High Frequency Trading: Its Role In Flash Crashes

A flash crash is a sudden and dramatic drop in prices - warranted or unwarranted - that seems to quickly spiral out of control.

= = =

This is part three of a four-part series on High Frequency Trading (HFT). In the first article the use of computers to quickly identify and execute trades was discussed. In the second article, the belief that data is unfairly disseminated was discussed. Today's topic is the possible role HFT plays in flash crashes.

= = =

THE CRASH OF 2010

The flash crash that occurred on May 6, 2010 was the largest single-day point swing and biggest one-day decline for the Dow in its history. There are still debates today as to how it started - a fat-finger error that sent Procter & Gamble's stock down, a large sell order in the S&P 500 E-mini contract are two popular culprits - but since HFT is hated by some many, especially lay people who know nothing about Wall St., HFT has received the bulk of the blame for exaggerating the movement.

Why not? Everyone hates HFT, so let's blame the single biggest one-day drop on the activity in hopes of drumming up support for legislation to be passed to ban the practice. Or when you lose money, let's use the flash crash as an excuse as to why you're losing to distract yourself from reality.

But while this is convenient, history tells us crashes have existed as long as the market has been in operation.

THE CRASH OF 1987

The market dropped 22.61% on Monday, October 19, 1987 --the largest single-day drop, both in terms of points and percentage, in Dow history. Volume was twice what the previous record had been.

It wasn't uncommon for individual stocks to drop 20% in one day when they missed earnings or released other bad news. But the entire market?

A bad day today would have the Dow down 500 points, but a drop today that matched the 1987 drop would be around 3,500 points. During the 2010 flash crash, the Dow dropped 1,000. Can you imagine 3.5 times that?

There were no bids. Once the selling started there was no way to exit. Fortunes that took a lifetime to build were wiped out in a single day. Traders with long track records lost everything.

Guess what? Computers, as we know them and currently use them, didn't exist in 1987. Yet this crash still happened.

THE CRASH OF 1962

On May 28, 1962, the market crashed. The Dow fell 6% in one day.

IBM dropped 5.3% in 19 minutes. Brunswick Corp. (BC) fell 22.3% from its opening price.

Volume was so heavy, it took almost 2-1/2 hours beyond the closing bell for the ticker to finish reporting all the floor transactions.

It was the biggest single day drop since 1929.

Guess what? Computers, as we know them, didn't exist in 1962. Yet this crash still happened.

THE CRASH OF 1929.

In late October the market melted down, and then, on what became known as Black Tuesday, the bottom fell out. On that single day, 16.5 million shares traded. To give you an idea how heavy that volume is, the market did not trade 16.5 million shares on a single day again until 1969 - 40 years later.

Guess what? Computers didn't exist in 1929. Yet this crash still happened.

THE BOTTOM LINE

Crashes have always occurred, and they will continue to occur. For better or worse, they're part of the market, computers or no computers, high frequency trading or no high frequency trading.

Crashes occur when bids disappear, and it doesn't matter if bids are being pulled by a human or a machine. The bids disappear and the market free-falls.

If you're not a profitable trader, you'll desperately blame anything. HFT is currently the du jour excuse, but don't kid yourself. If you're not making money, you're doing something wrong. Figure out what it is; make a change. Don't blame outside forces, and especially don't blame HFT for flash crashes because crashes have been occurring since before computers even existed.

= = =

Jason Leavitt is the founder of LeavittBrothers.com, a trading advisory service.

4 Comments

Join In on this conversation, post a comment below.
Visitor - Sinister Footwear: Mary Schapiro, ex SEC Chair correctly pointed out that May 6th differed vastly from 1962 and 1987 - and we are not just talking speed. On May 6th, "market makers" took liquidity aggressively, which is in sharp contrast to the volatility dampening and capital deployment done by market makers in 1962 and 1987. The HFT Market Makers on May 6th aggressively front ran each other to the point of hitting penny bids. Also, the May 6th crash was structural - the plumbing. 50 destinations woven together each with different speeds and capacities, and data disparity, plus data fueled algos makes for... fun times. Respectfully look at the difference between the crashes and you will see what I am talking about, and what Mary Schapiro has stated - privy to the real data.
Reply Flag
Visitor - Fred Simons: Spreads or relational trades work until they do not. Correlations are statistical. Always remember that for a long time that there was a positive correlation between the number of storks that annually nest in Sweden and the birth rate in that country. LTCM also discovered that correlations only last until they do not.
Reply Flag
Visitor - Fred Simons: The 1987 crash was caused by the delinkage of a hedge trade that was sold to institutional investors (pension funds). This was the deal.. you are long stocks, we will sell you a hedge based on the futures market.. long the physical, meaning stocks and hedged against a decline being short the futures. Futures traded at a premium to the physical. This reversed and that meant lock in the new spread.. sell the physical and buy the futures ad infinitum until the spread reverses. Finally the FED engineered the reversal by opening up the futures mnarket at a premium.. Fred Simons...
Reply Flag
Visitor - see3d: Crashes have happened every once in a while in the markets. However, after each one, the causes are dissected and changes are made to the market trading rules to prevent out of control crashes in the future from the same circumstances. The advent of HFT has significantly changed the trading environment and it must be examined for its role in creating conditions that increase instabilities or excess volatility. Rules have to keep evolving with the changing trading methods to support an orderly trading environment. Of course, it HFT is not found to be a benefit to efficient markets for price discovery and liquidity, it should be banned as an unnecessary overhead. Whichever way it goes, it should be examined objectively by those without any conflicts of interest.
Reply Flag