A new study from Leibniz University Hannover finds that the 1/3 of German professional fund managers who believe human biases to be important in financial markets have better calibrated estimates and rely more on momentum and contrarian strategies, but otherwise think and act like other fund managers:
I'm a little confused by this study. Lets go down the list of questions in Table 5. Assessment of one's own behavior:
[6] Hindsight Bias "The majority of economic news is not surprising to me". Is this testing hindsight bias at all? It seems to me that this is testing the difference in how endorsers and non-endorsers define surprise. This question leaves me wondering if respondents' answers imply they believe they were able to predict events. I think a better question would have been "Were you expecting (Big Recent Economic Event X) to happen?"
[7] Better Than Average "How do you evaluate your own performance relative to other asset managers." The author of this question is ignoring survivorship bias. I would imagine that every fund manager working today thinks they are better than average based simply on the fact that they are simply in business. When fund managers look around, they see the competition failing all the time. The average life expectancy of asset management groups is notoriously short. Give them some credit.
[8] Miscalibration "DAX 90% confidence interval for next month" This is the only question that directly measures an asset managers ability to benefit from lower biases. This is also the only question that proved statistically relevant. It indicates that endorsers might have a better understanding of volatility, risk, or their own lack of ability to predict near-random variables.
[9] Disposition Effect "I prefer to take profits when I am confronted with unexpected liquidity demands." Compare this question to note 7 "If the investor is faced with a liquidity demand, and has no new information about either stock, she is more likely to sell the stock that is up." I think a question phrased off note 7 would have been more effective.
[10] Home Bias "Allocate a portfolio" This seems to target the home bias well, but I probably still would have explicitly stated that the fund managers will not be able to reallocate funds for 20 years so that they know their familiarity with the German market won't help the portion allocated to Germany.
I would also have liked to see what kind of performance these funds have exhibited over the last 5-10 years. I think historic performance would be a much better indication of how much an understanding of biases is effecting a fund managers ability to make decisions.
Overconfidence bias is one of the most common and universal. One manifestation is that people choose too small a range when expressing uncertainty about some value. They undervalue the probability of values outside that range. Applied to markets, one might expect that this bias would lead traders to undervalue the probability of extreme market moves which take prices outside of a given trading range.
Yet from what I understand, this does not happen. Indeed, if it did, it would lead to an extremely easy money-making strategy; simply buy options and other instruments which will profit on extreme moves. If market prices undervalued these instruments then it would be an easy and obviously profitable investment. But markets move to nullify easy and obviously profitable investments, because investors will buy them and drive up the prices until they are no longer automatically profitable. This is one example of a way in which the cognitive biases of individual traders get nullified by market mechanisms.
The interesting thing is that this can seemingly happen in two ways in terms of individual overconfidence. One way is that some traders may happen to differ in opinion from the market as a whole, and overconfidently think the price will move substantially higher or lower than current prices. They will buy up these out of the money instruments until the profit opportunity disappears. This behavior is consistent with individual overconfidence bias.
The other possibility is for people who personally are inclined to overconfidence to be forced to confront their bias in the face of (let us suppose) a historical record of the market undervaluing extreme moves. They take positions which contradict their intuitive but biased beliefs, driven by the plain fact of an exploitable profit opportunity. With money on the line, greed and fear trump bias, and perhaps they are even led to change their beliefs and reduce their bias, accepting that prices may indeed move much more than seems intuitively plausible.
I wonder which factor dominates. The first doesn't seem strong enough to explain rational market pricing, while the second would predict that traders, especially successful ones, would be less biased about factors relating to their investments than non-traders. I would say anecdotally that most traders I have spoken with seem to me to be quite overconfident and biased about their investments, but perhaps that is just posturing and their internal beliefs are more moderate.
This suggests that we focus not on trying to correct our personal biases but instead on identifying and promoting institutions, such as prediction markets, to reduce biases.
The lesson to learn from the study is that knowledge of our own biases is not enough to overcome them; we must take the further step of using that knowledge to adopt a bias-avoiding strategy. It is, however, an open question whether this strategy should involve institutional mechanisms. An alternative is to adopt anti-bias heuristics, such as Bostrom and Ord's.
Regarding the opinion; "what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time?"
