Liquidity issues have been at the forefront of the equity market structure debate since decimalization. The current shortcomings have spawned a cottage industry of technological innovations, including crossing networks and so-called “dark pools,” by new market entrants as well as exchanges and brokerage firms in the quest for the Holy Grail of trading — liquidity. While crossing networks and other ATSs have certainly enhanced liquidity sourcing capabilities, not everyone has access to all of these trading venues for reasons ranging from budget and commission restrictions to accessibility constraints. In addition, the more successful crossing networks become, the greater the danger to centralized liquidity and price discovery since by definition these dark pools minimize the extent of order interaction and the possibility of price discovery. Thus, exploring other solutions is necessary, and turning back the clock by “going back to nickels” is neither realistic nor desirable. So what, then?
The inherent conflict with institutions seeking liquidity without revealing their hand has the most-intelligent market structure observers and algorithmic mathematicians scratching their heads, looking for an answer to the seemingly unanswerable question: How can we increase our participation rate without unduly impacting the market? The importance of answering this question in order to make markets more efficient and to revitalize liquidity has everyone scrambling for new and improved ways to access different markets and liquidity pools through the use of technology. It is no secret that leading algorithmic providers are always on the prowl for the latest improvements to enable them to pull ahead of the competition and ultimately increase their market share. There is, in fact, an improvement that may be in the pipeline on the algorithmic front that can be implemented across all dark, passive and active liquidity aggregators that could substantially alleviate liquidity concerns, providing a safer way to advertise liquidity with a lower risk of missing incoming orders.
The use of technology to reduce manual intervention while maximizing sourcing capabilities at speeds unimaginable, which began in the late ‘90s, has become an industry standard. That standard is known as a smart router or aggregator. Smart routers, through both general and specific instruction by a trader or algorithm, take direct market access to a new level. In milliseconds, a smart router acts as a sweep tool to probe the spreads, find hidden liquidity, and access multiple exchanges and ATSs one by one, all in an effort to satisfy best execution and increase performance intelligently with as little footprint as possible. When it has exhausted its sources, the smart router follows whatever the instructions require at that point, performing less intelligently than its name would imply. Usually the next instructions are for it to camp on an electronic communication network (ECN), posting an order that shows only a small quantity along with a hidden reserve in the event that a larger order comes in. If stock trades on another ECN, generally it will cancel the order where it currently resides and migrate the balance of the instructed order to where the volume is. But I believe there’s a better way — I call it smart posting.
Let me take a step back and explain the need for the approach. The hesitation by institutions to “open up” due to the potential for information leakage has erected a roadblock between buyer and seller. We are all worried about gamers, frontrunners, shoppers, and ultimately anything or anyone that will impair our performance while trying to get our job done, hence the need for anonymity. One thing that has opened us up a bit is technology. There are parameters set in these electronic routers and algorithms that are scientific and mathematical, and we have developed a certain level of trust in them. The problem here is that the national best bid and offer (NBBO) is usually crowded with a number of buyers and sellers. Your algorithm might have your order camped on Arca waiting for a seller to hit your bid, but Brut, which might also be best bid, could get hit instead. After your algorithm sees Brut being hit, it will migrate your order to that marketplace in an attempt to participate in the volume. If it is too late getting there and you have any sort of a “volume participation” instruction (which many algorithms have), your smart router, acting on instruction from your algorithm, would likely start the same bid process described earlier in an attempt to maintain a percentage of the volume. While the algo certainly has still served a number of functions, including allowing you to concentrate on the more-difficult trades that need special attention, it has arguably failed in several important respects. It may be too late, and you may have missed not only the volume traded that was advertised, but also any hidden reserve. In order to catch up, you may have to chase the stock higher in order to satisfy your desire to participate. This is a double whammy as now you are not only likely lifting offers at higher prices, but also paying the spread.
What if instead of being represented in just one venue you had an order represented on Arca and the NYSE? In other words, what if the algorithm were “smart” enough to slice your parent order into multiple child orders on multiple venues and ECNs simultaneously? These questions sound elementary, but oddly enough, many household name algorithmic providers I surveyed currently are not doing this. And yet I believe the implementation of this simple idea could have considerable benefits for the industry. There are some providers who already have algos that are customizable enough for a trader to enter parameters into their algorithm, which then post liquidity on multiple venues. However, requiring such active involvement of the trader is impractical in many instances because the trader is typically working a list of stocks through the algorithms. This individual attention eats up much of the efficiency gains of using the algorithms in the first place, while order flow shifts between venues are too dynamic to be capitalized upon by an algorithm in which a static choice of venues is communicated at the outset anyway.
Clearly an algorithm that focuses on smart posting rather than just simply smart routing is a more passive approach to filling orders, but the performance enhancements can be substantial. Using the above example, if you were represented on all of the venues simultaneously as best bid, or just behind the best bid with a little discretion or some combination thereof, you would be a part of any stock that trades on that bid. In addition to helping you avoid missing stock, your average price should improve because you are buying on the bid and capturing the entire spread, rather than having your algorithm aggressively pay the offer in order to maintain its instructed participation rate once the stock was missed. You would be first in line (or at least tied) for any algorithms attempting to sell through “hitting,” which would attempt to exhaust your reserve, getting you better performance in every aspect of your execution.
Furthermore, algorithms using this concept will offer a number of benefits. If they are on as many venues as they can be without creating the appearance of a strong stock, they would be in the position to post liquidity rather than take liquidity resulting in rebates from these venues rather than paying ECN access fees. By using this strategy in all dark and passive servers, the performance numbers will undoubtedly improve over time and lead a marketing edge for those algo providers. Additionally, the amount of liquidity that would be posted to the market, which often sits hidden on a single ECN, should increase, especially with respect to illiquid stocks. Aggressive algorithms will have more potential to find liquidity, and those passive trades would reap the benefits of being patient.
Although potential problems with over executing orders and making the stock look too strong are potential negative consequences from such an approach, these are two issues that can be mitigated with a simple tweak of the algorithm. Slicing the order could alleviate concerns of over execution, while not posting the order is the most obvious solution. Also, making stocks look too strong would be a non-issue if representation in multiple venues were executed intelligently. Remember, multiple ECNs/exchanges already dominate most NBBOs of over-the-counter (OTC) stocks and increasingly listed stocks. They are most likely already on the inside. Joining whatever is top of book on each ECN or venue with a small order of 100 shares rather than beating the top of book, while potentially pegging it as well, will allow you to capture any flow preferencing that venue. Being behind the top of book, with some discretion, and having the algorithm actively change its position while maintaining its exposure on each venue also will guarantee increased participation. Basically, what some low-touch algos currently are doing while camping on one exchange, they should be doing on all venues where there is a potential for stock to trade.
The only question now is what effect will Reg NMS and the NYSE Hybrid system have on this model? While these two concepts are creeping along in their respective implementations, this theory of having low-touch algorithms post liquidity wherever possible should be put into production without further delay. There are few incentives that can be used to entice institutions to post more of their liquidity. However, recent history proves they are willing to dump orders into electronic systems with mathematical and scientific parameters, so why not capitalize on that fact and maximize its potential? I don’t know traders or firms that are happy to see stock trading around them, including one of the most respected firms on the Street, which recently began the research and development of implementing this strategy in its dark, active and passive next-generation algorithms. If this becomes an industry standard, we could see liquidity increase substantially —liquidity that was, for the lack of a better explanation, right in under our nose!