How To Avoid Getting Whipsawed By High Frequency Trading After Data Releases

by: Steve Hammer

On November 29, 2011, at 10:00 a.m., the Consumer Confidence number was released. Everyone understands that economic releases move markets. Long before high frequency trading (HFT), markets have always responded with increases in volatility associated with the level of importance of the particular news release. Among all such releases, consumer confidence would probably rank somewhere a bit above average.

The extent to which HFT systems can influence prices is wonderfully illustrated by the November 29th event. It’s a chilling example of how much the trading landscape has changed and sends a warning to those trying to trade during these events.

Shown below are two algorithmic detection graphic monitors. The upper graph identifies the number of stocks trading with detected algorithmic sequences in them. In other words, the number of stocks involved in a program trade per second. The large spike at 10:00 a.m. shows 450 individual stocks in program trades at the instant the confidence number was released.

HFT Alert HFT/Algo Monitor

Again in the image above, the yellow line identifies the number of stocks being quoted per second. We can see that the initial HFT burst at 10 a.m. included 800 individual issues. Quote rates, that are orders, are what HFT systems use to slow the tape by flooding the CQS system with traffic mostly consisting of expired quotes. It is commonplace to see quote rates on individual stocks exceed 5,000 per second (the record I’ve seen is 42,000 in 1 second). This means that of those 5,000 quotes, possibly one might not be expired before it got to your screen.

This is called quote stuffing or financial spam. During the Flash Crash in 2010, quote levels rose to 650 stocks which was enough to trigger the worst part of the decline. This is because as the tape fell further behind, no one could trust what they were seeing on their price screens. So during a period of market stress, just as we need the liquidity provided by individuals and institutions, investors step back and the machines sell into the void.

The chart above shows SPY during November 29th. The arrow points to the large expansion bar at 10 a.m., which was the release of the consumer confidence number. As the quote rates rose to 800 stocks, the tape was slowed enough for HFT systems to run the price of SPY substantially. The high and low for that minute shown on the price bar was 120.84 to 120.13, or a little more than a half percent.

What is astonishing is that those HFT systems caused that range to be made in the very first second of that minute. The bar opened at 120.21 surged to 120.84, then collapsed to 120.13 before closing that first one-second period of that minute at 120.44 on 495,767 shares representing 965 individual trades.

The table below shows this information. The highlighted blue line is 10:00:00 or the first second of that minute. The prices are shown open/high/low/close/volume and trades. You can scan down the list and I am sure you will be astonished at the ranges between the high and low for each additional second. It surges again at 10:03-10:04 which is associated with the additional spikes shown on the HFT alert graphics above.

Click to enlarge

List of 1 second bars on November 29, 2011 near 10am

So what’s the lesson here? Lesson No. 1: Don’t try to trade or have any limit orders anywhere close to the current price before the news release. If you do, your risk protection in the form of a stop loss or other limit order may actually end up causing more risk to you than you are seeking to avoid. Lesson No. 2, if you are going to trade, it’s best to be able to monitor for algorithms so you can be informed as to the level these systems are operating. Lesson No. 3: As I have seen dozens of times, most economic news releases display this type of pattern.

The opportunity was in monitoring for HFT/Algo activity levels knowing that once the machines were finished whipping price up and down, prices would reflect the fact that the consumer confidence number came in above expectations, hence there was a reason for prices to rise. In this case one had the opportunity to enter long at the level SPY was prior to the release of the number.

Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.