For as long as I have been around a trading desk, the phrase limit order, limit business has been a popular mantra among traders. The expression works on multiple levels. On the surface, its a truism that speaks to the nature of brokerage, a commission-based business where traders only make money on orders they fill, not orders they receive. On another level it serves to chastise the entitled blowhards who cover larger institutions: its not the size of your book, but the amount of business you generate from it.
Its most frequent usage, however, is as an homage to the Glengarry Glen Ross school of motivational speeches and is drilled into to you the minute you pass your Series 7. For David Mamets characters, coffee is for closers. On Wall Street, its the trade that gets you paid.
In the early 2000s, the equity market made an abrupt and unrelenting move toward electronic trading. Big, meaty institutional orders that were previously matched in chunks between humans were now being sliced up and shipped out to multiple execution venues where the regulators hoped that a more competitive trading environment would better the quality of execution.
But even though serving sizes for equity orders shrunk from Porterhouse to Carpaccio, brokers were still raised as carnivores, and crossing quickly migrated from the free range telephone exchange between sales traders to the factory-farmed, sunless pens of dark pools.
In Part 1 of Tales From the Dark Side, we discussed how this paleo diet was nourished by enormous inflows, some from institutional liquidity but a greater percentage from other brokers algos and HFT. Serving this Spam-like order flow alongside the more nutrient-rich institutional liquidity gives dark pools the appearance of a heartier meal, but, in these highly processed electronic markets, even full plates can result in empty bellies and discourage customers to come back for seconds.
In order to maximize liquidity, dark pools rely on efficient matching logic. To understand the mechanisms of matching, we will first look at the degrading economics behind running a dark pool. Then we will review their baseline functionality and examine the good and bad of anti-gaming filters. Finally well discuss simple guidelines that help you refine your dark routing choices.
The Achy Breaky Product Life Cycle
Some fads peak and decline almost in the same breath. For dark pools, it been less than a decade since we grew this mullet and already the music has begun to fade. Looking at the volume and revenue trends, you get the feeling there are natural limits how long we can perform this two-step.
When dark pools were block crossing engines, the economics were great. Two million shares a day executed at 2 cents per share between institutional clients generated around $10 million in gross annual revenues.
But with the boom in electronic trading and order slicing, block crossing engines gave way to streaming dark pools and competition triggered pricing compression. Suddenly, commission rates were closer to 50 mils a share, and to achieve that same $10 million in annual revenues, a dark pool now had to execute 8 million shares a day.
Then, overnight, everyone had a dark pool. Confused and overwhelmed, institutions stopped routing directly to them and dark pool inflows were mainly driven by brokers and HFT. With average commission rates plummeting to closer (and in many cases under) 10 mils per share, to maintain that $10 million watermark a pool needs to attract 4 to 5 billion of shares of inflows and cross 40 million shares a day.
So less than a decade after Reg NMS and the introduction of streaming dark pools, many brokers are struggling to maintain profitability along with the threat of more regulations that may increase costs and limit crossing opportunities. With numbers like this, it makes sense that brokers dont want to give up any chance to trade an order. Therefore, most of dark pool matching logic tend to mimic each other, like line up of parrots repeating Polly wants a cracker in unison, with a few flashing more colorful plumage.
Exhibit F
The best place to understand how matches are made is from the broker dealers themselves. All dark pools register with the SEC via Form ATS-R. After the recent accusations leveled against certain pools, many of the larger operators have released their Form ATS via the web. Scrolling halfway down one of these documents will get you to a section called Exhibit F which will more or less explain how the dark pool operates.
A scan of the larger dark pool filings reveals some general consistencies. Most pools, like exchanges, use price/time prioritization to match orders. Limit, market and pegged orders are all standard and most will allow a trader to place minimum fill size. All pools claim to fill orders at or better than the current NBBO, although some highlight their use of direct data feeds from exchanges which are generally considered more accurate and less prone to arbitrage. Most choose not to execute during a locked or crossed market, none (should or do purposefully) when a stock is halted.
The one component of dark pools which has the greatest variance from filing to filing is the way they handle participant categorization and counterparty exclusions. These rules are the proverbial doorman controlling the velvet ropes and can be the difference between VIP status and general admission.
Anti-gaming: The TSA of ATSs
We all share universally mixed feelings about the TSA. Our understanding of their vital importance juxtaposed with what one day appears to be complete ambivalence and the next spiteful totalitarianism. We know we need them but we dont always trust that they are effective.
As markets rapidly evolved and dark pools matured (again, seemingly overnight), customers (mainly institutional ones) began to demand greater protections against contra parties who could potentially deconstruct their larger orders, even with the risk of lower fill rates. As a result, many brokers retrofit a caste system into their pools which allowed for order segmentation and participant exclusions.
The net result is more good than bad, but be prepared for some arbitrary nail clipper and liquid/gel restrictions at the security checkpoint.
Some pools care how the order in entered (Is that a fast connection? Must be a fast trader then). Others are fixated on who specifically entered the order (I know those guys. They are usually informed traders. Stick them in that bucket.). And others throw out those variables and make strictly quantitative assumptions based on historical actions (See, every time they trade, the stocks move. They must know something. Block em!)
If a pool wants to draw in institutional flow, participant segmentation has become a prerequisite. But some of these filters seem marginal, illogical or, at worst, self-serving. For example, what happens when a large, passive institutional customer uses high capacity pipes or an informed trader connects on standard transmission lines? Also if a huge percentage of order flow comes from other broker algos and SORs anyway, how can a pool categorize an order when it cant be sure of its origin?
In modern markets, the general feeling is that categorization and exclusion can be executed with enough effectiveness to be worthwhile. Just like the challenges associated with inflows, as complexity increases so does the need for vigilance.
Champagne of Table Water
In some ways, dark pools are like bottled water. Although the advertising portrays a crystal clear spring in some pristine land, your gut tells you its most likely drawn from an industrial tap. What makes you choose bottle A over bottle B is, more or less, the taste (or lack thereof) it leaves in your mouth. Although Perrier claims they are the Champagne of Table Water I have always strongly preferred Pellegrino. Then I found out they were both owned by Nestle.
So, when choosing which dark pools to route to, put the marketing material aside and follow these simple guidelines. First, Occams razor applies. The more convoluted the rules of a pool, the less likely you will really understand how your order is being handled and whether your executions are a result of intelligent filtering or plain happenstance. Stick to the pools that keep it simple. Second, ask for greater specifics about categorizations. And make sure you ask the product manager or the principal who runs the dark pool. They are more knowledgeable and have more at stake.
Finally, if all else fails, only interact with a pool using midpoint peg and a minimum quantity. That cocktail is the closest thing to a natural diuretic for toxic liquidity.
In Part Three of Tales From the Dark Side, we will sum up this topic as we explore the most practical way of access multiple dark pools at once.
A 20-year veteran of Wall Street, Craig Viani has built and managed dark pool and smart-order routing technologies. He is currently an independent consultant and mainly a pescetarian so he apologizes for any inaccurate portrayals of any beef or beef byproduct and assures no animals were injured during the writing of this article.
