When our silicon brethren finally take over all securities trading, will there still be evidence of biases and empirical anomalies in market prices, or will the rational robots elbow out us earlier, more emotional, more behavioral lifeforms?
Like you, I’ve ridden the behavioral finance roller coaster for decades. I used to be a purely rational guy. I thought behavioral finance was a joke. Pick a decision-making bias, pick a price pattern, see if they match, lather, rinse, repeat. If people can be underconfident or overconfident, obviously you can explain both mean reversion and momentum. If people can’t tell that “Linda is a feminist bank teller” is a subset of “Linda is a bank teller,” then people can probably believe almost anything.
Then I took my first behavioral finance course. The instructor was Richard Thaler, one of the fathers of the field, and, at the time, a future Nobel Laureate. He turned me the way a foreign asset is turned by master spycraft in every 80s movie. I’d never been turned before or since. I didn’t know I was about to be turned. I thought I was just going to take a fun course. But on the first day, almost his first sentence, I abandoned all my previous rationalistic loyalties. “Behavioral finance,” he said, “is really just open-minded finance.”
How can you argue with a statement like that? You’d have to admit to being closed-minded.
His argument was that conventional finance assumed a very specific type of investor following a very specific definition of rationality. Maybe there’s better models of human decision-making? Very few mothers bounce their children on their knee and teach them how to calculate expected utility. Most people are born and raised to rely on heuristics. We have consistent biases. Almost all of us experience a loss as more painful than an equal amount of gain; very few go the other way. Combine our systematic biases with some limits to arbitrage, and behavioral finance can become an important explainer of market phenomena.
But what if humans get replaced by machines in the markets? Sure, one machine against a billion humans won’t make a difference, but what about when it’s all machines and no humans?
The answer may come down to bananas.
People think there are two types of experiments, but there are really three. One type is an actual, real-world experiment. You observe, make some measurements, and see if it matches one hypothesis more than another. According to legend, Galileo dropped objects of different weights off of the Leaning Tower of Pisa and noted that they fell at the same rate. He also pointed telescopes at the night sky and recorded celestial bodies. Those measurements led him to the heliocentric model of the solar system.
MORE FROMFORBES ADVISOR
The other type is thought experiments. The most famous of these was Albert Einstein’s question of what happens if you hitchhike on a beam of light and then turn on a flashlight. Will the light from the flashlight go at twice the speed of light? The answer led to the theory of special relativity.
But there’s a third type of experiment, which is neither a real experiment nor a thought experiment. It’s a fake experiment. One of these fake experiments has been very influential. It’s about to influence you. You too can experience being turned.
You may have heard of this experiment. Monkeys are placed in a room with a banana hanging from the ceiling and a ladder underneath it. Monkeys love bananas. One monkey tries to climb the ladder to get the banana. At this point, the nefarious researchers spray water into the room. Monkeys hate being sprayed with water. But these researchers don’t spray the monkey climbing for the banana, oh no. That would be too simple. Instead, they spray all the other monkeys. Perhaps this happens a few times. Those soaked monkeys rebalance the weights in their neurons in a flash of backward propagation and learn the devastating pattern. Then, they beat up the banana-retrieving monkey.
After a few rounds of this collective negative reinforcement, the monkey society starts to police themselves. Anyone even sniffs near the banana, they get stern looks. If they start to climb, they get pounded. Soon, there’s no more need for the researchers to spray anyone. Bananas are safe in the ceiling.
Now, this community encounters some changes. One of the monkeys is replaced with a new monkey, one who has never seen water being sprayed due to banana hunting. This monkey looks around, assumes he’s the only smart monkey in the room, and tries to go for the banana. He is quickly and violently introduced to the local culture. Defeated, the new monkey does not try too many times. As other new monkeys replace existing monkeys, they soon start to join in on the roving bands of sentries preventing banana capture.
Eventually, after several generations, the entire room is full of new monkeys, none of whom have ever seen anyone sprayed with water, nor have they seen anyone who has ever seen anyone sprayed with water. Yet the cycle continues.
That’s what may happen when machines do all our trading. The earliest machines will encode the same human foibles that are present in the markets today. We don’t, as a species, have to have foibles. We could have all been rational. But that’s just not the world we live in. In order to fit in, the machines will need to participate and survive in the markets as they are, as we have shaped them to be. Eventually, even when all people are replaced by machines, and even those machines by untold generations of other, newer machines, hundreds of years in the future, they may all continue trading the same way as humans once did, because once upon a time, banana-grabbing was punished.
The only problem is that this monkey experiment never happened. It’s a famous fable, perhaps as legendary as Galileo hanging over the balcony of the Leaning Tower of Pisa. But fake experiments can contain truth too. In 1971, astronaut Dave Scott, on the last Apollo 15 moon walk, replicated the Galilean legend and dropped a feather and a hammer together. Because there was no air resistance, both hit the ground at the exact same time.
Maybe replicating the monkey/ladder/banana experiment would give different results, even if it weren’t considered too cruel to actually run. Who knows. But one thing is fairly certain. Within a few weeks, you will forget that this experiment was never run. The images are just too vivid. You will think it was real. And you’ll mention it to someone, and they’ll pass it on. Years from now, people will still remember this experiment that never was, just like machines will remember the biases the ancient humans never needed to have.
