PORTFOLIO POINT: High frequency trading is poorly understood by many investors, but the reality is that HFT affects everyone in the market.
High frequency trading (HFT) is subject to many different interpretations, from those who are practitioners of these techniques through to those who have heard the term, but do not think they are really impacted by it. This article is aimed at the latter category, and hopes to identify the fact that these investors are, in fact, participants in HFT – but on the wrong (losing) side.
As has been well-documented in recent years, HFT is a particular form of algorithmic trading, or trading that is conducted by pre-built mathematical systems. The 'frequency’ component refers to the fact that these systems execute very frequently, often holding investment positions for very short periods of time (often sub-second).
Non-HFT participants are sometimes ambivalent about the process, or believe it is immaterial to them. I disagree.
You may ask, “How can high-speed trading cost me, if I am not in the race myself?” How does the fact that Formula One cars drive faster than mine, impact on me? Obviously it doesn’t, unless Formula One cars are allowed to drive on the general road network with no speed limits in place, which is not the case. However, this is the case in electronic trading venues; there is no separation between low and high-speed users. And here is where the problem begins.
Think of a world in which you visit a supermarket and browse through a number of fairly priced products, but which are continually fluctuating in price in real time, depending on the supply/demand in the marketplace. Instead of prices changing daily, they can potentially change faster than the human eye can observe. Imagine algorithms that are constantly evaluating weather data, to see if the probability of a cyclone has just increased based on the last satellite image, and whether this should have an immediate impact on fresh fruit. This could be considered a fair and justified use of technology to ensure that commodities are priced correctly for both buyers and sellers.
Now imagine an algorithm that performs surveillance on customers, and observes customers buying certain items. This algorithm, seeing the intention of a customer to buy something at $1, quickly jumps in to buy the item at $1, driving the price up to $1.01 and then selling the item to the consumer at this higher price. To the consumer, this is not a big deal since prices are continually changing anyway, due to the fact that satellite images and other supply/demand data is causing prices to fluctuate rapidly. For the consumer, this 1c movement would just fall into the 'noise’ created by a normal real-time market. No matter how quickly a consumer tries to buy something, they are always beaten by the algorithm. This consumer may not realise they are paying out (over time) money to this algorithm, with no benefit to them. This is unfair, and not something that should be supported by a free market.
This is exactly what is happening today in the US stockmarket, due to a fragmented market structure with many different trading venues and regulations which, while introduced in the guise of increasing fairness, actually create opportunities for algorithms.
It should be noted that this scenario does not apply in Australia – yet. For this HFT advantage to be enjoyed, high-speed players need to have advance knowledge of order placement. However, it is actually the regulations designed in the guise of promoting fairness (US SEC rules 610, 611) that create this opportunity. The SEC created what on the surface seems to be quite a sensible rule: exchanges can’t execute orders at prices that they know are inferior to prices offered on another exchange at that point in time. Sounds reasonable right? This should protect investors from being short-changed by buying a stock for $10, when it could be bought for $9.90 somewhere else. The problem is that in the high-speed world of algorithms, everything you can see is already out-of-date. Due to the “slow” speed of light, everything you know about prices on other exchanges is already out-of-date by the time it arrives.
Imagine this rule being enforced on multiple execution venues in the days when carrier-pigeon was the fastest available communication technology. That is the current scenario under which algorithms operate – communication is extremely slow relative to processing speed, and information is always out of date. To “work around” these rules, exchanges introduced “flash orders”: orders generated on the exchange, only given to high-speed traders to allow them to knowingly trade on a price that the regulations consider to be inferior to prices available elsewhere. These high-speed traders are usually more informed (via their technology) than the exchanges, and can tell the difference between a good and a bad deal here. These orders only last for 0.1 of a second, and if no high-speed trader responds in that time, the orders are re-routed to other exchanges where it is believed a better price exists.
Regardless of the outcome, the exchange has essentially given a 'heads up’ to the more informed high-speed community that someone is just about to buy a certain stock at another exchange. This then gives them the opportunity to manipulate the market price and make 1c in the process, similar to the supermarket thought experiment. While this strategy involves some risk, in the long run the odds are in the favour of the 'house’, or in this case the one with the fastest technology.
The moral to the story is that regulations designed to help you, can actually hurt you. This is particularly pertinent in Australia as we now have multiple execution venues, with the associated “NBBO” rules, and ASIC is considering how to regulate algorithmic trading. Will we make the same mistakes as the US markets in actually creating opportunities for the high-technology players to feed off the rest of the market? Whatever you do, do not see the US as model case.
Many have responded to these risks with ideas of regulation in the form of speed limits, taxes and other brakes on the fastest players (read: most successful). I am against all these approaches. I believe there is a much simpler solution to this problem, and it is free-market based.
The problem we have comes back to the original fundamental rules of electronic trading: first in, wins. If you analyse it, you begin to see that the market structure we’ve ended up with (fast = profit, slow = loss) is actually a function of the market rules, not an inevitable outcome.
What I am advocating is this: instead of operating a simple limit-order book (LOB) that works on two factors – your limit price and the time you entered the order (known as “price-time priority”) – you also allow orders to include a time period. That is, instead of the current $10 sell-order meaning “please sell my stock of XYZ at $10 to the first person who enters a $10 or higher bid to buy it”, you instead enter an order with a set time period. For example, you could say sell XYZ at $10 for 0.5 seconds, meaning: “Once someone bids at least $10 for my stock, please operate an auction for 0.5 seconds, selling to the highest bidder at the end of that period”.
In this scenario, once the high-speed trader bids $10 in order to pick up a “flash bargain”, there would be a 0.5 second auction period. Of course, all market participants would be able to see the time period associated with any order, and be able to decide whether they want to bid based on that. In fact, the bidder would need to provide a time period as well, and the auction would only be initiated if the time period on the bid was at least as long as the offer.
What this allows for is the market to be 'slowed’ – but by the market itself, not by regulation. High-speed markets are only advantageous to those who can move the fastest. The slower majority, by definition, are placed at a disadvantage. Liquidity providers should be allowed to operate at the speed at which they are best served, not forced to run the gauntlet of an endless technology war.
Fil Mackay lives in the world of high-speed computing and trading systems. He has more than 20 years’ experience in building software, and combines that with equivalent experience in financial markets. He is also active on Twitter (@filmackay).