Latency Arbitrage with Predatory Algorithm for Trading

Latency arbitrage is the practice of one party, perhaps a predatory HFT firm, exploiting a time disparity and earning profits with a computer algorithm for trading, when that trade is executed solely because of a latency advantage.  Latency arbitrage has raised many heated discussions among all market participants, the SEC and government law makers for many years, yet this unfair unequal access to US equity markets is still the main strategy of many predatory firms.  

Let’s take a look at what exactly latency arbitrage means, why it occurs, and what we can do about it.  

fiber opticsLatency Arbitrage Defined

An arbitrage occurs when a simultaneous purchase and sale of a stock asset executed by an algorithm for trading earns a profit based on a difference in price.  Latency refers to the time that a firm receives the same publicly traded stock information compared to other firms, as not all firms receive the same information at the same time.  Thus, a latency arbitrage occurs when a firm earns a profit from the purchase and sale of stock, and when those transactions were executed because of a latency advantage.  

Firms will pay huge premiums to co-locate their equipment next to the exchange’s servers, and pay a hefty price for premium data feeds, basically to reduce their latency.  A reduced latency combined with cutting edge technology and purchases of raw data feeds allows these firms to see the NBBO substantially quicker than what is available publicly through the SIP (Securities Information Processor).

Let’s say that a firm issues a buy order to pay the midpoint of the NBBO (the National Best Bid and Offer) for stock XYZ, and the current market for XYZ $10.10 x $10.11. Their bid will be at $10.105. Predatory HFT firms, using faster data feeds and co-location to reduce the transmission times, may see that the $10.10 bid is now gone and the market is now $10.09 x $10.10. However, the NBBO has not changed yet, since the SIP is slower than the HFT firms, so that midpoint order is still resting at $10.105. For the HFT firm, it is simply a matter of selling at $10.105, and immediately buying back at $10.10, making a half penny.

While the half penny earned may seem miniscule, keep in mind that their computers are doing this all day long, and all those pennies per share add up to billions of dollars.  Billions of dollars that is basically skimmed off the top of your average middle class retirement fund, college savings, and hard earned investments.  

Add to that the fact that the firm performing the latency arbitrage assumed no risk whatsoever.  By seeing the true market before the rest of the world adjusts their orders, they know that their trade is a pretty sure bet.  All due to a speed, or latency, advantage.

This is extremely frustrating for the firm placing the original order.  When they originate the order, they are trying to get a fair price in the middle of the spread.  But at the time of execution, they actually ended up paying beyond the best offers in the market.

Another scenario arises due to the multiple venues where stocks can trade. Various dark pools, as well as the national exchanges, can each have liquidity available, but as you go through collecting that liquidity, HFT firms can jump in front of you. Say the same firm above was trying to buy 5000 at $10.11, and the NBBO showed that much available at that price. But as the firm starts to send orders to different venues to buy that amount, HFT firms see that activity, and jump ahead of the order to take everything at $10.11. The original firm might only get a few hundred shares, instead of the 5000 they saw when they placed the order. Now they have to pay $10.12 if they want to buy the balance.

A trader can get whiplash from this ‘now you see it, now you don’t’ scenario.  Just by placing their order, they alert the HFT’s of their intentions, and can almost plan on closing at the higher price, unless they find a way to play in the latency game.  

Why Latency Disparity Exists

Of course, Reg NMS and the current fragmented market structure are obvious contributing factors to latency problems, although there a couple of reasons that stand out as to why firms do not receive the same information at the same time.  One cause contributing to this latency disparity is of course the speed of the SIP, another is the fact that many firms co-locate equipment to the servers of the exchanges.  This speed advantage, combined with a predatory algorithm for trading, grants a very unfair advantage with access to see NBBO data before trading firms with a slower connection.  