I think its a matter of attitude one would have in such a situation
Take for example if the attitude is similar to one has to a watch or tape measure that one is certain is accurate and unbiased - one would not feel any irritation because one is aware of the consequences of going against the watch or tape measure
I think we might need to differentiate between biases in thought and speech and actual biases in action. People can act biased all day long when they don't suffer any consequences from doing so, and I'd wager this behavior isn't even irrational. There are social advantages to acting overconfident in situations where the price to do so is slight or non-existence. Of course this behavior can be a bad thing in some situations, which is why I generally carry a lot of cash on hand and often make bets with loud-mouths (unfortunately, they almost always back down, except I think in situations where they stand to loose too much face by doing so, which is rare).
Just because the part of our conscious thoughts seem biased doesn't mean our actions are; I'd wager people's actions are often much more rational. This is why I'm skeptical of this study which looks at people who claim to be aware of biases. I don't think their conscious understanding of biases is all that important to how they act. In my opinion, the ability of human action to diverge from human thought and speech is rather remarkable.
I find I can get rid of conscious biases by imagining I am betting on the truth of my thoughts and words. That seems to shift the way my brain works from the 'socialite' setting to 'engineer'. I don't think this technique will get you any chicks, though...
To look at Mike's point another way, the vast majority of us would become more calibrated if we reduced the certainty of almost all of our judgements (perhaps excepting a few of the mathematical ones) by at least 1%, and probably by a lot more. Each particular person should assume that he is a member of that vast majority unless he has specific reasons to think that he is not (and they'd better be really good reasons.) So basically each of us should immediately reduce the certainty of all of our judgements by at least 1%, and probably by a lot more.
Nonetheless, we do not do this. Why not?
In essence, this is simply an illustrate of my point when I said that I have no intention of doing things that are immoral, but I expect to continue doing them anyway. In the same way, I have no intention of continuing to be overconfident, but nonetheless I fully expect to continue to be overconfident.
"what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time?"
I think these type thought experiments aren't particularly useful. It seems from your example that your robotic pal could come up a better solution to the "immense irritation" than you could without his help.
Robin writes: "This suggests that we focus not on trying to correct our personal biases but instead on identifying and promoting institutions, such as prediction markets, to reduce biases."
there seems to be this tension--i want to remove biases from myself generally, believe what is true, but then i know it's tough or impossible to really be very effective at overcoming my personal biases, if i read this post right. so there's an institutional focus. but does this mean i act with some indifference to bias on a personal level? 'i can't overcome it, so i'll just believe i'm right and above average and carry on with life.'
it seems this is how i'm going to act, but at the same time, i have learned that it's likely i am more biased than i think, which is an odd statement in itself. i know this but i forget it when i am thinking in a biased manner, i suppose. am i eternally bound to just get annoyed when i notice after the fact that i've been biased, or when i've been forced by the world to admit i was wrong? is the best i can do to act in a biased manner with some indifference, but separate from this thought process i also set up institutions that will keep this biased self from acting in too biased a manner? i'm like my own babysitter. i have two selves, both of whom want to ignore the other, the one who knows he's biased generally, and the one who doesn't believe he's biased in a particular case. do i tie myself to a decision-making process that forces me to make decisions that feel wrong, but are more likely right?
a thought experiment: what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time? defering to him would be more rational. there's sort of a tension between wanting to make your own decisions, and wanting to be rational. is there a way of resolving the tension?
I wonder, is it possible to abstract the information provided by markets in a way that does not run afoul of cognitive biases when analyzed? We do this sort of thing to get around cognitive limitations (whether biases or not) all that time. I wonder if something similar could be done with markets. I'd imagine it probably is already being done on many levels, whether consciously or not, so I suppose the question is how could it be improved?
tc, I think Robin was referring to prediction markets reducing biases in other areas, not necessarily in trading. I think they are a great example of information abstracted to a form which does not run afoul of biases.
Ed Ziemba wrote about a real and persistent cognitive bias which allowed him to make money off of place and show bets.
The basis insight is this: At times, a large group of people will not put place or show money on favourite. Individually, they reason that such a bet will return an unattractive amount compared to a winning bet. This sentence is true, until a number of people act upon it. The fewer number of people in the place or show pool, the greater their return independently of risk.
Joel Greenspan has also identified similar bets.
Of course, Soros was famous for making money an essentially the following illusion: if I can sell you an investment with a 10% return based on the quality of earnings, viz. that they are increasing, I can for a while sustain a self-enforcing illusion.
None of the strategies are for people who get a thrill out of gambling or who look to volatility as a fun experience.
Most of the behavioral tricks identified in the literature don't have any clear strategies; many of the clear strategic advantages don't seem to be identified in the behavioral literature.
