Skip to content

Quant Bust 2020: summary and key takeaways

The paper explains in non-technical terms why dollar-neutral quant strategies, mainly mean-reversion statistical arbitrage, suffered drawdowns in the COVID-19 selloff. It argues the strategies depend on stable stock correlations that forced long-only liquidation breaks. Illustrative backtests show mean-reversion drawdowns and gains for momentum and overnight signals.

Kakushadze, Z. · 2020 · SSRNSummary updated 20 November 2026
Published2020Working paper
Asset focusEquitiesQuant drawdowns
TypeResearchQuant Crises & Methods
Reading time4 minthis summary

Quant Bust 2020 at a glance

The question
Why did dollar-neutral quant strategies such as equity StatArb lose money during the COVID-19 market selloff, and when do they work?
Dataset & period
U.S. exchange-listed common stocks and class shares from Yahoo Finance, downloaded March 25, 2020. Backtest covers 252 trading days with P&L through March 23, 2020, daily frequency. Universe is the 2,000 most liquid stocks by average daily dollar volume. Assumed capital is $20M, split $10M long and $10M short.
Method
The main text is a conceptual argument. It treats StatArb profit as a trickle-down effect from long-horizon long-only investors, who create stock correlations that shorter-horizon strategies exploit. The appendix backtests stripped-down intraday strategies that open at the open and close at the close. Signals use close-to-close mean-reversion over several horizons, a one-day momentum signal and an overnight signal. Portfolios are built by bounded weighted regression or bounded mean-variance optimization, using SIC industries or a statistical clustering classification.

Key findings of Quant Bust 2020

  • In the paper's account, the selloff was driven by long-only liquidation. Index funds and futures were sold in ways that did not preserve normal correlations between stocks in the same industry.
  • Mean-reversion StatArb keeps adding to losing positions as correlations come undone. It also faces poor fills on short sales and hard-to-borrow stocks, so it loses on both sides of the trade.
  • The paper compares the fall in U.S. market capitalization, from 158.9% of GDP to 103.4% of GDP (a 34.93% peak-to-trough drop), with the far smaller assets held in equity market neutral funds.
  • In the 2019 rally the S&P 500 returned 30.43%. BarclayHedge's equity market neutral index returned -0.56%, equity long/short 6.59% and multi strategy 5.23%.
  • In the backtests, close-to-close mean-reversion strategies had drawdowns of around 10% for 1-day strategies and around 15% for longer-horizon SIC strategies, and deeper for longer-horizon statistical-classification strategies.
  • Momentum and the overnight (delay-0) signal showed P&L spikes during the selloff. A strategy short oil stocks could also have gained, as WTI fell from $53.86 to $23.36 a barrel.

Robustness, as reported in the paper

  • Backtests are reported both before and after simulated trading costs, averaging 10 bps to trade an equally weighted portfolio.
  • Two portfolio constructions and two industry classifications are compared. Optimized SIC strategies did better than regression-based ones.
  • Risk models are recomputed fully out-of-sample, and the universe is refreshed periodically.

Caveats and limits

  • The backtests are described as illustrative and not exhaustive. They cover one year of data.
  • The simulated intraday trading at open and close is not how strategies are run in practice, and the cost model likely understates selloff slippage on sells and short sales.
  • The overnight delay-0 results are flagged by the paper as needing caution because the signal uses the same-day open price.
  • The central explanation is argued conceptually rather than tested with fund-level data (inferred).

Items marked (inferred) are our reading of the design, not statements by the authors.

Why it matters

Dollar-neutral does not mean market-neutral: the paper likens mean-reversion StatArb to selling deep out-of-the-money puts. Practitioners can use it to think about which regimes suit liquidity-providing strategies and why diversifying across strategy types matters.

Questions about this paper

Why does dollar-neutral StatArb lose in a selloff?

Long-only liquidations break the normal correlations between stocks, so mean-reversion positions keep losing instead of snapping back. Liquidity for short sales also dries up.

In which market does mean-reversion StatArb work best?

A sideways market with decent but not excessive volatility. Very quiet markets offer few mispricings, and choppy ones bring high slippage.

Could machine learning have avoided the losses?

The paper says no. It argues ML cannot foresee black-swan events any better than simple linear regression, and can obscure what a strategy is doing.

Where can I read the original paper?

The original is at its publisher: https://doi.org/10.2139/ssrn.3570280. We link to the paper and do not host the PDF.

Follow new quant research

Download MacroFinance market data as CSV or Excel, or pull it by API, and test the ideas in these papers yourself. New summaries are added as we review them.

Create a free account
Summaries are written by MacroFinance from the published papers, in our own words, for information only. They are not investment advice and may omit detail. Papers are not hosted here: each page links to the original at its publisher.