I’ve seen plenty of articles lately about high frequency trading, most of which completely exaggerate the effects of high-frequency trading and algorithmic trading in general.
Let’s talk about HFT, what it is, and how it actually works.
What is High Frequency Trading?
In short, HFT is trading securities, usually ETFs, really fast. In most cases, high frequency trading firms are interested in arbitrage. Arbitrage being a statically risk-free, or near risk-free investment usually the result of pricing discrepancies between one market and other.
As an example, we’ll use the Gold SPDR ETF (GLD), which tracks the price of roughly 1/10th of an ounce of gold bullion. Let’s say the price of the ETF is $190 per share. Gold is $2000 per ounce. How can you make money with this trade?
Easy. Buy the ETF, and sell an equal amount of gold on the futures market. If you buy 1 ounce of gold through GLD for $1900, and sell short 1 ounce of gold for $2000 on the futures market, then you’ve essentially harvested $100. Eventually, buying interest will push up the ETF, and push down the price of gold, until equilibrium is reached. Risk-free profits!
Okay, so that’s the basic explanation for high frequency trading as it relates to arbitrage. MOST volume from HFT firms is arbitrage, which is a good thing. Arbitrage creates more efficient markets, and arbitragers should be welcomed into any market.
Let’s say I want to bet on rising hamburger consumption. There’s not a really good way to do this. I could go buy a bunch of fast food then let it sit in my closet until demand (and hopefully prices prices) go up. But who is going to buy a dusty, month old hamburger? This isn’t very practical.
I could create my own synthetic option, though, by buying one security and hedging out all the parts I don’t like.
For example, I might buy 1000 shares of McDonalds, then short 20 shares of a smoothie maker to hedge out the impact of smoothies, short a French fry producer to isolate out French fries, and buy a derivative from Goldman Sachs as a bet on the average fast food employee hourly wage. Tada! I created a synthetic option, and pending that the Ph.Ds. and NASA astronauts that crunch numbers for me do it right, then I have a portfolio which will change in price due only to a change in consumption for hamburgers.
This is hardly complex by high frequency and algorithmic trading standards. I partially covered this in a post about how Wall Street brain drain actually isn’t brain drain.
High frequency trading firms also participate in market making, which tends to be controversial. Some say that a high frequency firm can “front-run” the market by buying a security at one price and selling at another before a trader can enter the market themselves.
A HFT firm might buy Apple (AAPL) at $360.02 and sell it immediately to another trader for $360.03, effectively bringing together buyer and seller. Nowadays this is supposedly a crime. Just ten years ago humans, not computers, did this work and the bid/ask spreads were far greater than a penny per share. I guess it’s all relative. I’d rather have a computer doing the processing than a slow and costly human being.
Interestingly, no one complained when Wall Street used fractional quotations, which made brokers about as much money on each trade. One-sixteenth of a dollar was often rounded up to $.07, meaning investors got short-changed by three-fourths of a cent on each share they traded anyway. Now we get faster execution and pay less in spreads.
Who’da thunk trades would take less time to complete, cost less in transaction fees and yet we’d still find a way to complain about it?
Ticker Tape Trading
There’s a ton of insight that comes from who is buying and selling. This is why even individual investors look to see insider buying and selling activity.
Complex computer algorithms use the order book to “see” into the market before others can. Think about it. If Morgan Stanley just dumped 1 million shares of Sears Holding Corp. then maybe there’s a downgrade coming down the pipeline. A high frequency firm might hop on the momentum, using data “learned” (more accurately, guessed) from the available ticker information.
This borderlines on privileged information. But, again, the benefit is due to speed, not information asymmetry. You could do the same thing, you’d just have to pump billions of dollars into some serious fiber optic pipes. 😉
Events and Behavioral Finance
Most high frequency trading models don’t affect individual investors in a measurable way. This one does…no doubt about it.
Behavioral finance is a combination of economics and finance. In particular, it is a study about how we feel about things corresponds with the financial markets. For example, if investors start worrying about the future for the economy, they sell off small cap stocks. Small caps are included in the “risk-trade.” When investors are confident, they buy small caps. When they aren’t, they sell them.
So if the non-farm payroll report misses the mark on the first Friday of the month, investors go to sell off small caps. The NFP is released to everyone at the same time, but access to the market isn’t at a single speed. If you can get in on the trade first, then you’re set to profit on the following momentum.
Real Danger of High Frequency Trading
There is a very real danger with high frequency trading, and that is that absolutely none of the data used to make a trade as part of an algorithm is related to actual business valuation.
In the example of events based trading, a selloff of small caps may push down small cap values big time. Many of the companies that make up the Russell 2000 index of the smallest companies on the market trade fewer than 50,000 shares per day. As you can imagine, a ton of activity on the short-side of the market can push securities down due to an event that wasn’t at all related to the company.
But it’s what happens when HFT firms come into the market to buy and sell quickly on changes in the market’s mood. The Russell 2000 index includes 2000 stocks; you don’t really think the HFT firm cares to figure out which is worth owning and which isn’t, do you? They know they don’t want small caps! Sell the lot of ‘em!
Same thing with commodities; they’re part of the risk-trade. Sell ‘em off in mass on a loss of confidence! Buy the whole consumer staples sector for protection! Rawr!
Anyway, the moral of the story is the HFT creates both more efficient markets and less efficient markets. It also creates volatility. However, banning it, taxing it, and protesting about it won’t do all that much.