I'm a little confused by this study. Lets go down the list of questions in Table 5. Assessment of one's own behavior:
[6] Hindsight Bias "The majority of economic news is not surprising to me". Is this testing hindsight bias at all? It seems to me that this is testing the difference in how endorsers and non-endorsers define surprise. This question leaves me wondering if respondents' answers imply they believe they were able to predict events. I think a better question would have been "Were you expecting (Big Recent Economic Event X) to happen?"
[7] Better Than Average "How do you evaluate your own performance relative to other asset managers." The author of this question is ignoring survivorship bias. I would imagine that every fund manager working today thinks they are better than average based simply on the fact that they are simply in business. When fund managers look around, they see the competition failing all the time. The average life expectancy of asset management groups is notoriously short. Give them some credit.
[8] Miscalibration "DAX 90% confidence interval for next month" This is the only question that directly measures an asset managers ability to benefit from lower biases. This is also the only question that proved statistically relevant. It indicates that endorsers might have a better understanding of volatility, risk, or their own lack of ability to predict near-random variables.
[9] Disposition Effect "I prefer to take profits when I am confronted with unexpected liquidity demands." Compare this question to note 7 "If the investor is faced with a liquidity demand, and has no new information about either stock, she is more likely to sell the stock that is up." I think a question phrased off note 7 would have been more effective.
[10] Home Bias "Allocate a portfolio" This seems to target the home bias well, but I probably still would have explicitly stated that the fund managers will not be able to reallocate funds for 20 years so that they know their familiarity with the German market won't help the portion allocated to Germany.
I would also have liked to see what kind of performance these funds have exhibited over the last 5-10 years. I think historic performance would be a much better indication of how much an understanding of biases is effecting a fund managers ability to make decisions.
Overconfidence bias is one of the most common and universal. One manifestation is that people choose too small a range when expressing uncertainty about some value. They undervalue the probability of values outside that range. Applied to markets, one might expect that this bias would lead traders to undervalue the probability of extreme market moves which take prices outside of a given trading range.
Yet from what I understand, this does not happen. Indeed, if it did, it would lead to an extremely easy money-making strategy; simply buy options and other instruments which will profit on extreme moves. If market prices undervalued these instruments then it would be an easy and obviously profitable investment. But markets move to nullify easy and obviously profitable investments, because investors will buy them and drive up the prices until they are no longer automatically profitable. This is one example of a way in which the cognitive biases of individual traders get nullified by market mechanisms.
The interesting thing is that this can seemingly happen in two ways in terms of individual overconfidence. One way is that some traders may happen to differ in opinion from the market as a whole, and overconfidently think the price will move substantially higher or lower than current prices. They will buy up these out of the money instruments until the profit opportunity disappears. This behavior is consistent with individual overconfidence bias.
The other possibility is for people who personally are inclined to overconfidence to be forced to confront their bias in the face of (let us suppose) a historical record of the market undervaluing extreme moves. They take positions which contradict their intuitive but biased beliefs, driven by the plain fact of an exploitable profit opportunity. With money on the line, greed and fear trump bias, and perhaps they are even led to change their beliefs and reduce their bias, accepting that prices may indeed move much more than seems intuitively plausible.
I wonder which factor dominates. The first doesn't seem strong enough to explain rational market pricing, while the second would predict that traders, especially successful ones, would be less biased about factors relating to their investments than non-traders. I would say anecdotally that most traders I have spoken with seem to me to be quite overconfident and biased about their investments, but perhaps that is just posturing and their internal beliefs are more moderate.
This suggests that we focus not on trying to correct our personal biases but instead on identifying and promoting institutions, such as prediction markets, to reduce biases.
The lesson to learn from the study is that knowledge of our own biases is not enough to overcome them; we must take the further step of using that knowledge to adopt a bias-avoiding strategy. It is, however, an open question whether this strategy should involve institutional mechanisms. An alternative is to adopt anti-bias heuristics, such as Bostrom and Ord's.
Regarding the opinion; "what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time?"
I think its a matter of attitude one would have in such a situation
Take for example if the attitude is similar to one has to a watch or tape measure that one is certain is accurate and unbiased - one would not feel any irritation because one is aware of the consequences of going against the watch or tape measure
I think we might need to differentiate between biases in thought and speech and actual biases in action. People can act biased all day long when they don't suffer any consequences from doing so, and I'd wager this behavior isn't even irrational. There are social advantages to acting overconfident in situations where the price to do so is slight or non-existence. Of course this behavior can be a bad thing in some situations, which is why I generally carry a lot of cash on hand and often make bets with loud-mouths (unfortunately, they almost always back down, except I think in situations where they stand to loose too much face by doing so, which is rare).
