For years, buying more after a fall made Carlos look brave. Carlos is one of the constructed traders in Nassim Nicholas Taleb's Fooled by Randomness. He buys emerging-market bonds, then adds to his holdings when prices drop. Earlier declines recover. Each recovery rewards his confidence and makes the next purchase easier to defend.
Then, in nineteen ninety-eight, his Russian bonds keep falling. Carlos keeps buying. His bank had earned about eighty million dollars from his earlier trading; that summer he loses three hundred million. A long winning record had concealed the size of a possible failure.
Taleb's answer is to judge a bet by what each outcome costs or pays, and whether the person taking it can survive. Losing frequently can make sense when losses remain affordable and occasional gains justify their cost. Winning frequently can still destroy a career.
Buying more during a decline lowers the average purchase price and increases the money exposed. This is averaging down. Carlos treats cheaper bonds as a better opportunity without adequately questioning his idea of their value. Even an eventual recovery could arrive too late to save his career.
John, another constructed trader, increases exposure through borrowing. His own fourteen million dollars supports a fifty-million-dollar portfolio. Losses on the larger holding fall on the smaller sum he actually owns. His capital disappears during the crisis. Borrowing and buying more during a fall are different ways to make an error more expensive.
What the losing bet costs
A call option shows why a discouraging record can still be reasonable. In Taleb's example, it gives its buyer the right to purchase a share at an agreed price one month later. If the share then costs less than that price, exercising the right offers no advantage, and the option expires worthless.
In Taleb's simplified example, the option costs one dollar. It has a ninety percent chance of expiring worthless and a ten percent chance of paying an average of ten dollars. Expected payoff weights each amount by its probability. One chance in ten of an average ten-dollar payment contributes one dollar to that expectation. The other outcomes contribute nothing.
That is one dollar before paying for the option. Subtract the one-dollar price and expected profit is zero, before any additional costs. The example establishes a fair price, despite frequent losses. It does not establish a profitable strategy. For that, the expected payments would have to exceed what the buyer pays.
The median gross payoff, the middle result when outcomes are sorted, is zero. The mean gross payoff is one dollar. Those numbers describe different features of the same gamble. Neither promises what an individual buyer will receive, or when a gain will arrive.
Taleb believes some rare-event exposures are underpriced because people dislike repeated small losses. That judgment about price matters. A rare payoff is no bargain merely because it is rare. The buyer also needs enough resources to bear the losing stretch. Running out of money ends participation before any later gain can help.
The seller receives frequent premium income but faces occasional much larger payments. A calm record may omit the loss that determines the value of the whole activity. Some distributions give extreme outcomes much more weight than familiar bell-curve intuition suggests. A short sample can leave that feature invisible. Small regular gains feel reassuring, while frequent small losses feel unpleasant. Choosing the smoother record can therefore postpone the pain into a catastrophe. Removing extreme losses from an assessment as irrelevant would erase precisely the outcomes that can bankrupt someone.
Deciding before the loss
Carlos needed a response to being wrong before the decline began. Taleb favors deciding how much to expose, what would undermine the original judgment, and when to leave. A falling price should prompt scrutiny of the method that made the asset seem cheap. Otherwise every setback supplies another reason to increase the bet.
A preselected exit, often called a stop loss, aims to limit the damage from a mistaken forecast. It cannot guarantee an exact cap on losses. Historical evidence can suggest a trade without certifying its safety. Crashes elsewhere can widen the possibilities worth considering; the largest loss in a recent record cannot set the limit for the next one.
There is pressure to ignore this restraint. After a year without a fire, an insurance premium can look wasted. The expense is visible; the disaster never happened. Risk managers similarly get blamed for blocking profitable trades, then blamed again when a permitted position collapses. A reassuring report changes little unless someone changes the exposure.
The bargain also depends on who bears the consequences. Taleb criticizes executive arrangements that reward success generously while leaving shareholders with much of the downside. An attractive risk for the person choosing it can be a terrible risk for its owner. He distinguishes entrepreneurs who expose themselves to failure while pursuing their ideas.
Survival remains a continuing requirement. Financial expertise doesn't ensure equal care with physical danger, and present prosperity leaves the future unresolved. A sensible losing strategy must still be affordable and worth its price. There is no prize for correctly predicting a recovery after losing the ability to participate in it.
Three things to keep
Winning frequency means little without the size and cost of possible outcomes.
Frequent losses need affordable exposure and gains whose size and probability justify the cost.
Plan for being wrong before losses make changing course emotionally harder.