Since the controversial acceptance of IEX as a stock exchange in 2016, high-frequency trading (HFT) has almost completely fallen off the pages of financial newspapers and replaced by fintech as the hot topic. In the background, there has been a slowdown in the world of HFT with lower profitability causing consolidation in the industry. Although some commentators argue this slowdown is a temporary outcome of very low market volatility, others argue that the trading edge obtained by some high-frequency traders (HFTs) has been eroded by competition. According to Tabb Group, US market makers reported $1.1 billion in revenue in 2016, compared with $7.2 billion in 2009.

The increased costs of installing communications technologies to reduce latencies, which is the time taken for an order to reach a trading venue, has hurt profitability for certain players and encouraged others to collaborate. Furthermore, these network latencies are approaching fundamental physical speed limits, and thus not leaving much scope for improvement.

Rediscovering Trading Ideas

Increasingly, HFTs are saying that speed is no longer sufficient for their business model and are attempting to find a competitive advantage in predicting markets through quantitative models and artificial intelligence (AI). This feeds into a developing narrative of the “Quant comeback” or Quant 2.0. Quantitative funds have been relatively out of favor since the financial crisis, but there is a renewed hype that machine learning–enabled AI will be able to develop “fundamental” trading ideas for individual stocks, rather than simply taking advantage of arbitrage opportunities as previous quant strategies did.

Although some may welcome this return to value-adding activities, an article in the Financial Times raises an interesting point. Who will be on the losing side of this AI-enabled quant resurgence?

The Game Changes, the Loser Stays the Same?

Paul Woolley and Dimitri Vayanos argue that asset managers who are tightly judged against market and peer benchmarks end up chasing those benchmarks to some extent — resulting in, for example, overpaying for stocks they do not like and selling stocks they do. On aggregate, this behavior is somewhat predictable and easy prey for early-stage momentum buyers, a common Quant 2.0 strategy, and one made even more effective by machine learning–enabled AI.

The authors present a stylized worldview in which benchmarking corrupts the price mechanism, exacerbated by Quant 2.0 early-momentum investors that buy early in order to sell to “benchmarkers” later in the cycle. Concurrently, these quant investors opportunistically buy the “artificially” undervalued stocks. The authors argue this stylized representation can explain why momentum and value investing seem to outperform other strategies. Essentially, “benchmarkers,” or late momentum traders, end up buying high from and selling low to early-momentum quant traders.

According to the authors, this is a problem if, for example, a pension fund has investments in a quant manager as well as the typical investment in a benchmark-tracking fund. Pension fund investors would essentially be paying fees to both managers even though one strategy (the quant early momentum) is designed to profit at the expense of the other (the benchmark tracker).

There is concern that AI-enabled strategies may exacerbate the problem by devising more intelligent and profitable ways to exploit the weaknesses of the system.

HFT: The Future of Fintech?

Is a fear of AI quant just the latest in a long line of technological doom-mongering? HFTs were once seen as profit-generating machines and destructive participants in the market ecosystem. But now it seems competition has tamed and commodified the HFT product. This shift has partly been caused by competitive pressure, and partly by the fact that the market is reflexive. In this case, reflexivity refers to the phenomenon of market participants (e.g., hedge funds or HFT) losing their competitive advantage because their presence and actions in the market change market behaviors as other participants adjust (find out more in the CFA Society Switzerland video presentation about adaptive markets).

With HFT firms now turning an eye toward machine learning–enabled AI, what lessons can be drawn for fintech firms? Izabella Kaminska argues that fintech growth (for example, Quant 2.0 AI-enabled traders) may hit a brick wall, just as HFT growth has done, once these firms’ competitive advantage is eroded or the financial market itself adjusts to their existence. In particular, market reflexivity may be the biggest stumbling block for the roll out and long-term success of AI-based investment strategies.

It is likely that fintech, like Quant 2.0, is at the stage HFT was when the speed race could still be won. It is possible that more blockchain-based decentralization or more machine learning can give firms a competitive edge — for the moment. But the experience with HFT suggests that eventually the technology will diffuse through the market and fintech firms will face the difficulties that HFTs are now experiencing.

These arguments suggest that technology should not be a source of fear for investors. They still face the most technologically complex yet favourable trading conditions ever known in terms of low explicit trading costs, easy access to markets, and the availability of innovative products and strategies. Instead, investors should be vigilant to more mundane and traditional sources of disadvantage, such as excessive fees and misselling or having their assets used to fund the kinds of zero-sum games described by Woolley and Vayanos.

If you liked this post, consider subscribing to Market Integrity Insights.

Photo Credit: ©Getty Images/alexsl