By Hitesh Mittal, Founder & CEO, BestEx Research
Single Dealer Platforms (SDPs) are an alternative source of liquidity for execution algorithms, alongside exchanges and ATSs, and are quite popular because they offer liquidity at no commission to brokers. But the way SDPs operate and the implications of using them are not well known. To make matters worse, myths related to SDPs create misunderstanding around their effectiveness. We answer a number of questions in detail in our paper “Accessing SDPs in Execution Algorithms: Penny-Wise and Pound-Foolish?” but address the biggest concern here—parent order information leakage due to serial correlation among child orders.
The information leakage for an algorithm sending a market order (or marketable limit order) to an SDP is far greater than other ATSs or exchanges. This is a result of a fundamental difference between SDPs and other venues—the counterparty an algorithmic broker faces when accessing these venues. Orders sent to an SDP trade only against the market maker operating the SDP, whereas orders sent to a dark pool or exchange can trade against a variety of counterparties.
When sending a marketable order to an exchange or ATS, the order is anonymous to the counterparty trading against it. Hence, they have no ability to reconstruct the counterparty’s trading patterns. Contrarily, in the case of a market order sent to an SDP, the SDP knows a great deal about the order. For example, it may know that the order originated from an institutional broker-dealer’s execution algorithm, and hence is likely a small slice of a larger parent order. If a broker algorithm sends a series of 100 share market orders to buy to an SDP, it is a lot more likely that they belong to a large order to buy that stock. Since they know that the order is coming from a certain broker-dealer and they also have historical information about the serial correlation of child orders on a given side for that broker, they can derive information that would be very hard to uncover if they were making markets at an exchange. When a market maker’s offer is lifted on an exchange, it is very difficult to determine whether a large parent order is behind the trade because there is little serial correlation in marketable orders on exchanges.
We often hear the common myth that, because a single broker may represent multiple client orders sent to an SDP, an SDP would not know whether the order is coming from a particular asset manager. While this is true, there is still a lot of valuable information known to an SDP that is not known to a market maker at an exchange. For example, SDPs receiving orders know that the orders are from an institutional broker (informed order flow), that they are slices of larger orders, and whether the aggregate imbalance of the institutional flow from a given broker is a buy imbalance or sell imbalance. And none of this information is known by market makers on exchanges.
To make matters worse, SDPs aim to further segment the flow by asking each broker to categorize order flow into different toxicity groupings. Not all brokers allow segmentation, but the more segmentation a broker provides, the easier it is for the recipient to establish patterns.
Given the available information, the question is whether SDPs use only the information about a given child order (likely less harmful) or if they use the serial correlation of child orders, the identity of the broker, and segmentation information to inform market making and/or non-market making activities.
There are no regulations requiring SDPs to provide transparency around how they use this information in their market-making or non-market making activities. While exchanges are completely anonymous, and ATSs must provide complete transparency on how information about certain orders can be used by subscribers and affiliates via public ATS-N filings, SDPs do not have these requirements.
All an asset manager can rely on as it relates to how their information is used is what is disclosed in an agreement with an SDP. However, since the broker-dealer trading the order is the client of the SDP–rather than the asset manager–they seldomly have access to those disclosures. We couldn’t find public disclosures for most of the SDPs, but we did find one that is alarming; [The SDP] “generally may use information about current and historical client orders, cancellations, and fills (both full and partial), internal positions, real time market data, and commercially available data, information that clients provide about their customers (if applicable), information about clients’ and, if applicable, their customers’ trading activities and patterns, and any other information clients transmit or otherwise provide to [SDP Name], in determining whether to fill a client order (in whole or in part), the price at which to fill an order (including whether to provide price improvement), and, more generally, in connection with any other trading or business purpose of [SDP Name], including its market making and nonmarket making activities.” [emphasis added]
Despite such alarming disclosures from some SDPs and relative lack of transparency from others, the real question is whether brokers are measuring the potential harm or benefit of using SDPs effectively in their best execution analyses. We routinely hear that both SDPs and brokers discuss child order markouts and child order fill rates to imply that accessing SDPs does not hurt the performance of execution algorithms. We discuss in our paper at great length why these metrics are flawed, but the biggest flaw is obvious. The potential risk of using SDPs lies in whether they can use the historical patterns from a broker and serial correlation of child orders to establish information about the parent orders–not the performance of individual child orders.
The most effective way to analyze the impact of SDPs on the implementation shortfall of parent orders is to run a controlled experiment where a large sample of parent orders from a fixed set of clients are randomly distributed to two equal algorithms—the only difference being that one routes to SDPs and the other does not. By statistically evaluating the difference in implementation shortfall of the two algorithms, the positive or negative impact of trading with SDPs could be uncovered. In the study we describe above, the important element is that the measurement is matched with the specific concern of information leakage at the parent-order level.
Such experiments require a very large number of parent orders and an interest from the algorithmic brokers in running them, which may be misaligned given the huge explicit cost reductions associated with SDPs (or the fact that many run their own or have plans to run their own SDPs). In our estimate, it would take approximately 50,000 full day orders to confirm a 1 basis point impact on implementation shortfall by using/not using a specified execution venue.
SDPs do provide an alternative to exchanges and ATSs and add to the diversity of venues that institutional brokers can access. In our paper, we address some of the less well-known facts about SDPs and dispel associated myths. Given their widespread use, we believe regulators should apply the same level of scrutiny to SDP operations as ATSs and exchanges and require complete transparency around operators’ use of order information
