DARK POOL DIFFERENCES.
Henry Yegerman, Global Head of Sales at ISS LiquidMetrix.
Are there significant differences between Dark pools and, if so, what are they? Here we look at some key performance criteria between different types of Dark Pools.
We categorise them into 2 groups:
(a) Dark Pools owned by Brokers/Market Makers;
(b) Non-Bank Dark Pools, which are largely owned by exchanges and ATS/MTF’s.
Our dataset includes over 10 million executions since the start of 2020. Here we focus on 3 main performance criteria, Trade Size, Spread Capture and Price Impact/Adverse Selection.
Key Points:
• With the exception of Dark Pools explicitly targeted at Blocks (BIDS, TRQM), average execution sizes are similar for Broker/Market Maker and Non-Broker Dark Pools. There is, however, a greater probability of getting a larger fill in a Broker/Market Maker Dark Pool.
• Spread Capture tends to be higher for Non-Broker Dark Pools at around 50% indicating predominantly mid-point executions. Broker/Market Maker Dark Pools tend to have lower spread capture as crossing the spread to access liquidity is more common.
• Price Impact is larger for Broker / Market Maker Dark Pools. This is especially pronounced in trading less liquid stocks. With a few exceptions, adverse selection is also greater for Broker/Market Maker Dark Pools.
One of the most important metrics in evaluating execution quality is the size of the execution. The larger the execution the sooner the trade can be completed minimising the risk of the price moving away from you. Examining Figure 1, we see that most of the dark pool venues have similar fill sizes. The Broker/Market Maker dark pools appear to have slightly larger average fill sizes, if you exclude two dark pools that are specifically focused on Blocks – BIDS and TRQM.
One needs to be careful when drawing conclusions from averages. When we look at other metrics such as the median and the distribution of fill sizes (see Figure 2), we can refine some observations. The Box and Whiskers chart in Figure 2 shows the median, the 25th percentile (the Lower Hinge) the 75th percentile of trade sizes (the Upper Hinge) and a criterion for being an outlier – the Upper Whisker, which is 1.5 times the size of the range from the 25th to 75th percentiles.
Figure 2 shows that the median fill sizes are similar, but the 75th percentile for Broker/Market Maker executions and large outliers is significantly larger. In 25% of executions a Broker/Market Maker fill will exceed $9,375 compared to $7,468 for a Non-Broker Dark Pool.
The chart also suggests that there are a relatively small number of outliers with very big execution sizes that account for the larger average fill sizes on BIDS and TRQM. While the averages for these two venues are much larger than other venues, the median and the distribution of costs for BIDS and TRQM are in line with other venues. Among all dark pools, JPMX has the largest execution sizes with both the highest 25th and 75th percentiles.
In addition to fill size, another key component of execution quality is spread capture. We are not making a judgement as to whether a higher spread capture is always preferable. That depends on the overall trade strategy and alpha model. In general, though, it is preferable to minimise spread cost. Figure 3 shows that spread capture is higher for Non-Broker Dark Pools which tend to be around the mid-point. Some of the Broker/Market Maker’s have considerably smaller percentages of spread capture resulting from crossing the spread to access liquidity often from internalisation. Of particular note is CDED, where the spread is almost always paid as part of the market making function.
We will look at one more metric which is information leakage after the execution. We measure the movement of the mid-quote after each fill from 1 millisecond onward in basis points and control for the execution either adding or removing liquidity by taking the absolute value of the change in price. The larger the change in price post execution, the greater the information leakage.
Overall, there is more price impact in the Broker / Market Maker compared to Non-Broker dark pools. After 25 milliseconds the price has moved more than 1 bps post execution only one Non-Broker Dark Pool (LEVL). In comparison, the price has moved post execution more than 1 bps for BARX, CAES, SGMT and UBSA among Broker Dark Pools.
These are only some of the criteria that can be used for evaluating execution quality, but it suggests that there are significant differences between types of dark pools in how they perform. Broker / Market Maker dark pools may get you larger trade sizes and be better in sourcing liquidity when there is urgency to complete the order faster. The trade-off is that you may pay a higher price due to lower spread capture and additional price impact.
©BestExecution 2020
