Skip to main content
LTCM 1998

Crowded Trades and Liquidity: When Everyone Holds the Same Position

Pomegra Learn

Why Did a Hedge Fund's Strategy Become the Entire Market's Problem?

LTCM's convergence arbitrage strategies were, in principle, strategies that should not have been vulnerable to crowding. Arbitrage, by definition, involves identifying mispricings and trading against them — which should make arbitrage self-limiting. If everyone is doing the same arbitrage, the mispricing should be eliminated and the trade should no longer be attractive. But in practice, several features of LTCM's strategies made them susceptible to crowding in ways that were not captured by the fund's models. The on-the-run/off-the-run spread, the European convergence trade, and the equity volatility short all attracted imitation from banks' proprietary trading desks and other quantitative funds that had observed LTCM's early success. By 1997–98, LTCM was not the only holder of these positions — it was the largest, but far from the only. When LTCM's losses became apparent and the fund was forced to reduce positions, it was selling into a market where dozens of other large institutions were simultaneously trying to sell the same things. The result was a liquidity crisis that amplified LTCM's losses and threatened the stability of the broader market. The crowded trade concept — and its relationship to liquidity risk — became a central concern for risk managers and regulators in the years following 1998.

Crowded trade: A position held by many market participants simultaneously, such that aggregate selling (or buying) pressure from a single participant would be amplified by similar actions from others, creating market impact that is disproportionate to any individual participant's size.

Key Takeaways​

  • LTCM's strategies were imitated by many banks' proprietary trading desks and other quantitative funds, creating crowded positions that were not visible to any individual participant including LTCM.
  • Standard risk models treat positions as independent of other market participants' positions — they measure the market risk of a position but not the additional risk that arises from other participants holding the same position.
  • When a crowded trade reverses, all holders experience simultaneous losses and simultaneously try to reduce their positions, creating market impact (selling moving prices down further) that amplifies losses for all holders.
  • Liquidity — the ability to buy or sell at market prices without significantly moving the price — is not a fixed property of a market; it is determined by the current supply of buyers (or sellers). In a crowded trade exit, liquidity collapses because everyone is on the same side simultaneously.
  • The crowded trade mechanism is a form of endogenous risk — risk that arises from the actions of market participants rather than from external shocks — that standard models do not capture.
  • Contemporary risk management incorporates crowded trade assessments through measures of strategy distinctiveness, position concentration, and liquidity stress testing.

How LTCM's Trades Became Crowded​

LTCM's success in its early years was widely observed and discussed. The fund published research papers; its partners gave conference presentations; it was the subject of financial press coverage. Competitors who observed LTCM's strategies could and did implement similar positions.

The imitation process worked differently for different trade types:

On-the-run/off-the-run spreads. This trade was simple to implement — buying off-the-run Treasuries and shorting on-the-run Treasuries — and the rationale was clearly articulated in academic literature. Government bond desks at major banks, which needed to hedge inventory positions in any case, easily implemented the trade as part of their normal operations.

European convergence trades. As European Monetary Union approached and it became clear that spreads between European sovereign bonds would narrow, many funds and banks implemented similar trades. The theoretical rationale was compelling and widely understood; the trade was not proprietary to LTCM.

Equity volatility shorts. Selling long-dated equity options to capture the volatility risk premium was a recognized strategy among options traders. Banks' derivatives desks and specialized hedge funds all implemented various versions of the trade.

Swap spread positions. Fixed income arbitrage between swap rates and Treasury yields attracted practitioners from the entire interest rate derivatives community.

By 1997–98, LTCM was the largest holder of all these positions but far from the only holder. Goldman Sachs, Deutsche Bank, Lehman Brothers, and dozens of other institutions held similar trades of varying sizes. LTCM's risk models, which measured its own positions against market volumes, did not model the aggregate size of its strategy class across all market participants.

The Mechanics of Crowded Trade Collapse​

When LTCM's losses became apparent and news of its distress spread through the market, the crowded trade collapse unfolded in a specific sequence:

Step 1: News of distress. Rumors about LTCM's losses spread through the financial community from August 1998. Counterparties that had derivatives exposure were independently estimating LTCM's losses from observable market moves.

