconfidence, though it didn’t eliminate it completely.
This finding suggests that brokerages could mitigate overconfidence by making trading history and relevant comparisons more easily accessible when their customers are making trading decisions. They could, for instance, provide a data dashboard showing how past trading decisions have played out relative to some reasonable benchmarks like a buy and hold strategy or the performance of common indices like the S&P 500 or Nasdaq 100
In the past, many brokerages may have been disincentivized to do this because they made significant income from trading fees. With the proliferation of low- or no-fee trading at brokers including Robinhood and TD Ameritrade , data provision like this is likely to be mutually beneficial to brokers and traders alike.
It’s common for our view of the past to be overly rosy. For instance, memory distortion was found among college students who remembered their high school grades as being higher than they achieved in reality and among patients who recalled their cholesterol scores and cardiovascular risk categories as more favorable than shown on a recently viewed test.
Those effects are typically attributed to people’s desire to maintain a positive self-image, which, of course, is important. But the ego-boosting benefits of biased memory about investing performance carry a significant cost, and most investors we’ve spoken with would rather have accurate beliefs and better returns.
The implications are clear: Don’t trust your memory. If you do so, you are likely to have an inflated sense of your own abilities, and this bias can lead to big mistakes, such as taking on too much leverage and failing to sufficiently diversify.
We recommend keeping track of past investing performance and studying your record on occasion, especially before making a big bet. You will be in a better position to protect and grow your wealth if you have an accurate view of your own abilities.
Nostalgia has its place, but it shouldn’t determine how you manage your investments.
Philip Fernbach is a professor of marketing and the director of the Center for Research on Consumer Financial Decision Making at the University of Colorado. He is co-author of “The Knowledge Illusion: Why We Never Think Alone.” Daniel Walters is a professor of marketing and the director of the Marketing Insights Lab at INSEAD.