Skip to Content
logologo
AI Incident Database
Open TwitterOpen RSS FeedOpen FacebookOpen LinkedInOpen GitHub
Open Menu
Discover
Submit
  • Welcome to the AIID
  • Discover Incidents
  • Spatial View
  • Table View
  • List view
  • Entities
  • Taxonomies
  • Submit Incident Reports
  • Submission Leaderboard
  • Blog
  • AI News Digest
  • Risk Checklists
  • Random Incident
  • Sign Up
Collapse
Discover
Submit
  • Welcome to the AIID
  • Discover Incidents
  • Spatial View
  • Table View
  • List view
  • Entities
  • Taxonomies
  • Submit Incident Reports
  • Submission Leaderboard
  • Blog
  • AI News Digest
  • Risk Checklists
  • Random Incident
  • Sign Up
Collapse

Report 413

Associated Incidents

Incident 2830 Report
2010 Market Flash Crash

Loading...
What actually caused 2010 "Flash Crash"
businessinsider.com · 2016

Reuters The search for causation over the May, 2010 "flash crash," a one hour near 1,000 point loss and recovery on the Dow Jones Industrial Average, is a complex topic but one important to accurately document.

Such quantitative market mishaps only have the potential to cause more economic damage as society becomes increasingly dependent on technology and "artificial" intelligence.

Initially regulators publicly blamed a "fat finger" on the market sell-off that had no fundamental basis as a computer, not a finger, was involved. But a more nuanced and detailed report pointed to a single market participant for causation.

In support of the initial regulatory finding is a January 25 report challenging U.S. prosecutor conclusions, but does this report de-emphasize a key component?

Is analysis of what drives pricing — and a market participants providing two-sided liquidity that was simultaneously withdrawn during crisis— missing from meaningful analysis? This is not to diminish the report — it provided interesting insight on several levels.

It is to point out that any flash crash analysis might do well to consider how traders at the time were aware of certain market manipulation techniques relative to the underlying algorithmic market structure.

ValueWalk

Navinder Sarao was not solely responsible for the flash crash

A new report from two University of California and one Stanford University academics claim to know who should not be blamed. In a draft report titled "The Flash Crash: A New Deconstruction," they say it is "unlikely that… Navinder Sarao's spoofing orders, even if illegal, could have caused the flash crash."

Sarao faces criminal charges in Chicago for intentionally manipulating markets using the "spoofing" tactic during the flash crash and on other occasions.

Spoofing is a trading technique that is designed to deceive the electronic market making system into moving markets in a beneficial direction to the trader. The technique has been a concern as it could damage market structure security, particularly the case in 2010 as the market making systems at the center of the flash crash were in the early stages of development.

This "spoofing" statement is perhaps most interesting not for claiming Sarao didn't cause the flash crash - causation is likely a more complex and not singular in its conclusion and claims he was singularly responsible have been widely disputed.

But the report, in acknowledging the existence of Sarao's "spoofing" orders, points to a commonly known tactic that has been the subject of criminal attention from authorities.

In part, spoofing works by intentionally creating order imbalances in both quotes — messages sent into the exchange that have been interpreted as indications of market balance — as well as actual sell orders designed to be executed immediately.

ValueWalk

Key market security issue at center of spoofing charges

Such spoofing has been successfully prosecuted. In the past, authorities would treat harshly anyone who engaged in behavior deemed damaging to overall market structure security, including manipulating market systems or abusing segregated funds laws.

While the report comes to Sarao's defense in regards to his role as a "but for" cause of the flash crash — meaning if he was not involved, the crash would not have taken place — it might not tip the scales in court as it does not absolve Sarao of the more encompassing charges of market manipulation, an acknowledgement Sarao himself made in emails sent to his computer programmer designing the system.

Regarding flash crash causation, the report tends to side with the initial joint SEC-CFTC staff report on the topic that said "prevailing market conditions combined with the introduction of a large equity sell order implemented in a particularly dislocating manner" was the guilty party, pointing a finger at Kansas-city based mutual fund Waddell and Reed, the alleged "fat finger."

The report noted a "disruptive" trading pattern but did not provide analysis of the Waddell and Reed market order that was said to have rested on the higher "ask" side of the bid-ask spread and was done over multiple smaller orders.

This was hotly debated among practitioners at the time, and a then CFTC source claimed it caused significant tension between the two regulators. The losing force in the battle wanted to raise questions regarding known methods of manipulating the market making system and disputed the notion of pointing blame at any one organization.

The report then hits a key market security topic: "If Sarao's relatively small-scale trading could in fact generate the large-scale effects asserted by the government, modern equity market structures could be viewed as alarmingly fragile."

Those fighting a battle inside regulators, generally defending market security issues, wanted full academic discussion of the issue to be public.

ValueWalk

Market making software had "eye" feature that detected anomolistic behavior and order imbalances, but th

Read the Source

Research

  • Defining an “AI Incident”
  • Defining an “AI Incident Response”
  • Database Roadmap
  • Related Work
  • Download Complete Database

Project and Community

  • About
  • Contact and Follow
  • Apps and Summaries
  • Editor’s Guide

Incidents

  • All Incidents in List Form
  • Flagged Incidents
  • Submission Queue
  • Classifications View
  • Taxonomies

2024 - AI Incident Database

  • Terms of use
  • Privacy Policy
  • Open twitterOpen githubOpen rssOpen facebookOpen linkedin
  • e1b50cd