What is high-frequency trading? It's certainly in the news a lot lately, what with the publication of Flash Boys, essentially an indictment of the industry, and recent scandals involving HFTs and dark pools. Let's take a look at what HFTs do, the trajectory of the industry, and where problems first arose.
What Do HFTs Do?
There are a lot of things that high-frequency, or high-speed, traders do, but the majority of them act as market makers.
What does this mean? It means that HFTs act as a middleman between buyers and sellers, buying and selling from their own accounts to fulfill trades that other people want to make. They compete by offering faster execution at a lower price, and they take a spread in order to make money on the service.
The idea was innovative because, like most things, market-making used to be performed by people -- that image you have of someone standing on the floor of the NYSE is not inaccurate.
HFTs proved to be powerful competition because they used algorithms, not heuristics, to see market movements more quickly than their human counterparts, and thus could respond faster and more accurately. With heuristics replaced by statistics, HFTs could meet liquidity demands more effectively and at a lower cost, meaning that bid-ask spreads could be reduced, and transaction costs with them.
So do HFTs actually perform this function well?
Research indicates that the introduction of high-frequency trading is followed by improvements in price discovery (meaning prices are more accurate), increases in liquidity, and reductions in the bid-ask spread. Take it away, or tax it, and spreads go up and liquidity goes down.
So, in a word, yes -- there is evidence that HFTs perform a beneficial service.
How is All That Working Out?
The problem with competing in this way is that the barriers to entry are relatively low, meaning that anyone with the technical smarts to do what you do can come in and compete alongside you.
Despite major investments by existing firms, that's exactly what's happened in the HFT world: As more firms came in to take the low-hanging fruit, there was less fruit and the exercise became less profitable.
Both revenues and margins are shrinking for HFT firms -- industry revenue went from $5 billion for in 2009 to around $1 billion in 2013, and average profits from 1/10 of a penny to 1/20 of a penny.
In other words, HFTs are making the markets so efficient that they're starting to put themselves out of work.
So What's the Problem?
Like any business trying to survive (you could argue this chronology if this if you wanted, I haven't see any evidence either way), HFTs started looking for other ways to wrest revenue from their activities.
One tactic is "momentum trading." Using the same principles that help them buy and sell efficiently, HFTs try to pick up patterns in where a stock is going and place big bets on the direction of that stock. Rather than trading to transact, they start buying and selling to win -- a potentially lucrative and rather risky strategy.
The tactic illustrates the primacy of information for HFTs. These firms are already focused on it, if and the advent of directional makes up-to-date intelligence on individual stocks even more critical.
So, in addition to analyzing order data, we now have vociferous news analysis. Not long ago, there was even a bit of a scandal about a hacked Associated Press Twitter account.
Of course, some of the most prized information has nothing to do with what's happening at a company, but rather about what's happening with the people who trade shares of that company.
Knowing what institutional traders are doing would obviously be very valuable to a firm that competes on information. And indeed, HFTs have followed beleaguered institutional traders into their dark pools. A major recent lawsuit alleges that a Barclays-operated dark pool gave certain HFT firms special treatment by providing access to private trading data. Lovely.
These are the unsavory sides to HFT, and they not only illustrate the perversely exaggerated incentives facing these firms but also some deep-seated conflicts of interest between brokers, exchanges, traders, and market makers.
It's these issues that are at the root of governmental investigations into HFT: Where is there potential for conflict (or where does it exist already), and how can it be stamped out and avoided in the future?
Other Problem: Rise of the Machines?
Another potential problem is the reliance on quantitative methods in themselves.
The Flash Crash was a great example of algorithms gone wild, and it's not the only example of technical glitches causing high-profile problems. These issues have plagued everyone from Knight (a big HFT firm), which nearly went bankrupt due to a software problem, to NASDAQ, which was forced to delay the IPO of Facebook because of a technical error. It's been suggested that redundancies like stop limits, trading halts, and kill switches would help mitigate this problem, not just for HFTs but for everyone.
Finally, there is the problem of HFTs competing with themselves to the detriment of everyone else. There is some evidence that when HFTs start aggressively competing for trades liquidity drops and volatility rises.
This is something to at least think about -- what are the implications of HFTs running in circles around each other, and how can that be avoided?
These are still open questions.