Financial Management
MICHAEL HARRISON | November 16, 2021
High Frequency Traders (HFT) play a ‘need for speed’ game day after day in a quest to capture smaller and smaller profits. To be a successful high frequency trader there are two mountains to summit.
The first, is the development of a system which can analyse data in as close to real time as possible. Here, this would involve not only monitoring prices, volatility, clustering; on multiple exchanges simultaneously. There is also the need to monitor slower items such as fundamentals, news streams and annual dividend pay-outs. It is likely that many high frequency traders are using artificial intelligence to do this, as they must process big data sets in as close to real time as possible. Thus, this need for speed has started an ‘arms race’ between traders to possess the fastest system in the market.
This links to the second of the two challenges: the minimisation of latency, or in plainer terms; the maximisation of speed. Information must be processed as quickly as possible, a trading strategy devised and a market order submitted. Ideally, quicker than any other trader. To be clear, there is always a first movers’ advantage when trading information into price!
To give a sense of this speed, consider that all of the above can be achieved in less than the time required for a human being to blink. This technology originates in the digitisation of exchanges, for example the NASDAQ exchange opened in 1971 as the first electronic exchange. Later in the 1980’s, proprietary information services became available. The most well-known of these became Bloomberg which began in 1981. The faster trading conditions and greater availability of information has opened up the gap between human traders and algorithmic traders, which has become wider to a point where their ability to interact with each other is questionable.
Estimates vary; however, one reliable source suggest that 73% of the US equity trading is now undertaken by HFT’s . In my own research in ForEx markets for highly traded pairs, the HFT incidence can be up to 8%, even in an over-the-counter market.
To illustrate an issue HFT activity creates, consider that a high frequency trader can observe information, identify which asset is involved and the new ‘correct’ price, and submit an order; all in the time needed for a human trader to blink. This mismatch is especially problematic where liquidity is provided to markets by high frequency traders. Here, it is very difficult for slower human traders to access liquidity as it does not remain in the market long enough as HFT matches against HFT.
In their defence, a high frequency trader would tell you that they provide liquidity for rare assets and also work to keep bid ask spreads tight. Also, it is not debateable that HFT does make a significant contribution to keeping prices correct, in terms of reflecting all known information and holding the law of one price across trading venues. So, it can not be as simple as concluding that HFT is all bad and progress should be revered.
At this point we need to know that a HFT is typically an inventory neutral trader who will trade to capture spreads. Typically, individual trades earn very little return. Where the spread capture takes place over short time periods; this is known as a ‘scalping’ strategy. A trader who can work at a sufficiently low latency, will be able to run a scalping strategy over six ticks of the market which in some cases, can occur in less than one second.
High Frequency Traders in their efforts to capture spreads could also increase price volatility. This may especially be the case where a trader, is placing a large order which incentivises HF traders to try and provide liquidity whilst capturing a spread. As this increases the price impact of the large order, there is an incentive to find a way to avoid the HF trader’s attentions. Such places exist away from the regulated exchanges; these off exchange liquidity sources are known in the trade as ‘dark pools’. These remain legal for trades above a certain size.
A more pertinent issue in terms of the fragmentation of markets, is the creation of trading venues which seek to protect slower traders from predatory high frequency traders. Many regulatory tools have been proposed to constrain HFT activity. National regulators have over the last decade developed governance standards for algorithms used in trading, to incentivise developers to have full awareness of the behaviours of their creations under the full range of market conditions.
On balance, I am not in favour of solutions involving lags or generally slowing down trading, as the race to the front of the queue still exists. Rather, any intervention needs to alter the desire/incentive for some activity to take place. For this reason, minimum order resting times, perhaps as little as one whole second would remove the incentive for aggressive scalping strategies.
Read More