Dev Loses $440 Million in 28 minutes | The Night Capital Catastrophe

On August 1st, 2012, a quiet morning in New York soon erupted into chaos on Wall Street. Knight Capital, one of the largest trading firms, faced a catastrophic failure, losing a staggering $440 million in less than an hour. The question remains: who is to blame for this debacle – man or machine?

The Prelude to Disaster

As the sun rose over the New York skyline, pre-market trading was set to commence at 8:00 a.m. EST. Night Capital's internal system sent out 97 email alerts a minute later, indicating an error labeled "Power Peg disabled." Unfortunately, these alerts were not prioritized and went unnoticed by the staff, setting the stage for a disaster.

The Unraveling

When the market opened at 9:30 a.m., traders quickly realized something was amiss. Within a minute, the market was flooded with an unusually high order of specific stocks. By 9:31 a.m., the surge in trading activity had caught the attention of many on Wall Street. By 9:34 a.m., quantitative analysts traced the issue back to Night Capital.

In a desperate attempt to contain the situation, the New York Stock Exchange (NYSE) tried contacting Night Capital's CEO, Thomas Joyce, who was recovering from knee surgery at home. Meanwhile, Night's trades accounted for more than 50% of the trading volume on the NYSE, with positions worth $148 million being opened every minute.

The Core Issue

The chaos resulted from a new dark pool trading system called the Retail Liquidity Program (RLP), which launched on the same day. Dark pools are private exchanges allowing institutional investors to trade large blocks of securities away from the public eye. While these pools provide liquidity, they also raise concerns about market fairness.

Night Capital, a major market maker, had updated its trading infrastructure to prepare for the RLP launch. However, a series of critical errors in their code deployment led to the disaster. The new RLP code inadvertently activated an old, deprecated algorithm called Power Peg. This algorithm, designed to lose money for testing purposes, ran unchecked due to missing safeguards in the code.

The Deployment Failure

A week before the launch, an engineer deployed the updated code to seven of eight servers, leaving one with the old code. This oversight went unnoticed due to the absence of a secondary review process. When the markets opened, this lone server executed trades using the flawed Power Peg algorithm, resulting in a relentless barrage of orders that drove prices up uncontrollably.

The Fallout

In the ensuing 45 minutes, Night Capital's systems executed 4 million trades across 154 stocks, amassing positions worth $6.65 billion. With only $365 million in cash, the firm was on the brink of collapse. Attempts to cancel the trades through the SEC were denied, and a rollback of the code only exacerbated the losses.

The Aftermath

By 9:58 a.m., engineers finally identified the root cause and shut down the system. Goldman Sachs intervened, buying Night's positions at a discount, but the damage was done. Night Capital faced a $440 million hole, receiving a $400 million cash infusion from investors a week later. Ultimately, the firm was acquired by GetGo LLC the following summer.

Timeline of the Night Capital Debacle

August 1, 2012

  • 8:00 a.m. EST: Pre-market trading on the New York Stock Exchange (NYSE) is set to commence.
  • 8:01 a.m.: 97 email alerts were sent from Night Capital's internal system indicating an error labeled "Power Peg disabled"; the staff ignored these alerts.

Market Opens

  • 9:30 a.m.: Market opens; an abnormal surge in trading volume is immediately noticeable.
  • 9:31 a.m.: The market is uncontrollably flooded with high orders on specific stocks.
  • 9:32 a.m.: Wall Street traders and analysts outside of Night Capital notice the ongoing issue.
  • 9:34 a.m.: Quantitative analysts trace the surge in volume back to Night Capital.


  • 9:35 a.m.: NYSE Chief Executive Officer Duncan L. Niederauer tries to contact Night Capital CEO Thomas Joyce, who is recovering from knee surgery at home.
  • 9:36 a.m.: Orders from Night Capital start pushing stock prices by over 10%.
  • 9:40 a.m.: Night Capital's trades constitute more than 50% of the trading volume on the NYSE.
  • 9:42 a.m.: Night's Chief Information Officer is contacted to flip a kill switch, but no such kill switch is in place.

Uncontrolled Trading

  • 9:45 a.m.: Engineers at Night Capital struggle to diagnose the issue in a live trading environment.
  • 9:50 a.m.: Night Capital realizes they are losing $48 million per minute and orders $2.5 million worth of trades every second.
  • 9:55 a.m.: Night Capital's positions have reached $6.65 billion in open trades with only $365 million in cash.

Attempted Resolution

  • 9:57 a.m.: Night Capital tries to roll back the code, believing the bug is in the new deployment.
  • 9:58 a.m.: Engineers identify the root cause and shut down SMARS on all servers.


  • 9:59 a.m.: Goldman Sachs buys Night Capital's positions at a discount, amounting to a $440 million loss.
  • 10:00 a.m.: The NYSE continues average trading as the situation is resolved.

Subsequent Developments

  • August 8, 2012: Night Capital receives a $400 million cash infusion from investors.
  • Summer 2013: Night Capital is acquired by rival firm GetGo LLC.

Key Events Leading to the Debacle

  • October 2011: NYSE proposes the Retail Liquidity Program (RLP), sparking controversy.
  • June 2012: The SEC grants approval for the RLP, which is set to go live on August 1, 2012.
  • Late July 2012: Night Capital's software development team rushes to update trading infrastructure for the RLP launch.
  • One Week Before August 1, 2012: An engineer mistakenly deploys the new SMARS code to seven out of eight servers, leaving one with the old code.
  • August 1, 2012, 8:01 a.m.: Initial email alerts about the "Power Peg disabled" error are ignored.

The Night Capital incident unequivocally underlines the fragility of high-frequency trading systems and the absolute necessity for rigorous testing and robust fail-safes. Human error and technological flaws undeniably contributed to the debacle, emphasizing the imperative need for vigilance in an increasingly automated financial world.