SIP

The SIP is tasked with consolidating all stock market data, then distributing this data at the same time to all subscribers.  While direct data feeds are historically faster than the consolidated feeds, the SIP has put much focus on improving its latency.  While some may think that reducing the SIP latency would create a more efficient and fair market, the issue of latency arbitrage is much more complicated than just reducing latency of the SIP.  

The latency of the SIP refers to the time that it takes to receive all information, compile it, aggregate and assemble, and then disburse it out.  The problem with the SIP is not the time that it takes to perform these functions, even if and when that time is improved, it is with the method of disbursement.  The transportation of that data is where the challenges lie.  

Because there are various sources for the SIP information, there will always be a latency issue involved with consolidated data compared to direct data feeds.  The NYSE, for example, has their data center in Mahwah, NJ, while information from Nasdaq comes from their server center in Cartaret, NJ.  Just a different geographic location can make enough of a speed difference to create a latency issue.

Flash Trading by Co-Location

Flash Trading occurs when exchanges ‘flash’ buy and sell order information to premium subscribers, typically HFT firms, a fraction of a second before publicly available.  This is quite controversial as HFT firms use this advantageous information to trade ahead of orders pending, also known as front-running.

This has been going on for quite some time, as in 2009 Senator Charles Schumer asked the SEC to ban flash trading altogether, stating that it contributed to a two-tiered market, the privileged few and everyone else.  

While other factors such as technology, computer algorithms for trading, distance, and regulations, contribute to latency issues, these two main factors of consolidated feed vs. direct feed and flash trading, both lay the framework for the current two-tiered market structure.  

Latency Arbitrage and NBBO

What all of these terms boil down to is that co-location, HFT, flash trading, and SIP latency all contribute to an altered NBBO.  The NBBO, (National Best Bid Offer), is supposed to be the one most accurate price of a stock across the market, all exchanges and off exchange venues. A representation of the “best price.” The problem with this endeavor is that it is impossible to instantaneously update every participant in the market, even the exchanges, when a change occurs to the NBBO.  

Information about stock price changes must travel between all market participants and the speed at which that occurs varies greatly depending upon the technology a firm is using, and the distance between them.  This means that all market participants, including the exchanges themselves, see a different view of the NBBO at exactly the same moment in time.  In fact, information leakage is not the only downfall of latency arbitrage.  When these privileged firms conduct their privileged trades, the NBBO is actually changed; altered by the front-running executions.  

While there is no quick solution to this problem, and perhaps not even a long drawn out solution, there are some tools available now that can offer some assistance to traders dealing with this problem.  If we cannot stop the latency arbitrage as we know it, we can tackle it another way – with modern day latency arbitrage tactics.

One tool addressing this issue is the IEX Signal used in their D-Peg and Primary Peg orders.  The signal acts as a yellow traffic signal, alerting a trader to an upcoming change to the NBBO.  Used as a predictive tool, the IEX signal utilized in these proprietary orders is also termed a ‘crumbling quote indicator’, as it predicts an upcoming change to an NBBO, basically by observing a stock’s NBBO activity, any shifts in that price, and compiling a prediction of which way it’s leaning, or moving.  IEX’s signal attempts to predict the NBBO change, and will move their D-Peg orders out of the way if need be, protecting the investor.

This is one small tool in a large fight against a long-time practice, a practice that the privileged participants are not giving up easily.  With a computer algorithm for trading conducting trades at lightning speed, we should all realize that we are now at a crucial point in terms of the structure of our financial markets.  While latency arbitrage must be tolerated for the most part, the time to take action whenever possible is long overdue.

Great Point Capital has been serving the trading community since 2001 and our 100+ prop traders actively trade the firm’s capital, specializing in equities and equity options.  We are headquartered in Chicago with a location in Austin, TX.  Contact Great Point Capital LLC today, in either our Chicago Office, or our Austin Office, to learn more about how we can successfully trade together with high performance results.  We are one of the very few firms able to offer access to Takion Software Platform, enhancing your online equity trading performance.

Comments are closed