A Tale of Two Stories...

As Deep Throat said in All the President's Men: "Follow the money. Always follow the money."

Two, seemingly, unrelated events over the past couple of months have triggered significant debate over equity market structure. First, the disclosure that ITG secretly conducted proprietary trading, despite marketing themselves as an un-conflicted "agency" broker, set off a series of renewed objections to "Dark Pools". More recently, the imbroglio caused by NYSE invoking rule 48, to "speed up the opening process." has re-opened debates about their floor-based model. What's interesting about these two controversies, is that they are very much related. As Deep Throat said in All the President's Men, "Follow the money. Always follow the money". In both situations, the firms derived profits from these activities. As a result, both situations are examples of conflicts of interest and need to be viewed as such.

Most people, upon hearing about ITG operating an internal trading group, understood the conflict of interest, intuitively. The NYSE's motives, however, do not seem to be well understood. Most know that the NYSE, despite dominance in the business of corporate listings, currently struggle for market share in trading. Despite this, I have read nothing which explains their motive to maximize market share around the open.

To understand the importance of the open to the NYSE, consider their own opening "print" is the second largest trading volume point of the day, on average (the closing auction is the largest). Also, the NYSE chargesboth sides of the trade, 10 cents per hundred shares, to transact instead of paying "rebates" to the liquidity providing firm. They do, however, cap this fee per member firm at $30,000 per firm.

Also consider, the market is not considered "open" by the average investor, until the official opening print occurs. This means that most retail and institutional trading interest is either fully or partially benchmarked to the open price, which forces most broker dealers to either submit orders to the NYSE open or take a risk that they will miss out on that price. As a result, the NYSE's relative market share at the open is much higher than during the bulk of the trading day. After the opening auction, the NYSE market share declines quickly to its intra-day average, until the closing auction. The NYSE, therefore, has a vested interest in controlling the opening print.

The NYSE justifies their manual opening process by claiming that they inject "human judgment" to control volatility in individual stocks. In situations where individual stocks might over-react to company specific news, this is probably true. Many believe that experienced traders are better able to determine when individual stock prices move "too far", than computers and are happy for that element in the NYSE market structure. Unfortunately, however, when the volatility is systemic, that human judgement becomes an impediment to the overall process, due to the fact that the markets and securities are so interconnected.

In situations such as the large move in the market on August 24th, the opening prices would have been better determined by processing and distributing all the orders electronically, or slowing down the opening process on all markets in a coordinated fashion. Nuance was not a significant factor, compared to the vast number of orders in correlated securities that were being sent to the brokerage community and then to the stock exchanges. In this type of situation, not only does the human element fail to add value, it can result in serious adverse impact to investors.

The problem is that the NYSE does not operate in a vacuum. There are many other exchanges trading NYSE listed stocks which have market makers, traders and investors that either wait for the NYSE to open stocks or are forced to take risks on where those securities will open, and with rule 48 in place, with less than normal information. Worse, there are investors, market-makers, and traders in related securities, such as Exchange Traded Funds, that are essentially forced to operate blindly. The result, which has been well-chronicled, is that ETFs, with meaningful numbers of underlying stocks waiting to open, experienced extreme volatility, relative to both historical norms and the market as a whole. This caused many investors in ETFs to suffer losses and provoked a general lack of confidence in the market. Considering the importance of ETFs to both individual investors and the investment advisory community, we should strongly consider the better synchronization of opening and trading halt processes, as well as, new methods of measuring ETF execution quality. During periods of volatility, ETF trades should also be compared to the actual NAVs of the instruments, as opposed to, exclusively, measuring them based on the quotes on the exchanges. If investors truly understood that they were sending orders that would result in selling their holding at 20%, or more, below their actual value, they wouldn't. Moreover, if brokers measured trades that way, its likely market makers would have been forced to greatly improve the quality of their executions.

