The original title of this post was “Quantitative trading and Macroeconomics”, however more than Macroeconomics per se, this post wants to underline the impact that macro dynamics in general have on financial markets (where with “macro dynamics” I want to indicate any dynamic caused by human behaviour, including Macroeconomics but also Behavioural Finance, Psychology, Politics, etc. All in all, they are just different -codependent- aspects of the coin )
This post may seem to divert from the usual thematic discussed in mathtrading and in some sense it probably does. However, changes of market conditions due to some structural change in the global socio-political economy can be as important for quant traders as they are for discretionary ones.
Realising when such a process is taking place and be ready for it is a challenge that requires ahead-of-time planning and keeping in mind that black swans are a possibility in the fat-tailed world of financial markets. What makes the global web of socio-political interactions even more important is that changes in its structure are often the reason behind other dynamics that quant traders can find themselves facing with little or no notice.
One day your strategies may just stop working, cointegrated pairs can start diverging and never look back, long best performers/short worst performers can stop being a Momentum strategy and actually become a Mean Reversion one. And you may have no clue on the reason behind and your model may not be trained to look at the colours of swans.
In some aspects, this is what happened to some quant trading superstars (AQR, PDT) during the 2007-2008 financial meltdown, as told by Scott Patterson in his book “The Quants” (http://www.amazon.com/The-Quants-Whizzes-Conquered-Destroyed/dp/0307453375).
In regards to this, a recent article from George Friedman gave me some food for thought. Here are some relevant extracts (you can read the full article from here: http://www.stratfor.com/weekly/financial-markets-politics-and-new-reality):
“Louis M. Bacon is the head of Moore Capital Management, one of the largest and most influential hedge funds in the world. Last week, he announced that he was returning one quarter of his largest fund, about $2 billion, to his investors. The reason he gave to The New York Times was that he had found it difficult to invest given the impossibility of predicting the European situation. […]
The purpose of hedge funds is to make money, and what Bacon essentially said was that it is impossible to make money when there is heavy political involvement, because political involvement introduces unpredictability in the market. […]
The investors’ problem is that they mistake the period between 1991 and 2008 as the norm and keep waiting for it to return. I saw it as a freakish period that could survive only until the next major financial crisis — and there always is one. While the unusual period was under way, political and trade issues subsided under the balm of prosperity. […]
Once the 2008 crisis hit external factors that were always there but quiescent became more overt. The internal workings of the financial system became dependent on external forces. We were in the world of political economy, and the political became like a tidal wave, making the trading cycles and opportunities that traders depended on since 1991 irrelevant. […]
The real problem for the hedge funds was not that they didn’t understand what they were doing, but the manner in which they had traded in the past simply no longer worked. Even understanding and predicting what political leaders will do is of no value if you insist on a trading model built for a world that no longer exists.[…]
A strong argument can be made — corruption and stupidity aside — that the real problem has been a failure of imagination. We have re-entered an era in which political factors will dominate economic decisions. This has been the norm for a very long time, and traders who wait for the old era to return will be disappointed. Politics can be predicted if you understand the constraints under which a politician such as Merkel acts and don’t believe that it is simply random decisions.”
Failure of imagination. Or intellectual laziness.
These two are main challenges for anyone trying to develop a quantitative trading strategy, but they can show their effects only once the strategy has been live for a while.
How is this possible? I think one of the reasons lies in some necessary steps in the development of a strategy: approximation and simplification.
When we translate an idea into an algorithm, there are usually a huge number of variables/situations that could affect our model and prevent it from working. However, the only way to get started is to operate some simplifications and consider the problems one by one. In doing this, it’s natural to strive to find workarounds for empirical and obvious obstacles and to leave for “later” other possible issues.
An example could be a pair which has been cointegrated since the beginning of our historical data and which implicitly you assume to continue behave this way.
Another interesting example comes from Patterson book in the form of a strategy long top performers stocks and short worst performers. What (likely) happened was that many people were running the same strategy at the same time, and when some of them simultaneously exited during 2007 because in the need of funds, in a vicious circle the strategy not only stopped working but actually became a mean reversion strategy.
A third example comes from Friedman, and although more macro oriented, shows a similar pattern: people getting used to one market regime (using the term “regime” in a loose fashion here), and unwilling, unconsciously to some degree, to consider the challenges involved in a change of regime.
What’s in common in the 3 examples is that in each case one could have been prepared by “simply”:
1) Considering the actual facts as being a possible scenario (pairs breaking up – best performers and worst performers showing mean reversion behaviour – markets being extremely sensible to political developments)
2) Re-evaluate the likelihood of these happening by actively following global political and economical developments and have a good intuition on their consequences.
The good news is that our quantitative methods can provide some help for both aspects.
The bad one is that our models will always be…just models, and hence imperfect.
But having them tuned and prepared for the otherwise unexpected is already a big step
(and, intellectually, can also be an easy step, e.g. use stops losses, diversify your strategy across a number of assets, diversify your trading across different strategies).
Finally, an even bigger step is to have a good understanding of what’s happening in the world and of what could happen.
PS: For interested readers, here are similar calls for a better understanding of macro matters (including links to some material in Quantivity’s post):