Asset allocators use the so-called Modern Portfolio Theory in attempt to measure risk. But linear equations like standard deviation try to simplify the data to the point it grossly fails to capture the true potential for loss. Equations like standard deviation, in fact, aren't very useful in understanding market data. A good way to understand the problem is the flaw of averages: a 7 foot tall guy can drown in a pond that averages only 3 feet deep. If you don't consider the maximum drawdown (historical decline), in this case the maximum depth, you will greatly underestimate the risk.

Source: http://www.flawofaverages.com/

Market trends move far higher and lower from its average to fit nicely into a bell shaped curve. The bell curve for market data is asymmetric, not a picture of symmetry. It’s people who use such an equation that have outlier losses that are much larger than they expected.

The conventional wisdom of stock picking is the “value investing” methods popularized by Ben Graham in his first edition of Security Analysis in 1934. Ben Graham actually argued against measures of risk based upon past prices (such as volatility), saying that price declines can be temporary and not reflective of a company’s true value. He argued that risk comes from paying too high a price for a security, relative to its value and that investors should maintain a “margin of safety” by buying securities for less than the true worth.

I completely disagree with this part of the Graham “margin of safety”. His margin of safety assumes that you can actually buy securities for less than they are worth. In other words, you need to know that in advance and others don’t. It assumes the outcome is determined by the point of entry. I believe just the opposite. I believe the outcome is controlled by the exit – we don’t know the outcome in advance, so the results are determined by what point you sell. Graham’s “margin of safety” assumes you are right to begin with. so your risk management is achieved from paying a price so much lower than true value that you have room (a margin for error). That may have been more possible prior to 1934. Anyone who believes they are so right at the point of entry because they are paying such a low price, then, certainly won’t exit the position if it goes a lot lower. If you have no predefined exit on the way down to protect against large losses, I guess your exit is $0, and your risk is "all of it".

That’s what’s happened in nearly every major blowup in history. If you walk into a pond that is an average of 3 feet deep and you start walking faster as it gets deeper and deeper, you are increasing the chance you may drown. Starting out deeper in the water isn't going to change that.

Long Term Capital Management, one of the biggest blowups in history, assumed the markets they traded wouldn’t trend agsinst them any further, so they kept betting against the trend buying lower and lower until they couldn’t anymore. With a few Nobel Prize winning PhD’s on staff, they certainly believed they were right and got caught tight in the loss trap. They lost billions. 

In 2008 the Wall Street Journal printed The Stock Picker's Defeat about Bill Miller. Bill Miller’s Legg Mason Value fund has been criticized much since it’s large losses during the 2008-2009 period. Miller buys stocks he considers to be undervalued. That very worked well for about 15 years, but they learned how risky buying declining stocks can be. He was simply following his objective. It said nothing about any aim to manageme risk or capital preservation. If you keep buying what is falling, betting against the trend, you have to realize the fact that you are probably risking it all. A value investor believes a lower price is simply a higher “margin of safety” and that probably works until it doesn’t. When it doesn’t, you may be whipped out.

A more recent fund example is Bruce Berkowitz and his Fairholme Fund. A few weeks ago CNN Money wrote "Fairholme Fund loses half its value and its co-manager". It's another "value" fund that buys declining stocks hoping they go back up. They may turn out to be right, but in the meantime their losses are apparently too large for their investors. According to the article, they manage billions but billions have been withdrawn from the fund in addition to the capital loss.

Conventional wisdom seems to believe that "buying low, selling high" is some form of risk management. So, when some people hear something that sounds just the opposite, like my recent interview in Investor's Business Daily, they perceive it as more risky. Yet, it's just the opposite. Buying things that are rising and selling when they decline is a tactic of managing exposure to loss that is absolute. Defining risk as the difference between the price today and the price we'll exit if to goes down is an absolute form of risk management. 

Another recent example is John Paulson, a hedge fund manager who become famous for profiting during the 2008 period crisis by betting that the strong housing market trend would reverse. It did... that time. Reuters recently wrote "Big losses for John Paulson". He apparently bet too big against the trend this time and took on huge losses. In this case, reportedly even he is losing sleep over the losses, according to "Losses mean sleepless nights for John Paulson". You can probably see how the more you believe you are right, the more that loss trap is likely to keep you awake at night. When your ego is tied to your opinions, it's hard to let it go and cut loose from the trap. 

I was reminded of these things today when I read this in IBD about MF Global. 

The brokerage bought up bonds of Italy, Portugal, Spain and Ireland last year, betting they would have bounced back by now. But the euro-zone debt crisis continued to fester, and MF Global started running out of cash.

The bold highlight is mine. You can probably see how conventional wisdom is the very tactics that lead to such blow-ups. Efficient Market Theorist point to such managers as examples of the problem with "active management", but it's really only an example of the problem of the most common approaches - the kind that get risk wrong.