Ending Market Abuse: Limits on Anonymous Trading

The recent focus on enhancing the structure of US equities markets has seen a variety of proposals to make our markets better. Given the open atmosphere to new ideas, some of which have been quite radical, it may be appropriate to consider one more: changing the current paradigm of complete anonymity in trading. Market participants want to be anonymous for good reasons, and I believe there’s a way to satisfy their legitimate needs while still addressing what’s become a very opaque trading environment.

I can think of four reasons why traders would want to be anonymous:

  1. They want to hide their intentions to trade until they’ve completely reached their desired position.
  2. Highly successful traders don’t want to be blacklisted by counterparties fearful of trading with them.
  3. They’re afraid that their secret-sauce algorithm could be reconstructed if their orders were identifiable.
  4. They’re ashamed of what they’re doing.

I think almost everybody would agree that the first and second reasons are legitimate and that any change to the current system should address these concerns. The third concern encompasses a broad range of activity and may be a bit more controversial. A trader seeking anonymity for the fourth reason is likely harming market quality and we should consider ways of exposing them.

One way to expose undesirable trading activity would be for the SEC (or FINRA/exchanges) to make public de-anonymized order data long after the fact (say, 90 or 180 days). That way, even if the SEC is too overworked to investigate all imaginable types of market manipulation, other traders could. There would definitely be motivated people studying the data in order to learn about markets. And, given the SEC bounty for manipulation and fraud, there’s a sizable incentive to search for and report suspicious trading. One of the chief complaints about HFT is that it engages in legal but unethical behavior. For those traders within the letter of the law who would still be embarrassed by their trading, the prospect of public shame could be a useful deterrent.

The data would need to have fairly precise timestamps, and while that could pose a challenge, I think it may be readily achievable. The current MIDAS system looks like it provides a complete and precise view of the market. Exchanges (I believe) keep track of the MPID (Market Participant ID) associated with every order ID, which could be integrated with MIDAS to provide Broker-Dealer level identification of the entire market. Once the Consolidated Audit Trail is complete this task would be even easier, could include hidden orders, as well as identify traders at a more granular level like Large Trader ID.

Complete market data files can be quite large; for example Nasdaq order data is several gigabytes per day. The SEC/FINRA/exchanges could charge for bandwidth to cover any associated costs. They could even make a profit by selling this data, potentially covering some of the considerable costs associated with the Consolidated Audit Trail. This would, of course, require a change in the regulation.

This proposal clearly accommodates traders with reasons 1. and 2., as well as the most important swath of traders with reason 3. The substantial delay before the data is published should provide ample time for a position target to be realized. And certainly counterparties couldn’t use the stale information to avoid trading with the savviest participants. High-frequency traders afraid that their secret-sauce could be uncovered with analysis may have more cause for concern. Some algorithms would be immediately visible even by eye, but I’d argue that these would tend to be more latency-oriented strategies. Such strategies may as well be publicly known anyhow, since the main advantage their proprietors have are not the actual strategies, but their speed. Whether or not this data were public, if a competitor were faster than them, then they probably wouldn’t be successful. Strategies that police complex market relationships or interpret disparate data sources would be much harder to reconstruct, but in some cases it could still be possible. One of the principle arguments behind algorithmic trading is that it aids price discovery beyond the ultra-short-term. If an algorithm is truly making longer-term price predictions, then it would be very hard to decipher from de-anonymized data. If it isn’t, then maybe it deserves to belong to the fastest traders in the market. I’ll also mention that trading strategies need constant tuning; dissemination of information that’s 6 months out of date is probably not too significant a threat to operators of complex strategies.

HFTs usually have the most bare-bones interaction possible with the public. I think that’s partly out of fear that their algorithms will be discovered but also stems from a more mundane aversion to the public eye. This proposal would significantly reduce their privacy. There are arguments on both sides about whether this privacy is reasonable. Without getting into them, I will point out that large investment managers already have to report their long positions and that in some countries detailed financial information must be made public [1]. For instance, in the Netherlands annual financial statements are public for all large companies and, in the UK, employee compensation is public for certain financial institutions. These countries are obviously not finding the publicity to be an obstacle to retaining trading talent (plenty of HFTs are in both, e.g. Optiver or Jump). Nasdaq Nordic until recently required counterparty disclosure across the board and still does for many securities (p.21). So, if we were to see a shift away from the cloak-and-dagger stuff in the US, I’d expect that it would turn out just fine.

Because it is relinquishing significant privacy, I’d expect there to be resistance to this proposal. But there are significant gains to be had for HFTs as well. Conspiracy theorists paranoid about HFT are becoming increasingly influential. In order to be taken seriously, they would need to back up their claims with real data and show not just isolated examples of market weirdness, but systematic patterns of abuse. And, because this proposal would make abuse so easily discoverable, there would be much less of an impetus for more radical reforms that would put HFTs out of business entirely.

[1] One potential issue with this proposal would be that big short-sellers might find themselves in more danger of getting squeezed. Large long positions already must be disclosed, but to my knowledge currently big short sellers *could* keep their position secret. If something like the above rule were put in place, that would change. By seeing a trader’s executions, it wouldn’t be hard to estimate their position (although it’d be out of date). It’s true that overall short interest is already public, so this change wouldn’t be monumental. But if Hedge Fund A accounts for 90% of a stock’s short interest, then Hedge Fund B could start buying up shares until Fund A runs out of cash and is forced to liquidate at a massive loss. Fund B could probably even cause a short squeeze just by going on TV and announcing that they might do this sort of thing. Leaving aside whether Hedge Fund A deserves secrecy in this case, you could probably avoid these shenanigans by having a sufficiently long delay before the data is public. I don’t know if 90 days would suffice, but 1 year seems like it ought to. Amsterdam Trader tells me that in the Netherlands (and the rest of Europe) large shorts are public the very next day. And European markets function just fine with information like this, which is possibly recent enough to reveal a large short-seller before they’ve finished accumulating their position.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s