Step 2: Anticipatory selling. Other holders of similar positions — banks and funds that had implemented similar convergence strategies — began reducing their own positions in anticipation of LTCM's forced selling. Each institution rationally calculated that LTCM would eventually be forced to liquidate; better to exit before LTCM's selling drove prices down further.

Step 3: Anticipatory selling triggers exactly what was anticipated. The anticipatory selling by other funds made the spreads move wider before LTCM had even started selling. LTCM observed its positions deteriorating further; margin calls accelerated; LTCM was forced to sell sooner and faster than the situation otherwise required.

Step 4: Amplified market impact. When LTCM did sell, it was selling into a market where every other holder of the same positions was also selling. Market-making banks widened bid-ask spreads dramatically; the depth of market collapsed; prices moved substantially on relatively small volumes.

Step 5: Further losses and further selling. Each round of selling drove prices further against all remaining holders, triggering more losses, more margin calls, and more selling.

This self-reinforcing exit dynamic — in which each participant's rational individual response created a collective outcome worse than cooperation would have produced — is the prisoner's dilemma structure that the crowded trade literature describes.

Endogenous Risk​

The crowded trade concept is one manifestation of what economists call endogenous risk — risk that arises from the behavior of market participants rather than from external events.

Standard financial risk models treat market prices as exogenous — driven by external news, economic data, and events. A company's stock price falls because its earnings disappoint; a bond's yield rises because inflation increases; a currency falls because its trade deficit widens. In this framework, markets aggregate external information and prices reflect fundamental values plus noise.

Crowded trades create endogenous risk: prices move because of who holds positions and what they do with them, not because of fundamental economic developments. The on-the-run/off-the-run spread widened in August 1998 not because the economic difference between a six-month-old Treasury and a new Treasury had changed — it hadn't — but because many institutions were simultaneously trying to sell off-the-run bonds to raise cash. The price movement was entirely endogenous.

Endogenous risk has a specific implication for diversification: if the source of risk is the behavior of other market participants, holding a "diversified" portfolio of positions all owned by the same market participants provides no diversification protection. When those participants sell, everything in the portfolio falls.

Measuring Crowded Trade Risk​

LTCM's models did not measure crowded trade risk because they had no information about other participants' positions. This is the fundamental challenge: measuring crowded trades requires knowing what others are holding, and that information is typically unavailable.

Contemporary approaches to estimating crowded trade risk:

Strategy distinctiveness analysis. Portfolio analysis that identifies whether a position is likely to be held by many other managers. Positions in highly liquid markets with transparent theoretical rationale are more likely to be crowded than obscure positions in specialized markets.

Short interest and borrowing costs. For equity positions, short interest data and stock borrow costs provide indirect evidence of crowding. High short interest suggests many managers hold the same short position; high borrow costs suggest heavy demand for shares to short.

Factor exposure analysis. If many funds have similar factor exposures — similar loadings on value, momentum, carry, and other systematic factors — their portfolios will behave similarly in factor stress events even if their individual holdings are different.

Position concentration indicators. In markets where position data is available (futures, some options), concentration metrics show when a small number of large holders dominate open interest.

Implied volatility asymmetry. When market participants pay for downside protection but not upside, the skew in options pricing implies crowded long positions (participants hedging against a position they hold).

None of these measures is definitive, but together they provide signals about crowded trade risk that LTCM's models could not have captured.

Policy and Regulatory Implications​

The crowded trade problem has specific implications for both investment management practice and financial regulation:

Position reporting to regulators. If regulators had access to aggregate position data across all market participants, they could identify crowded trades before they became crises. The absence of such data was a key gap in 1998. Post-crisis regulatory initiatives — including mandatory reporting of large derivatives positions under Dodd-Frank — partly address this gap, though reporting remains incomplete and regulatory analysis of systemic crowding is limited.

Leverage limits. Crowded trades are most dangerous when combined with leverage. A crowded trade without leverage produces losses when it reverses but not a cascade. A leveraged crowded trade produces forced selling that amplifies losses for all holders. Leverage limits on investment funds — proposed in various forms but not comprehensively implemented — would reduce the amplification component of crowded trade risk.

