Mathematical Traders Fail Again
Back in August, I reported on how traders using computer programs based on advanced mathematics and statistical patterns had lost huge sums of money as August experienced dramatic market fluctuations that contradicted the mathematical models, as there was for example market movements that were 20 standard deviations from the perceived normal-events that really never should occur.
But in reality, events considered impossible in mathematical models occur quite frequently. Examples of this include the stock market crash of October 1987, the market movements that destroyed the Long Term Capital Management fund and the movements in August 2007. And while perhaps not considered impossible, we have also seen market movements that contradict what statistical patterns indicate. For example, normally September is the month of the year when the stock market is weakest while January is the month when it is strongest. But this September, stock prices rose sharply while they have fallen sharply this January.
Here is an interesting Bloomberg News article about how last year, the great losers were the "quants", i.e. the traders relying on mathematical models.
This again illustrates how human behavior cannot be captured in mathematical models. Instead, what matters is understanding of sound economic theory of how the economy and the markets function and the understaning of how to apply this to market movements.
But in reality, events considered impossible in mathematical models occur quite frequently. Examples of this include the stock market crash of October 1987, the market movements that destroyed the Long Term Capital Management fund and the movements in August 2007. And while perhaps not considered impossible, we have also seen market movements that contradict what statistical patterns indicate. For example, normally September is the month of the year when the stock market is weakest while January is the month when it is strongest. But this September, stock prices rose sharply while they have fallen sharply this January.
Here is an interesting Bloomberg News article about how last year, the great losers were the "quants", i.e. the traders relying on mathematical models.
This again illustrates how human behavior cannot be captured in mathematical models. Instead, what matters is understanding of sound economic theory of how the economy and the markets function and the understaning of how to apply this to market movements.
posted by stefankarlsson at 9:47 AM
2 Comments:
Trying predict tomorrows market events based on mathematical models and statistics is akin to trying to predict tomorrows news based on yesterdays headlines, folly.
It doesn't matter how well the models model certain events, they are still only able to model a small subset of an infinite amount of possible events, some that may never have occured before.
I was almost laughing my ass off last fall, when I was reading that some quant funds had lost massive amounts of money because "events they had predicted would only occur once every 10000 years happened three days in a row".
A good book on the subject of unpredictability in the world is "Black Swan", I believe it was written by.. a former quant trader, of all things..
I would revise your comments to include not just sound economic theory but also an understanding of how markets move. I'm a technical trader, and as such, I incorporate little of my understanding of economics into my analysis of markets and none at all into my selection of individual trades.
This is not to say that such theory doesn't have a place in trading the markets. Simply, it doesn't have a place for me; it's now how I personally trade. What's important, however, is to have a theory that guides how you trade the markets. And I completely agree that computers cannot capture the range of human emotions that encompass that markets.
That is not to say that quantitative trading is somehow bad or shouldn't be used. There are many successful system traders who only wish to use objective systems. But these traders are fully aware that such systems will incur periodic drawdowns due to changes in market structure, the laws of probability or unlikely events. However the consistently successful traders understand this and size their positions accordingly.
The biggest problems with many of the quant funds were their use of excessive leverage, their ludicrous supposition that the market structure had somehow changed forever (i.e "this time it's different') and their ignorance of the frequency of black swan events. In other words: hubris. And just like Icarus, they flew too close to the Sun, their wax wings melted and the their cash, quite literally, fell from the sky.
Post a Comment
<< Home