Dark Pools: Going Beyond The Dark

By: Jim Ross

File Under: MatchPoint, NYSE

• 0 Comments
• Email Entry

In my last post (which was an eternity ago back in July (just ask my 401(k) -- it is kind of a 201(k) now), I highlighted the curious phenomenon that most of the dark pools, which employ anonymity and opacity in order to attract larger order size, have average trade sizes well below 600 shares. And who needs to cloak a whopping 300-share order anyway? But the small-trade-size phenomenon is more than just an intellectual curiosity. It is a symptom of a much bigger problem. We have lulled ourselves and the investors into thinking that anonymity and order opacity will fully answer for the negative performance attributed to transaction costs. And while "darkness" is a critical component of the transaction-cost puzzle, it does not in and of itself complete the whole transaction-cost “picture.”

Wayne Wagner, a pioneer and a leading figure in transactions cost analysis over the past 2+ decades, places transaction costs into two categories: visible and hidden costs. Visible costs like commission, spread costs and market impact are relatively easy to identify and quantify. Competitive pressures, technology and regulation have greatly decreased both commissions and spreads but anonymity and order opacity have been driving factors behind minimizing (though not eliminating) market impact.

Market impact is an elusive quarry, though. For while an order in a continuous dark pool or dark algo may be completely anonymous and totally non-displayed (pre-trade), it can still incur substantial market impact post trade. Jamie Selway gave a good example in our last blog chat when he said, “What about the fact that as you traded at the midpoint 123 times over the course of an hour, the midpoint moved against you 2%? No ‘tactical’ market impact, sure; but obviously tremendous impact overall.” The multiple-small-trade “footprint” of a continuous dark pool or dark algo can give away post trade all the transaction’s cost benefits (and more!) that pre-trade anonymity and opacity can bring to an order.

But, hang on; we have not even gotten to hidden costs yet! The hidden costs that Wayne identifies are delay and missed trades. In a recent presentation, Wayne described them in this way: “Hidden costs are 2-3 times the visible costs. Hidden costs arise from the need to trade in sizes that swamp the market. Waiting -- or searching -- for liquidity creates these costs. They are often as poisonous to performance as any other cost.”

The obvious consequences of delay and missed trades are information leakage, slippage, opportunity cost and even gaming. Each one festers the longer it takes to complete an order whether it is sliced into small orders over time or shopped to multiple dark venues.

But there is also a much-overlooked consequence and that is the lost benefit of the original investment decision (first-mover advantage, if you will). We focus so much on avoiding the costs of a transaction through slicing and dicing that we fail to capture the timely VALUE of an investment decision. Executing a block transaction quickly is still a valuable and perhaps preferable execution strategy to slicing a block into tiny orders.

For a dark pool to be truly effective across the transaction-cost spectrum (and thus capable of trading large blocks), it needs to be more than just “dark.” And clearly one of the more problematic aspects of today’s dark-pool and dark-algo offerings is their continuous, real-time nature. (Interestingly, all of the dark pools with small trade sizes are also continuous.) Each trade becomes a footprint and each subsequent footprint becomes a billboard betraying the investor’s trading intentions.

For those who know me, yes, it appears that I am simply trumping up the fact that a totally dark point-in-time cross like MatchPoint is the solution to the continuous issue (and it IS an elegant and effective solution at that!) but there are dark pools (continuous and not) that do go beyond “darkness” and are constructed in a way that smartly and effectively minimizes post-trade market impact and the hidden costs mentioned above while still catering to anonymity and order opacity.

Clearly point-in-time dark pools (like MatchPoint, Instinet VWAP cross and POSIT matches) break the cycle of continuous trade “footprints,” and “pinging” them is impossible. They also aggregate all trades from multiple parties into one trade report “footprint,” limiting even further any unnecessary information leakage. Other dark pools use minimum order-size requirements or functionality (like Pipeline) to inhibit algo “pinging” and attract true block liquidity, significantly reducing delay.

And block negotiation systems (like Liquidnet and BIDS) provide a pre-qualified vetting of contraside interest to reduce information leakage and promote natural block discovery as well as a “scorecard” concept to punish gaming. All of these are innovative, effective solutions to post- trade market impact, information leakage, slippage and gaming that can occur in a continuous dark pool environment, and there is more to be done for sure.

Block trades as much as algo order slicing are integral parts of the investment and transaction cycle. Understanding the transaction-cost “balance” between the two will help us evolve our marketplace to better accommodate and serve the diverse and sophisticated needs of all investors.