Heavy Rotation
One of the supposed truisms of our modern-day US market structure is that access is synonymous with opportunity. If you’ve heard this song before, feel free to hum along with me: the more places we route our orders, the better the execution. This maxim was first popularized in the late ‘90s when ECNs ambushed a third of the exchanges’ flow seemingly overnight1. A few years later, it was remixed for Reg NMS when the number of dark pools exploded like Mentos in a Diet Coke. In both those cases, substantial liquidity (often at better prices) appearing on new trading venues created a catalyst for technological pivots such as smart order routing and dark pool aggregation.
On today’s Wall Street playlist, conditionals sound like an acoustic cover of these greatest hits stripped down to the essentials. No new venues, just new routes. No new liquidity, just streamlined access to blocks.
Recently, the ballad of conditionals has been on such heavy rotation that it would make even Taylor Swift blush. Brokers are feverishly building conditional routing into their dark pools while simultaneously creating outbound conditional connectivity to everyone else’s2. The prize this time is simple enough: efficiently capture more blocks across the spectrum of the non-displayed markets. Intuitively, this makes sense, but before you ask your brokers to set this anthem on repeat (in essence to “spray and pray”), let’s first hear what the recent FINRA data is telling us to see whether conditional routing is hitting all the right notes.
But first, some background music…
Blocks Get Piggish
Over the last two decades, buy-side traders have relied on block-crossing networks to provide one of the more efficient ways for large institutional buyers and sellers to find each other. Crossing networks all offer variations of “blotter sync” technology that read indications residing on buy-side OMSs and leverage them to initiate bilateral negotiations.
As dark aggregators grew in popularity, conditional orders became a creative way for dark venues – both crossing networks and standard dark pools – to simulate the efficiencies of the blotter sync. A conditional order allows algorithms to represent interest to trade across multiple dark pools simultaneously without committing the order to any one venue until an opportunity arises. Today, all six block-crossing networks and (at last count) thirteen dark pool providers allow conditional access to their venues.
This is where things get interesting. Based on what now appears to be a network of shared electronic block information, traders may assume that conditionals would all but eradicate the scrutiny of venue selection. But, in terms of block executions, this isn’t the case at all. In fact, the data that follows will show us that the distribution of block fills in ATSs is more Orwellian than equitable. That is, when it comes to where blocks print, the pigs will tell you “all animals are equal, but some animals are more equal than others.”
Or, in other words, as the use of conditionals has proliferated, the sources of blocks have not.
The Song Remains the Same
When we compare Q1 2017 FINRA data (the earliest available block statistics posted) to Q1 2019, arguably the time period where conditional routing has flourished, what initially stands out is that crossing networks have provided and continue to consistently provide roughly two-thirds of all ATS block volume (Exhibit 1). Quarter vs quarter, blocks in crossing networks slightly increased 1.3% while blocks in dark pools with conditional access gained 2.1%. This is all to the detriment of venues without inbound conditionals which now capture 10% of all ATS block volume.
At first glance, even though the largest percentage shift went to venues that offer conditional routing, the allocation of block fills across ATSs by these categories remains relatively stable. [IMGCAP(1)]
Breaking out those block volumes by venue for Q1 2019, crossing networks represent the top five providers and six of the top ten (Exhibit 2), providing possibly the best visualization of the hierarchy of blocks in US equities ATSs.
EXHIBIT 2: [IMGCAP(2)]
Not So New Kids on the Block
Additionally, crossing networks continue to provide larger block prints than do dark pools.
Although the average size of blocks in crossing networks has dropped marginally over the two-year interval (32,921 to 28,494 shares) while pools with conditionals have ticked up a shade (18,230 to 19,875 shares), crossing network fills are still 43% larger than fills in dark pools with conditionals (Exhibit 3).
The reason for this disparity may be simple: a venue’s block size is likely dictated by that ATS’ main sources of liquidity. Whereas crossing networks derive a good percentage of their block liquidity from OMS blotters, which is assumed to result in the buy side’s full position being represented in the venue, dark pools offering conditional access regularly draw their inflows from algorithmic orders that often represent only a portion of the buy side’s larger indication size.
