There is an extremely famous psychology paper written by Daniel Kahneman and Amos Tversky named ‘Judgement Under Uncertainty: Heuristics and Biases’ (Kahneman won a Nobel Peace Prize in 1992 for his work in the field, specifically on prospect theory) which explores the decision making process
As trading requires decisions to be made constantly – stop loss adding, lot size, whether a trade is right to take etc – I think a quick write up would be highly applicable.
Essentially, there are several ‘heuristics’ or ‘biases’ which I will attempt to put into a trading context.
1) Reliability. Making sense of data on the spot is a difficult task to undertake. When you look at a chart, you are looking at a representation of the market in candlesticks and not the actual market. Adding more and more indicators causes the reliability of this data to further decrease, possibly leading to a distorted view (however, if you are profitable with indicators then that is all that matters). Decreasing the reliability is pretty self explanatory. Anyone familiar with this equation will know why, mathematically:
2) Representation. We normally feel that if a pattern is forming that it will play out in the way we expect based on knowledge of previous events. Let’s say we have a man in a suit. Would you say he is more likely to be a gardener or a lawyer? You’d probably go lawyer based on an understanding that most lawyers wear suits and most gardeners don’t.
When back testing, you may look for data to represent the notion you have about a certain set up and ignore the set ups that have failed, therefore leading to a skewed view of that strategy. Indicators represent a potential set up and not what is actually occurring – indicators are used to fit a concept in your head. The fact that something is more representative does not necessarily make it more likely to occur.
3) Anchoring. People place too big a value on the first number or value given to them. They then compare from that. An easy, easy sales tactic to get someone to buy something is to offer the most expensive first, then offer the cheapest, and then the product you want to push is to be between these two values. Little tip there for retail FX account managers.
In trading, you could consider it like this. Say you see Cable trading at 1.50 in January, but now it is at 1.40. A lot of people will start to say this move is too overextended; ‘it has to rally’. Instantly your bias is skewed based on the number you see at the start of the year (I don’t know if this is a good example, but you get the picture).
4) The Gambler’s Fallacy
. When an event occurs more or less is a short time period, you may believe that it will happen less or more in the future. As said before, the market is impartial. Past events do not change the probability of future events occurring. Think of it this way. You’re at a Roulette wheel only playing black and red (and for example’s sake it’s a European wheel with no 0). You hit 20 blacks in a row. You might assume that the ball has to land on a red the next time. But the probability of it landing on black again is the same as it landing on red (only a 1 in 2 chance of either event occurring).
5) Availability bias. I hate trading gold now. The reason being is that back in December I just had a torrid time with it. This is all I remember when I think of gold now. However, I don’t remember the times I had profited. Again, here is a bias based on frequency/magnitude of past events, that are not indicative of future behaviour (apparently; I don’t buy that quote though).
Biases allow us to make decisions quickly. However in an environment where subjectivity has to be eliminated as much as possible, they can be detrimental.
Automated traders do not have the problem of biases as the emotion is taken out of the trade, which is why possibly developing an algorithm can be hugely beneficial if you have a stringent set of rules that you can programme into a computer. This can be done via coding which I have no clue about, or by using pictorial programming applications such as Qubitia