Nanex hft

By Eric St-Cyr

Do you remember HAL 9000 from “2001 A Space Odyssey”? It was a mega computer on a spaceship that was prepared to do anything to conclude its mission. Arthur C. Clarke, the author of the book, was 10 years too early however. Since 2010, his vision has taken shape under another name: the HFT.

On Feb. 13, the New York Stock Exchange experienced its lowest volume for non-holiday trade in over a decade. If we consider that High-Frequency Trading platforms (HFT) now account for 60%-75% of daily volume, we can easily conclude that HFT is the only game in town. The volume of transactions today is 40%-50% lower than it was three years ago.

Investopedia describes HFT as: "A program-trading platform that uses powerful computers to transact a large number of orders at very fast speeds. High-frequency trading uses complex algorithms to analyze multiple markets and execute orders based on market conditions."

HFT presents several distinguishing features:

1) It is a quantitative system

2) The trades are kept for a very short period of time

3) All trades are closed before the end of the day

Today the bulk of the market volume is taking place between the largest brokerage houses in the US, buying and selling to each other for a fraction of a cent profit on every trade. No wonder the market goes up on low-volume days and drops on high-volume ones. The phenomenon became easy to observe in the summer of 2010 and has reemerged even stronger since August 2011 (strangely, on the day of the U.S. downgrade...). There is emerging evidence that HFT was the catalyst behind the May 6, 2010, flash crash. (How can they not be if they account for the most of the volume?)

If you doubt the influence of HFT on today's market, please visit Nanex , an expert in market data feed. Their visual demonstration illustrating all the trading days since 2007, named "The Rise of the HFT Machines" is stupefying. It clearly demonstrates the rising impact on the market, over the last five years, of computer-to-computer trading. This five-minute presentation should be watched by anyone trying to understand today's market behavior. The presentation is available at: http://www.nanex.net/aqck/2804.HTML .

Forget about fundamental or technical analysis. There is a new game in town. Like HAL 9000, HFT programs will do everything in their power to conclude their mission, generate millions in profits and push the market higher.

Over the past few months I have noted some observations on the impact of the HFT on the market. I don't have the data to back all of them, so some are just opinion, but if you follow the market daily, you will probably agree with them.

1) The HFT impact on the market is growing. Brokerage firms in Brazil and Malaysia are now installing HFT platforms in these markets. It is becoming a global phenomenon.

2) The efficiency of the algorithm employed seems to be at its maximum when volumes are extremely low, like we experienced in the summer of 2010 or since the beginning of 2012. This would explain why some of the largest brokerage firms experienced no trading loss for any days during some of the quarters of 2010 and early 2011. Historically, this would have been a mathematical aberration.

3) The algorithms are using aggressive stop-losses, and most firms seem to have a similar mathematical model. As the markets correct, they all rapidly go into selling mode, all at the same time. May 6 2010 comes to mind, however the meteoric rise of AAPL in February of this year, followed by the 5% intraday drop on Feb. 15 is also a sign that all HFTs hang on the same side of the boat.

4) HFT increases liquidity however it also increases volatility.

As an investor you can participate in this foolish game, but be aware of the danger. When the ice cracks, everyone ends up in the lake, and I can guarantee you that HFT will swim faster than you. This means that without a stop-loss, at least 75% of the market would have sold their positions before you realize what is happening.

My recommendation is to buy large-cap stocks with high liquidity. I would favor corporations with low downside volatility and a strong brand, so buyers can jump back in on a pullback.

Up to last week, the obvious choice was Apple /quotes/zigman/68270/quotes/nls/aapl AAPL -0.02% (see why I like AAPL here ). However, I now recommend you wait a few days to see if the stock stabilizes. Google /quotes/zigman/93888/quotes/nls/goog GOOG +0.88% , Whole Food /quotes/zigman/80039/quotes/nls/wfm WFM +11.32% or Coca-Cola /quotes/zigman/222647/quotes/nls/ko KO +2.38% would be good picks.

If you want to reduce the risk, go with the SPDR S&P 500 ETF Trust /quotes/zigman/714403/quotes/nls/spy SPY +1.65% . After you have established the position, set a 5% trailing stop-loss or manually increase your stop-loss on a regular basis. I recently moved my AAPL stop-loss from $390 to $430 to $450 to $470 and to $490 on Wednesday. I would also recommend some protection with a short EURUSD (see previous column ) and some long VIX /quotes/zigman/2766221 VIX -9.36% call options or, if you don't have access to options, the Barclays Bank PLC iPath S&P 500 VIX Short-Term Futures ETN /quotes/zigman/1312200/quotes/nls/vxx VXX -0.67% It isn't as good as the VIX, but should do the job.

For as long as volumes remain low and the algorithms work on the long side of the equation, we're good. Europe will present us with a big short opportunity in the coming weeks, however a market on steroids can remain exuberant for a long period of time. The stop-loss should protect your portfolio on the downside, and the EURUSD and VIX call option turn a profit when HFT turn the trade and move onto the short-side of the fence.

Strangely, HFT now does to the stock market what the Fed does to the economy, support it at any cost; and this is why my friend you must buy on the dip...

Clover is currently long AAPL, GOOG and VXX in some of its client's accounts. Clover has no positions in WFM, KO and SPY. Clover has not made any transactions in any security mentioned in the past 72 hours and will not make any purchases in the 72 hours following publication of this article.