Interestingly, in Q1 2019 the average block fill size in venues that don’t offer conditional routing is larger (21,184 shares) than venues that do (19,875 shares). Although some of that skew comes from pools that focus on VWAP or closing crosses, if we view these pools as the control group, we may be witnessing the placebo effect where the marketplace has inferred a usefulness on conditionals that doesn’t necessarily exist.
Based on both the market share (Exhibit 1) and execution size data (Exhibit 3), we could argue that the effects of conditionals on the dispersal of blocks across ATSs has been mostly negligible.
EXHIBIT 3: [IMGCAP(3)]
In advance of Reg ATS-N (which this blog will dive deeply into in the coming months), there’s one additional FINRA metric that reveals another dimension of the grey anatomy of dark venues.
Beyond the block, breaking out all executions across ATSs by venue or venue type, we can quickly spot the variation in fill between pure block crossing, standard dark pools, and hybrid versions that populate the landscape in between.
Two crossing networks in particular provide average execution sizes across all fills that are more in line with a pure block-seeking strategy (Exhibit 4). The majority of venues – dark pools with and without conditionals accounting for 25 of them – tend to have average fills nearly twice the size of the Exchanges 6 but far below the threshold of 10K shares, the standard minimum block size. The average fill size in hybrid pools, or venues that support crossing network functionality (i.e. blotter sync), as well as traditional dark pool resting orders tend to be dictated by which of the prior models dominate.
Without passing judgment on which type of venue best complements your investment style, the typical fill size from a pool in conjunction with its percentage of total ATS volume should influence your decision tree for order exposure length and frequency.
EXHIBIT 4 [IMGCAP(4)]
Going Forward: The Lessons of Milli Vanilli
The lowest point of pop music in the ‘80s – as witnessed intently by those of us who owned multiple pairs of parachute pants, and you know how you are – had to be the Milli Vanilli scandal. In short, the German pop duo and New Artist Grammy winner was caught lip syncing their hit songs whose vocals were recorded by someone else. Audiences quickly began questioning the talent of other artists, forcing most to prove their chops in live performances and likely spawning the more cynical music era of Grunge in the early ‘90s.
In the same way, as conditionals take center stage in our market structure today, it’s up to us to determine if our brokers’ dark aggregation strategies really do what they claim they are doing.
Algorithms are certainly the most important and impactful trading technology of these last two decades. They are the rock stars. Conditionals are an exciting innovation that can drive more efficient routing and better access to key liquidity. It’s up to us practitioners to scrutinize their function and determine if they are as integral as the cash register sample starting Pink Floyd’s “Money” or are they just more cowbell.
More words and music from me to follow…
1 According to the SEC’s “Special Study: Electronic Communication Networks and After-Hours Trading” from June 2000, ECNs account for approximately 30% of total share volume and 40% of the dollar volume traded in Nasdaq securities, https://www.sec.gov/news/studies/ecnafter.htm#pt2
2 As referred in https://tabbforum.com/opinions/conditional-orders-the-great-liquidity-aggregator/ which states “…conditional orders quickly are becoming the de facto method for buy-side traders to aggregate liquidity in a decentralized marketplace. Despite the widespread adoption of conditional orders, however, the complexities around evolving use cases, as wells as the fact that each ATS offers its own flavor of the order type, can be confusing.”
3 FINRA ATS transparency data, Tier 1 and Tier 2 ATS block volumes for Q1 2017 and 2019 https://ats.finra.org/Agreement
4 Excluding Dealerweb, an IDB which only trades ETFs and other non-equities
5 Venue categories derived from Liquidnet’s outbound connectivity as well as various public sources. ATSs listed individually reflect current venue names as per the SEC website (https://www.sec.gov/divisions/marketreg/form-ats-n-filings.htm)
6 Average execution size in Exchanges for July 2019 for all Exchanges in aggregate was 148 shares based on data calculated from the Historical Market Volume Data supplied by CBOE (https://markets.cboe.com/us/equities/market_statistics/historical_market_volume/)
