Bloomberg published an article called Trader VIP Clubs, Ping Pools Take Dark Pool Trades to New Level which described the rise of single dealer platforms. While the article is accurate that such platforms are gaining market share, that they provide market makers a method to provide better liquidity to better clients, and that MiFID II essentially promotes a close cousin to the US pools called Systematic Internalizers, it creates an interesting impression. For example, reading this article, one would not understand that the architecture of the larger Ping Pools is such that the market maker responds by algorithm inside of the pool to inbound orders. In many cases, orders that are not filled are not communicated to the market makers other systems at all, or the market makers have strict rules against utilizing the information in real time.
That said, if there was a reasonable disclosure regime on the part of routing brokers, the statistics would either show the value of these pools or call into question why brokers route there. I have called for specific, statistical disclosures in multiple comment letters to the SEC and described the idea in previous posts. Simply put, routing brokers should be required to show statistics per routed venue, broken out by order type. In this case, analysis of Marketable, Immediate or Cancel orders, sent on behalf of Not Held orders, grouped by size buckets, would paint the needed picture. If, for these orders, the number of shares ordered, fill rate, and execution stats such as price improvement and effective spread were provided per venue, it would help clients understand the routing. If, however, the statistics also included the short-term price movement after sending the un-executed orders, it would help to determine if there is consistent information leakage. This is a bit problematic, if the behavior of the routing firm is to immediately route the order elsewhere when it is not filled. In those cases, it is quite possible that the statistics would show leakage, but the cause might be subsequent orders and not the order routed to the single dealer platform. Brokers, however, would know their own behavior and should consistently look for this type of price movement in order to determine if the venue can be trusted.
The reason that clients should insist upon this type of analysis, instead of simply avoiding these pools, is that avoidance could cause them to pay higher transaction costs. This is because market maker liquidity, along with central risk book liquidity, if deployed properly, should be very desirable to institutional managers. The reason is that market makers and central risk books rarely trade aggressively, and are managed somewhat passively by hedging exposures instead of trading out of positions. As a result, there is less likelihood that they will create additional market impact. Of course, as the Bloomberg article pointed out, money managers dont believe this, based upon the statement that: One-third of equity traders surveyed at U.S. buy-side firms say automated market-making firms exist primarily to collect incentives paid by exchanges, according to a Greenwich Associates report.
That statement is truly sad. It either displays an appalling failure to understand how market makers operate, or points to very poor survey question design. The simple fact is that market makers provide liquidity to a level just above what economists would call an indifference point. In plain English, that means that if they expect a net profit trading 2500 shares at a price on a specific venue or to a customer, but that the next 100 shares would not make incremental profit, they will only offer 2500. The net profit, meanwhile is based on the combination of the bid offer spread they expect to capture and the fees they would pay (or be rebated by the exchange). Considering the difficulty in capturing bid offer spread when posting displayed quotes on exchanges, it is true that for many stocks, the rebate is an important consideration. That said, saying that the market makers exist primarily to collect rebates is flat out wrong.
A couple of years ago, I did a study for KCG on their ping pool which has since been acquired and re-branded by Virtu. We found statistically significant evidence that the pool offered unique liquidity to the brokers that routed orders to it. That finding is quite consistent with the notion that the clients of the pool were a VIP club, since they did receive a benefit. What is most interesting, however, is that so many money managers would prefer to sit in the economy section and not participate…
