Developed by Nassim Nicholas Taleb (Antifragile, 2012).
Classifies a system or position as harmed, unaffected, or improved by shocks, and points to exposures with limited downside and open upside.
A factory's year under supply shocks of growing size: does it break, hold, or gain?
Across the plot, a supplier outage lasts longer; the line is the factory's profit change for the year. The black line is the lean factory with one supplier: fine in a calm year, worse with every week of outage, and worse at an accelerating rate, which is fragile. Spare stock, set here at the six weeks the planners chose, holds the line flat until the stock runs out, which is robust. Add cheap options on backup suppliers and the line bends upward, since in an industry-wide outage the factory can serve customers its rivals cannot, which is antifragile. Illustrative.
This factory's year, $MLean, one supplier
An example
A maker of industrial pumps had spent five years making its main plant lean: one supplier for each critical casting, two days of inventory, and the lowest unit costs in its market. The chief executive asked a new question at the annual review: what would a long outage at the casting supplier do to us, and could any version of that outage leave us better off?
The planning team drew the plant's profit against outages of growing length. Lean, it lost more with every week, and faster as customers gave up and went elsewhere: a fragile position, however good the average year looked. Six weeks of spare castings made the line flat for any outage up to six weeks, which was robust and cost about $1.5 million a year. The third move changed the shape. For $300,000 a year each, the company bought standing options with two foundries abroad to take orders at short notice. In an industry-wide outage those options would let it serve customers its rivals could not. The board approved the stock and both options.
Why it works
Most risk work asks how to survive a shock. This tool adds a classification that changes the question: for any position, plot what happens to it as shocks grow. If the losses grow faster than the shock, the position is fragile, and its good average hides the year that breaks it. If the line is flat, it is robust. If it rises, the position gains from disorder.
The practical rule follows from the shape. Remove exposures with a capped upside and an open downside, such as a single supplier that saves a little and can cost everything. Add exposures with a small, known cost and a large possible gain, such as options, spare capacity, and small experiments. Prediction is not required: the classification works without knowing when or how big the shock will be, which is the point when the shocks that matter are the ones nobody forecast.
The thinking behind it
Nassim Nicholas Taleb Nassim Nicholas Taleb, a former options trader who writes on risk and probability, introduced the term in Antifragile: Things That Gain from Disorder (2012). He sorted systems into three classes, fragile, robust, and antifragile, by how they respond to volatility and stress, and argued that some systems need stressors to improve. His practical advice is to reduce exposure to large losses and increase exposure to cheap options on large gains, often through a barbell of very safe and small very risky positions. The book is an essay for a general reader; some statisticians fault his wider treatment of statistics.
How to run it
- Name the position under review. Pick one thing to test: a plant, a supply chain, a portfolio, a product line, or a career.
- Draw its response to shocks of growing size. Estimate the result for no shock, a small one, a medium one, and a severe one, and plot them.
- Read the shape. Losses that grow faster than the shock mean fragile; a flat line means robust; a rising line means antifragile.
- Cut capped-upside, open-downside exposures. Look for savings that are small and certain but carry a large, rare loss, such as single sourcing or zero slack.
- Add small-cost, large-gain exposures. Buy options, keep spare capacity, and run cheap experiments whose worst case is known and whose best case is large.
- Let small stresses do their work. Allow minor failures and drills in the parts that can learn from them, so they get stronger before a large shock arrives.
Where it goes wrong
The common failure is calling a position robust because it has survived. A lean plant that has never seen a long outage looks strong until the first one. It shows up as confidence based on history with no large shock in it. Test the response to a shock larger than any on record.
A second failure is buying insurance and calling it antifragility. Insurance can make a position robust; it rarely makes it gain from disorder. Ask where the upside comes from. A third is overpaying for options. Each one is cheap alone, but a drawer full of them becomes a steady drag that a long calm period exposes. Budget the total, and retire options whose scenario has passed.
When to use something else
- Tail-Risk Hedging. When the aim is protection against a named catastrophe at a known yearly cost, more than gaining from disorder.
- Barbell Strategy. When the question is how to split resources between very safe and small speculative bets across a whole portfolio.
- Convexity Reasoning. When the work is analyzing the curvature of a payoff precisely, beyond sorting a position into three classes.
Where it sits in SOLVE eX
Go to the source
- Antifragile: Things That Gain from Disorder, 2012