Concentration limits in fund mandates. Investment managers can implement position concentration limits that prevent any single strategy or market from representing too large a fraction of the portfolio. These limits reduce the crowded trade exposure by ensuring that forced liquidation of one strategy cannot destroy the entire portfolio.

Market maker capacity requirements. In a crowded trade exit, the collapse of market liquidity reflects market makers' unwillingness or inability to absorb selling at reasonable prices. Capital requirements for market makers that are sensitive to market conditions — rather than pro-cyclically shrinking when they are most needed — can reduce the severity of liquidity crises.

Common Mistakes in Analyzing Crowded Trades​

Assuming crowded trades are visible. The most dangerous aspect of crowded trades is precisely that they are invisible to individual participants. Each institution knows its own positions but not the aggregate position of all similar strategies. Post-crisis identification of crowded trades is easy; pre-crisis identification is difficult.

Treating crowding as always a warning sign. Some "crowded" trades — such as long equity positions during bull markets — are crowded because they are genuinely attractive, and the crowding does not prevent continued appreciation. The danger is crowded trades with leverage that require simultaneous exit in crisis conditions. Not all consensus positions are dangerous.

Ignoring the endogenous component of liquidity. Liquidity is sometimes discussed as if it were a fixed property of a market — Treasury bonds are liquid, distressed debt is illiquid. In practice, liquidity is endogenous to the current composition of holders and their financial condition. A market is liquid when potential buyers are ready to provide liquidity; it is illiquid when all current holders want to sell simultaneously. Treasury bonds became illiquid (for off-the-run) in 1998 not because of any change in their fundamental properties but because of the crowded trade exit.

Frequently Asked Questions​

Can individual investors be affected by crowded trade risk? Yes. When large fund managers all hold similar positions — for example, when passive index funds all hold similar weight in "popular" stocks — individual investors who hold those funds are exposed to crowded trade risk. The 2018 "momentum crash" and the 2020 growth factor decline both reflected crowded factor exposures unwinding. Individual investors can assess their crowded trade exposure by examining whether their holdings have high overlap with popular factor strategies.

How has the crowded trade problem changed with algorithmic trading? Algorithmic trading can both increase and decrease crowded trade risk. Algorithms that react to the same signals — price momentum, volatility changes — can amplify crowded trade exits by all responding simultaneously to the same triggers. Algorithms that provide liquidity (market-making algorithms) can reduce liquidity gaps. The net effect depends on the specific market structure and the types of algorithms present.

Did post-LTCM regulations prevent a repeat crowded trade crisis? Partially. The 2008 global financial crisis involved crowded trade dynamics — mortgage-related securities and the carry trade both exhibited crowded positioning. The crowded trade mechanism was not eliminated by post-LTCM reforms; it recurred in different instruments. The post-2008 reforms (Dodd-Frank position reporting, higher capital requirements) have reduced leverage but not fully addressed the fundamental problem that aggregate crowding is invisible to individual participants.

Summary​

The crowded trade problem was a critical but invisible dimension of LTCM's crisis. LTCM's convergence arbitrage strategies had been adopted by many banks' proprietary trading desks and other quantitative funds, creating aggregate positions far larger than LTCM's alone. When LTCM's distress became apparent, other holders anticipated forced selling and began exiting simultaneously — amplifying the price impact of LTCM's actual liquidations. Liquidity collapsed in the specific markets where LTCM held its largest positions; market-making banks widened spreads and reduced depth; prices moved substantially on small volumes. The crowded trade exit produced endogenous risk — price movements driven by market participant behavior rather than fundamental economic developments — that LTCM's models could not capture because they had no information about other participants' positions. The crowded trade concept has since become a standard element of risk management, though the fundamental measurement challenge — aggregate crowding is invisible to individual participants — means the problem cannot be fully solved through individual fund risk management; it requires either regulatory position reporting or macro-prudential surveillance capable of identifying system-wide crowding.

Next​

→ The Federal Reserve's Intervention