Just because the part of our conscious thoughts seem biased doesn't mean our actions are; I'd wager people's actions are often much more rational. This is why I'm skeptical of this study which looks at people who claim to be aware of biases. I don't think their conscious understanding of biases is all that important to how they act. In my opinion, the ability of human action to diverge from human thought and speech is rather remarkable.
I find I can get rid of conscious biases by imagining I am betting on the truth of my thoughts and words. That seems to shift the way my brain works from the 'socialite' setting to 'engineer'. I don't think this technique will get you any chicks, though...
To look at Mike's point another way, the vast majority of us would become more calibrated if we reduced the certainty of almost all of our judgements (perhaps excepting a few of the mathematical ones) by at least 1%, and probably by a lot more. Each particular person should assume that he is a member of that vast majority unless he has specific reasons to think that he is not (and they'd better be really good reasons.) So basically each of us should immediately reduce the certainty of all of our judgements by at least 1%, and probably by a lot more.
Nonetheless, we do not do this. Why not?
In essence, this is simply an illustrate of my point when I said that I have no intention of doing things that are immoral, but I expect to continue doing them anyway. In the same way, I have no intention of continuing to be overconfident, but nonetheless I fully expect to continue to be overconfident.
"what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time?"
I think these type thought experiments aren't particularly useful. It seems from your example that your robotic pal could come up a better solution to the "immense irritation" than you could without his help.
Robin writes: "This suggests that we focus not on trying to correct our personal biases but instead on identifying and promoting institutions, such as prediction markets, to reduce biases."
there seems to be this tension--i want to remove biases from myself generally, believe what is true, but then i know it's tough or impossible to really be very effective at overcoming my personal biases, if i read this post right. so there's an institutional focus. but does this mean i act with some indifference to bias on a personal level? 'i can't overcome it, so i'll just believe i'm right and above average and carry on with life.'
it seems this is how i'm going to act, but at the same time, i have learned that it's likely i am more biased than i think, which is an odd statement in itself. i know this but i forget it when i am thinking in a biased manner, i suppose. am i eternally bound to just get annoyed when i notice after the fact that i've been biased, or when i've been forced by the world to admit i was wrong? is the best i can do to act in a biased manner with some indifference, but separate from this thought process i also set up institutions that will keep this biased self from acting in too biased a manner? i'm like my own babysitter. i have two selves, both of whom want to ignore the other, the one who knows he's biased generally, and the one who doesn't believe he's biased in a particular case. do i tie myself to a decision-making process that forces me to make decisions that feel wrong, but are more likely right?
a thought experiment: what if i had a robotic pal who was unbiased and knew what was better for me than i did. wouldn't it still be immensely irritating to have to defer to him all the time? defering to him would be more rational. there's sort of a tension between wanting to make your own decisions, and wanting to be rational. is there a way of resolving the tension?
I wonder, is it possible to abstract the information provided by markets in a way that does not run afoul of cognitive biases when analyzed? We do this sort of thing to get around cognitive limitations (whether biases or not) all that time. I wonder if something similar could be done with markets. I'd imagine it probably is already being done on many levels, whether consciously or not, so I suppose the question is how could it be improved?
tc, I think Robin was referring to prediction markets reducing biases in other areas, not necessarily in trading. I think they are a great example of information abstracted to a form which does not run afoul of biases.
"promoting institutions, such as prediction markets, to reduce biases."
Aren't these managers already working with the most liquid and complete prediction markets in the world?
Ed Ziemba wrote about a real and persistent cognitive bias which allowed him to make money off of place and show bets.
The basis insight is this: At times, a large group of people will not put place or show money on favourite. Individually, they reason that such a bet will return an unattractive amount compared to a winning bet. This sentence is true, until a number of people act upon it. The fewer number of people in the place or show pool, the greater their return independently of risk.
Joel Greenspan has also identified similar bets.
Of course, Soros was famous for making money an essentially the following illusion: if I can sell you an investment with a 10% return based on the quality of earnings, viz. that they are increasing, I can for a while sustain a self-enforcing illusion.
None of the strategies are for people who get a thrill out of gambling or who look to volatility as a fun experience.
Most of the behavioral tricks identified in the literature don't have any clear strategies; many of the clear strategic advantages don't seem to be identified in the behavioral literature.