Is High Speed trading negatively impacting returns for the buy&hold investor?

With so much recent press, you are probably concerned that High Frequency Trading(HFT) is siphoning gains and making stocks more expensive for you to purchase.   

Both Algorithmic Trading and High Frequency Trading(HFT) were made possible by a SEC ruling in 1998 that enabled the creation of electronic exchanges. Once HFT trading became possible, the method didn’t come into existence until the early 2000s when computing power became inexpensive enough to justify the gains that could be obtained.

HFT is a method of trading that utilizes specialized algorithms to purchase and sell equities extremely quickly, taking advantage of minute spreads in prices and spreads. Typically, HFT operators compete with other HFTs. The fastest ones have the best algorithms, the fastest links to exchanges, and the fastest servers. Every aspect of the computing infrastructure has been tuned/overclocked/tweaked to run as quickly as possible. You’ll come across news articles on almost a weekly basis about some new data transmission method to reduce latency between the HFT operator and the exchange.

Here’s an analogy what will help you grasp HFT system architecture: Most servers are configured to handle high concurrent loads, such as a thousand users accessing a website at the same time.  Conversely, HFT servers are configured to run one application extremely quickly, and are tuned to prioritize that one application and a low number of network connections. Think of a multi-lane highway, vs a private one lane autobahn with no speed limit. You want one car going as fast as possible on an empty road, not thousands of cars/hour.

System Architecture: Differences between HFT & Enterprise servers

 

CPU

Memory

Disk

Network

OS

HFT System

1 per server, overclocked

High speed & overclocked

OS resident in memory

SSD, Application resident in memory

Close to exchange, low latency, fiber, custom designed NIC

One application has priority

Enterprise System

Many per server to process concurrent workloads

Large amount, commodity

HD, serves many transactions

Redundant connections, high bandwidth

Multiple applications share load

Since we’re told that the HFT outfits are all essentially competing against each other, how does that impact the long term investor? We don’t buy and sell in nanosecond timeframes, so why should we care if a stock goes up or down on such a short time period?

The first impact on the average investor is increased market volatility. To date, HFT systems have impacted the markets twice. At 2:42 PM on May 6, 2010, HFT systems created a flash-crash that caused the Dow to dive approximately 10% in minutes. (SEC report). While the market recovered much of the loss within the hour, there were significant impacts across the markets as stop-orders went into effect.  The Dow ended the day 3.2% lower, all due to HFT systems triggering massive selloffs.  Trading controls have been put in place since then, but we all know that murphy’s law has a way of returning. Then, on April 23rd 2013, HFT systems following twitter feeds caused the markets to drop by over 1% in three minutes after a false tweet stating that the White House had been hit by two explosions.  Other than existing market volatility safeguards, there are no guarantees stopping another flash-crash from occurring the next time these HFT systems go into a selling frenzy.

The second impact is an algorithmic procedure called Front Running. Some HFT systems try to detect when other market traders are purchasing an equity, and then get in front of them before the price goes up. Let’s say you read a positive article or tweet about a company and decide to buy the stock, you may be among many other people doing the same. If the HFT system catches on to that and makes a purchase before you, it just made your stock a bit more expensive. That impacts your gain. Conversely, it can be argued that when this occurs, your equity price is only negligibly higher. So for instance, you stand to lose out on 0.2% of the gain, essentially an insignificant amount for a small investor.  

However, that 0.2% or similar small percentage can mean a lot of money to larger investors such as managed funds, institutional investors or retirement funds.  If a HFT firm detects a pending purchase of 2 million shares from a large investor, it will buy shares just milliseconds before the purchase, and will then resell right back to the investor. If you participate in a large managed fund, this will certainly affect your overall return.  HFT systems will also try to get in front of monthly trades made by large managed & index funds that are investing 401k/pension savings.

Since only one significant market crash has been decisively blamed on HFT, and Front Running appears to be negligible for the small investor, we conclude that High Speed Trading doesn’t negatively impact the buy and hold investor in a significant way. However it can and does impact returns for larger investors.