Aurora Borealis
1 November 2015 CFA Institute Journal Review

High Frequency Market Microstructure (Digest Summary)

  1. Priyanka Shukla, CFA

US markets have changed considerably in the past 10 years. High-frequency (HF) trading now accounts for more than half of US trading. Trading is automated and faster, and market microstructure, orders, price discovery, and liquidity are fundamentally different. The author describes HF microstructure changes and HF trading, strategies, traders, and markets.

What’s Inside?

Markets have been transformed in far-reaching ways. Market fragmentation, co-location, and direct feeds as well as technological advances have contributed to increased complexity in various ways:

  • Liquidity dispersal across trading venues
  • Higher order volume, with most orders getting canceled
  • Price differences across venues, leading to arbitrage opportunities within and across markets
  • Competition and speed increases for all investors

The author examines high-frequency (HF) strategies and traders; HF market making; non-HF traders; and algorithmic trading, markets and exchanges, and policy issues. She also outlines gaps that exist in the research and data and suggests research areas to address these gaps.

How Is This Research Useful to Practitioners?

Understanding HF market microstructure may be valuable to investors, broker/dealers, portfolio managers, economists, policymakers, and regulators. The following summarize a number of the author’s observations.

HF trading categorization ranges from low latency (approximately 10–3 seconds) to ultra-low (approximately 10–9 seconds), with co-location and individual data feed options helping minimize latency, thus allowing HF traders to enter and cancel orders faster than others. More than 98% of orders are canceled.

HF strategies take advantage of exchange priority and matching rules, such as price, size, or time priority or pro-rata matching. Statistical arbitrage strategies take advantage of historical correlation, whereas other algorithmic trading strategies include profiting from time-weighted average pricing patterns, latency arbitrage, and momentum-ignition strategies. Manipulative and prohibited strategies include spoofing or quote stuffing by predatory algorithms.

HF market making increases informational efficiency and reduces spreads. But HF market makers have no obligation to provide continuous liquidity, an aspect that is often criticized.

Algorithms offered by broker/dealers include dark aggregation (nondisplayed orders for institutional investors), scheduled orders (parent orders split into child orders), volume participation, active trading (to minimize implementation shortfall), and smart algorithms. Retail trades are often internalized or delivered through broker/dealer purchase agreements. Dark trading and odd lot trades have increased, whereas trade sizes have fallen because smaller trades are used to prevent HF traders from spotting and profiting from human traders who tend to trade in round lots.

Exchanges use pricing structures, such as maker-taker pricing (market order traders pay fees, whereas limit order traders receive rebates, which is a profit source for HF traders), taker-maker pricing, and subscription markets. Different order types are rampant.

The author states that the US SEC’s naked access rule is well thought out, although the current trade-through regulations can be improved. She touches on nuanced fairness issues and suggests further policy research areas.

How Did the Author Conduct This Research?

The author’s discussion of market microstructure includes information about market structure and design, trade rules, market frictions, price discovery, information availability, and liquidity. Regulations to advance competition, such as Regulation National Market System, Regulation Alternative Trading Systems, and similar European regulations, helped grow HF trading and contributed to equity market fragmentation.

For researchers, HF trading has changed the nature of market information, liquidity, informational asymmetry, adverse selection, and transactional time. In algorithmic trading, the order sequence holds information so trades are not independent. This dependency, coupled with maker-taker pricing and midpoint orders, makes identifying informed traders difficult. Studying execution data from a broker/dealer, the author finds that, on average, 65% of trades were passive (buy at or below bid), and 13% were aggressive (buy at or above offer).

She pinpoints such issues as missing odd lot data, consolidated tape data out of chronological order (because of varying latencies and time stamping), quote volatility, and the possibility of using volume clocks and emphasizes the need for new analysis tools to handle empirical challenges.

Abstractor’s Viewpoint

HF strategies can be protean, and the author presents a useful overview of current HF strategies, market microstructure, and topics meriting further policy or research focus. Topics not covered include market transparency; costs, benefits, or options in the race to minimize latency; and HF trading’s broader economic impact.

We’re using cookies, but you can turn them off in Privacy Settings.  Otherwise, you are agreeing to our use of cookies.  Accepting cookies does not mean that we are collecting personal data. Learn more in our Privacy Policy.