Thursday, 16 October 2008

Black Swans: New Rule to Control the Risk of Fat Tails

If there ever was a Black Swan type of market, this is it. Statisticians call it a "fat tail." It's the kind of horror story that catches a lot of traders by surprise - markets doing crazy things that have little historical precedent and blowing all our data out of the water. Trader Nassim Nicholas Taleb wrote an excellent book about it called The Black Swan: The Impact of the Highly Improbable. He argued that markets don't work according to conventional rules of statistics and that surprising events - like we're seeing in today's markets - happen much more often than statistics would predict.

This brings me to a new risk-control rule I've been contemplating for a few weeks. It's my twist on a common trading rule: when your portfolio drops over six percent in a month, stop trading for the rest of the month. Until recently, I understood this rule to be mostly a psychological rule for a trader who is feeling messed up because of a string of losses that screw with your head and cause you to make more bad trades. I didn't see how it applies to mechanical trading.

In fact, I now see its real utility for me can be to control Black Swan risk. Here is my twist on this new rule I will be using. I call it the Black Swan Rule. It has two parts. One applies to the whole portfolio, the other to individual trading setups. (1) In any four-week period, if my portfolio experiences a loss greater than six percent (calculated using the weekly close value), I will close all losing trades on the following weekly open and close winning trades if they experience a drawdown of greater than two percent. I will also cease trading new signals for four weeks. (2) As well, I will cease trading any setup for four weeks that was stopped out because of a loss greater than the setup's past maximum drawdown. The idea is to give markets time to settle down and allow the data to come back into usual alignment.

The question is how to apply this rule now given the recent market craziness. I determined that my rule would have called for me to cease trading Sept. 22 for four weeks, with trading to resume next Monday. That would have helped me avoid much of the recent hurricane sweeping the markets. But since I didn't have this rule at that time, I am obviously still trading. In the latest four-week period, my portfolio saw a greater-than-six-percent decline as well. What to do? Here's my solution: I will check the situation after this week's close. If the last four-week period again shows a decline over six percent, I'll execute the Black Swan Rule for four weeks. For more details on all my risk-control rules, see my "How It Works" page. Good luck the rest of this week.


Anonymous said...

Hi Alex,

First, just wanted to thank you for your website. I find it invaluable. I look at my other trading resources and use your COT presentation for further insights. Your service adds a valuable piece to a complicated puzzle providing a better market view.

We are in a time of market anomalies. Hedge funds, a big part of the market, are going out of business as investors pull their money out or go bankrupt through bad investments.

I do not expect volatility to subside for quite some time - probably several months. Thought I think the market will likely fall another couple K in the next week and have a sustained rally thereafter, it will be a minefield of spikes. It will continue to create stop problems. But, this market is fraught with profit opportunities. Volalitlity can create big profits, or losses.

That said, your website provides tremendous market insights and I look forward to reading it each week. I feel bad for your recent losses, but we all take them every now and then. It almost seems the market is so complicated that any singular investment discipline will have trouble working, most especially with all of the recent black swan events.

Your website has saved me money in the past by stopping investments I would have otherwise made. In addition to your website, I have had good success lately using a combination of technical advising services.

May I humbly suggest you may find it helpful to augment your investment discipline with another. I only offer the following website with the idea of helping you make profitable investment decisions, i.e., to augment your investment discisions. I would like to see you remain profitable and stay online. We investors need your services. One service I have found invaluable is Robert McHugh Ph.D. at:

He offers a free one month subsciption, which requires no credit card. He really is good and is uncanny in his predictions. I have no affiliation with him other than being a subscriber. Once, again, I only recommend this to supplement your work, to help you augment your investment analysis.

Please keep up the good work and I wish you the best.

Best regards,
Mike P.

craigjones said...

Just a note, this market turmoil is not a "Black Swan" event as described in Taleb's book. He specifically states near the end of the book that stock market crashes and financial panics do not fall under this category, as they are, in fact, probable events.

Alex Roslin said...

Hi Craigjones,

Taleb called the current turmoil a Black Swan on his blog:


Anonymous said...


We call all these new tweaks rules: survivorship bias.
Any Mechanical system will work well in an up trend market, but not in down or sideways markets. The problem with your system, is that it relies on investors expectations whether they are small or large. That is not realiable. In a bull market, almost any investor groups expectations will come true over time as long as the trend is up. Another problem with your system is that the historic data only goes back to 1995, which have been pretty much a bulish cycle since then. Personaly, I have come the conclusion that Technical analysis is probably the only reliable way to invest. Trend is your friend. Don't try to fight it just try to get on board with it and have some good stop loss management to keep your profits.

Alex Roslin said...


Thanks for your input. The new rule is a very common one among traders. See, for example, Way of the Turtle. I just didn't see the utility of it until now. I think it's vital to be open to learning in trading, as with anything in life. Seems silly to me not to change my system as I learn more.

Historic data back to 1995 includes the major bear market of 2000-2003 and smaller corrections in the 1990s. The standard is to have at least one bear, bull and trading range within the dataset.

The system doesn't rely on investor expectations. It relies on the trader positioning in the COT data. This isn't the same thing.

I actually consider the system to be based on technical analysis - specifically, a type of Bollinger Band system. Except, the difference between it and many approaches out there is I'm trying to base mine on backtesting. Most technicians rely on mythology about the patterns they follow, without any reference to whether it's actually a robust pattern. As David Aronson has shown, many technical patterns don't actually work.