2004: US Housing Bubble — The Low-Volatility Illusion

Spring 2004. A real estate agent in Phoenix shows the same house to three buyers in one week, each competing with an all-cash offer above asking. Across the country, in city after city, the same scene repeats with minor variations. The Federal Reserve's federal funds rate sits at 1.0%, mortgage rates hover near multi-decade lows, and a generation of lenders has discovered that borrower credit quality matters less when rising collateral values absorb the risk. The mood is not euphoria—market participants in 2004 are not consciously aware they are inside a bubble. They are simply doing what the data and the incentives tell them to do.

The numbers that followed are well documented. In the 30 months from April 2004 through September 2006, the US 20-City Composite Historical Replay: 2004-04 to 2006-09 shows a market climbing from 158.47 to 205.8, a nominal gain of 29.9%. The maximum drawdown across those 30 monthly bars was 0.3%, and the largest single monthly move was 2.0%. These are not the statistics of a market in crisis. They are, in fact, the statistics of a market that felt extraordinarily safe.

The danger of such stability is that it trains participants to ignore what they cannot see. Phoenix offers the most extreme example within the US data set. The Phoenix Historical Replay: 2004-04 to 2006-09 records prices rising from 130.1 to 225.2, a 73.1% climb in 30 months, with a maximum drawdown of 1.0% and a largest monthly move of 4.9%. A participant in this replay faced what appeared to be a market that rewarded courage and punished hesitation—and indeed, directional bets during this window were largely rewarded. But the same data describes a system in which successive cohorts of buyers had levered themselves against collateral whose own appreciation was being used as justification for additional leverage. The smoothness of the path was not evidence of health; it was a product of the specific moment in the credit cycle.

The low-volatility environment had a second, subtler effect on judgment. When drawdowns stay near zero for an extended period, risk management begins to feel like an anchor on performance rather than a protective mechanism. FinFable's scenario data for this window shows that a maximum drawdown of 1.0% in Phoenix could plausibly lead a participant to increase position size under the assumption that the market had permanently changed character. The assumption that low historical volatility implies low future volatility is one of the most persistent errors in financial decision-making, and the 2004–2006 US housing market provided a vivid real-world laboratory for it.

Not all US cities moved at the same pace. Atlanta gained 12.5% with a maximum drawdown of 0.1%, while Boston managed only 7.3% with a 3.4% maximum drawdown—the highest drawdown among the 2004–2006 US city indexes. Seattle climbed 42.8% with a 0.0% maximum drawdown across the same 30-month window. These variations mattered because they allowed market observers to construct local narratives—tech spillover for Seattle, supply constraints for Boston—that fit the data without requiring acknowledgment of the shared credit expansion driving all of them. The Seattle Historical Replay: 2004-04 to 2006-09, with its 0.0% maximum drawdown and steady 1.9% largest monthly move across 30 months, presents a market that rewarded patience and consistency. Participants in that replay confront the psychological challenge of whether to take profit when no correction has provided a natural exit, or whether to maintain or increase exposure on the grounds that the trend is clearly established.

The 30-month replay windows capture a market at a specific, carefully chosen phase of its trajectory—specifically, the phase before the reversal. They are instructive precisely because the volatility they record is artificially compressed relative to the full cycle. Consider the contrast when the same indexes are viewed through a longer lens. The US National (FHFA, 1975Q1+) Historical Replay: 2004-04 to 2011-07 records a maximum drawdown of 18.7% and a largest single monthly move of 3.8% across the extended window. The same asset class, viewed across a period that includes the crisis, produces numbers that contradict the 0.2% drawdown and 1.6% monthly move of the 2004–2006 segment. The lesson is not that the extended data is more accurate; both windows describe real market behavior. The lesson is that a period of suppressed volatility can coexist with the conditions that will eventually produce its opposite.

The same principle holds across national boundaries. The UK National (BIS) Historical Replay: 2004-04 to 2011-07 records a net loss of 5.5% with a maximum drawdown of 22.5% and a largest single monthly move of 6.7%—numbers that dwarf the calm of the US city indexes. Bulgaria's national index delivered a 20.3% net gain but with a maximum drawdown of 40.1% and a largest single monthly move of 12.3% across the same period, the most volatile national record in the data set. Even China's modest 2004–2006 window, recording a net decline of 2.4% with a 5.0% maximum drawdown, illustrates that no national market navigated the era without intermediate stress. The international data forces a reckoning with the psychological assumption that US markets were uniquely stable during this period.

The deeper structural point is that suppressed volatility is itself a risk factor. When a market consistently produces maximum drawdowns below 1%, it generates institutional narratives of safety that become self-reinforcing. Lenders reduce cushions. Borrowers accept terms they would not accept in a volatile market. Regulators observe stable performance and relax scrutiny. Participants in replay scenarios confront this dynamic at first hand: a 0.0% or 0.1% maximum drawdown is not merely a data point about the past, but an active influence on how participants behave within the simulation. The challenge is not to identify the reversal that comes later—it is to maintain disciplined process during the period when the data actively rewards the opposite.

For financial institutions and regulators, the 2004–2006 housing episode reshaped several fundamental assumptions. Mortgage underwriting standards were tightened in ways that permanently altered the relationship between credit access and collateral values. Rating agency models for structured products incorporating residential mortgages were revised after the crisis demonstrated the inadequacy of historical default rate assumptions. Central bank frameworks incorporated asset price valuations into inflation assessment rather than treating them as outside the mandate. The regulatory response produced new requirements around disclosure, capital adequacy, and risk retention that altered how credit is allocated across the economy.

For participants using these scenarios as decision training, the 2004–2006 window offers a specific and reproducible teaching moment: how consensus about market direction can coexist with an underappreciated accumulation of systemic fragility. The compressed volatility of the pre-crisis years is not a curiosity from a distant past. It is the mechanism through which risk perception diverges from risk reality, and it is a pattern that recurs across asset classes and decades whenever credit expands faster than underlying cash flows can support.

Phoenix's trajectory captures the extreme end of the US experience: a 73.1% gain with a 1.0% maximum drawdown and a 4.9% largest monthly move. Seattle's 42.8% gain with a 0.0% maximum drawdown illustrates how sustained stability can create conditions for behavioral complacency. The US National (Case-Shiller, 2000+) index, recording a 26.0% gain with a 0.2% maximum drawdown and 1.6% largest monthly move, represents the aggregate picture that obscured these local extremes. The contrast between US city replays and international national replays—Bulgaria's 40.1% maximum drawdown, the UK's 22.5%—underscores that the lesson is not about a single market but about the universal tendency to mistake low-volatility environments for low-risk ones.