The ITG revelation has been described well by others, in particular by Matt Levine (Bloomberg) who published an excellent piece detailing, in the context of ITG, the broad problem with broker to broker routing strategies. While the piece was specific and well done, I believe conflicts of interest should be examined more broadly.

The real problem, is that conflicts of interest manifest whenever a broker charges a commission to their client and does not pass thru explicit trading costs on a "cost plus" basis¹. While it is not always as explicit as the "AlterNet" example, described by Mr. Levine, it is true, as my father would say, "you get what you pay for…" Routing strategies that attempt to minimize explicit costs often result in inferior execution prices, rather than those that don't.

To illustrate this point, compare a parallel routing strategy with a sequential, cost minimization strategy:

A client uses a Smart Order Router (SOR) to purchase 8000 shares of stock XYZ. In this example, there are 9000 shares of XYZ available at the offer price of 50.24, but the offer consists of 6 individual quotes as shown here:

If the SOR was instructed to send orders on a parallel basis and sent six orders simultaneously to all venues, the likelihood is that they would receive most of the shares they were seeking². If, on the other hand, the router attempted to fill the order by sending the full order to the cheapest market, followed by sending the full balance to the next cheapest, and so on, they would not likely succeed. (Most of these strategies would access the Boston and Edge A stock first, then try a number of cheaper ATS's next and then go to IEX, followed by BATS, NASDAQ and ARCA.) In all likelihood, some or all of the shares offered on ARCA, NASDAQ and BATS would vanish before the router was able to purchase them. This is not the result of nefarious activity on the part of HFT firms, but is a normal reaction to market activity. Market Makers are always cognizant of the dynamics of the market, and will react to the low priced liquidity disappearing and other reported stock trades on inexpensive venues, by pulling their quotes.

(As an aside, IEX's "Magic Shoe Box", in this example, would not be helpful. Round trip order times, for average routers operating in a sequential manner, are relatively long (500 microseconds or more) and would also include orders sent to other dark pools, in between sending orders to the markets listed above. Unlike the situation chronicled in "Flash Boys", this scenario involves multi-millisecond delays on average)

My point is that firms who opt to use explicit costs as the sole determinant in routing strategies are likely doing their clients a disservice. I'm not suggesting that most firms do this, nor am I suggesting that every situation using sequential routing will be problematic. I am suggesting, however, that the only way to manage this potential conflict is to accurately measure routing efficiency. The problem is that this is not done very often today. In particular, a lot of routing activity is not even measurable by the end client. As I have written previously, very little routing is subject to public disclosure and few brokers and fewer investment managers measure this specific aspect of the trading process.

Keep in mind that, in many cases, the investor is not trading directly with the firm making these routing decisions. In the case of ITG, the AlterNet system gave broker dealers a choice of a high explicit cost and a superior execution price or a low or non-existent explicit cost at an inferior execution. ITG disclosed this fact, but it is certainly an open question as to whether broker dealers should be able to make such decisions without disclosing them to their clients.

I am not going to reprise my arguments on dark pools here, except to say that the ITG disclosures changed none of my opinions. Dark Pools are still a useful tool for investors and they still need to be better measured. If anything, these stories may push the SEC to consider requiring such disclosures. If that happens, then the world will be a better place.

¹"cost plus" pricing is provided in the minority of instances due to technological complexity in both determining the actual cost and in fitting the information into existing confirmation systems

²It is not always true, particularly if the order was motivated by a very short term upward move in S&P Futures, but, all other things being equal, 8000 shares would get filled.

David Weisberger, Managing Director, Trading Services at Markit
Posted 10 September 2015

S&P Global provides industry-leading data, software and technology platforms and managed services to tackle some of the most difficult challenges in financial markets. We help our customers better understand complicated markets, reduce risk, operate more efficiently and comply with financial regulation.

This article was published by S&P Global Market Intelligence and not by S&P Global Ratings, which is a separately managed division of S&P Global.