The Dot-Com Bust, 2000Start here

A technology transformation that was entirely real, and a set of prices that were not. Why being right about the future is not the same as being right about the price.

What happened

• 1995–1999 — internet adoption accelerates. Companies with little or no revenue list at large valuations; a technology-heavy index rises roughly fivefold over five years.
• 1999 — the pace intensifies. Companies routinely double on their first day of trading, and analysts adopt metrics such as revenue multiples and "eyeballs" because earnings-based ones return no meaningful number.
• March 2000 — the technology index peaks.
• 2000–2002 — it falls approximately 78% from that peak. Many companies go to zero; some of the largest survivors fall 80–90% and take many years to recover their highs.
• Meanwhile the underlying prediction proves correct. The internet did transform commerce, media and communication, on roughly the timeline the enthusiasts described.

The mechanism: correct thesis, wrong price

• The thesis was right and the arithmetic was not. A company can grow into a valuation, and the question is always whether the required growth is plausible. That question has an answer, and it can be computed.
• When earnings are negative, valuation moves entirely into the terminal value. The DCF calculator reports what share of the answer sits beyond the forecast period; at these valuations essentially all of it did, which means the whole result was one perpetuity assumption.
• Invented metrics are a symptom, not a technique. When conventional measures produce uncomfortable answers, new measures appear that produce comfortable ones. Each one may be defensible individually; the pattern of replacement is the signal.
• Aggregate arithmetic constrains individual stories. The sum of the implied market shares in a sector's valuations frequently exceeded the plausible total size of the market. Every individual company's story can be coherent while the set of them is impossible.
• Index concentration transmitted it. Cap-weighted indices had accumulated very large technology weights by the peak, so "diversified" holdings carried a concentrated sector bet nobody had chosen — the structural point in active versus passive.

The drawdown arithmetic

• A 78% fall requires a 355% gain to recover. The drawdown calculator gives the number, and the number is the reason recovery took roughly fifteen years for that index.
• The convexity is the lesson, not the specific figure — see the arithmetic of drawdowns.
• Survivorship distorts the retrospective view. Charts of the companies that survived show a spectacular recovery. The companies that did not survive are not in the chart, and they were the majority — the bias described in how to read a market number.
• Recovery of an index is not recovery of a holder. Anyone who sold during the fall, or who was forced to, realised the loss permanently regardless of what the index did later.

What it teaches

• Being right about the technology is not being right about the investment. These are separate claims and the second requires a price.
• Growth assumptions have arithmetic limits. Compound growth extrapolated far enough always produces absurdity; the discipline is checking where the absurdity begins.
• Valuation is not a forecast; it is a translation. It converts a price into the set of assumptions required to justify it. Whether those assumptions are plausible is then a judgement, but at least it is the right judgement.
• New metrics deserve scepticism proportional to their convenience.
• Position sizing is what survives being wrong. No analysis reliably distinguishes a transformative company from an expensive one in advance; sizing does not require the distinction — see how to size a position.
• Concentration arrives without a decision. Cap-weighted holdings drift toward whatever has risen most, which is exactly the point at which they are riskiest.

The part that is genuinely hard

• Some of the extreme valuations were justified in hindsight. A small number of companies did grow into and far beyond their 1999 prices. This is not a story where everyone was simply wrong.
• The distribution was extremely skewed: a handful of enormous successes among a very large number of total losses. That is a specific distributional shape, and it has clear implications for how such exposures should be sized and diversified rather than for whether they should exist.
• Bubbles are only unambiguous afterwards. At the time, the disagreement was sincere on both sides, which is why "obvious bubble" is a description available only in retrospect — see behavioural finance.