Dr Darren Touslon, founder, LiquidMetrix
MiFID 2 has introduced some key structural changes to the equity trading landscape. Chief amongst these are the closing of Broker Crossing Networks (BCNs), their partial replacement with Systematic Internalisers and the Double Volume Caps applied to Dark Trading.
The broad aim of the new regulations seems to be to force smaller orders to execute on lit venues and for ‘dark’ trading to be reserved for larger‘block’ type orders.
Block trading has always been appealing to Buy Side participants. It offers the chance of getting large orders filled relatively quickly at a fair price with a ‘neutral’ counterparty. The lead-up to MiFID 2 saw the opening of new block trading venues (CBOE LIS, Euronext Block) as well as significant innovations on existing block platforms such as Turquoise Plato Block DiscoveryTM, Liquidnet and ITG POSIT.
Assuming block trading does indeed take off during 2018,the first challenge participants will faceis how, when and where to access this new liquidity. An additional challenge, created by MiFID 2’s more stringent Best Execution rules (RTS 28), will be to measure and quantify the benefits of any use of Blocks.
So how should we measure the performance/benefits of block trading? LiquidMetrix and Plato Partnership have formed a collaboration to study a large dataset of anonymised block orders (taken from the Turquoise Plato Block DiscoveryTM platform) and come up with innovative ways of extending traditional TCA to encompass modern block trading.
We will be publishing an in-depth study of our findings later in the year but here present some preliminary results.
The Data Set
The dataset we report on here consists of several million orders – with a filled value of €9.5bn- that were placed on the Turquoise Plato Block DiscoveryTMplatform over a 12-month period in the run up to MiFID 2 from September 2016 to September 2017.
The Good
We started by measuring execution prices versus market mid-price at arrival for all orders that filled. We found that the execution prices were,on average, very close to arrival prices with just under 1 BPS of Implementation Shortfall. To put this into context, using LiquidMetrix’s peer-based pre-trade estimates, the expected costs of executing these same orders in similar market conditions algorithmically would have been around 20 BPS. So,block orders executing on Turquoise Plato Block DiscoveryTM save an average of 19 BPS in IS cost.
The Bad
The Achilles heel for block venues is fill rates. For the orders in this study about 95% of orders were unfilled(this rate dropped markedly over the 12-month period as volumes doubled in the pool but was still >90% in September 2017).
At best, unfilled orders represent a ‘no benefit’ scenario, diluting the 19 BPS of filled order improvement to 0.3BPS when value weighting over filled/unfilled orders.
We also found that whilst orders were resting,market prices tended to drift on average 9BPS in an unfavourable direction (i.e. up for a Buy order). This represents a large potential‘opportunity cost’ on the unfilled orders, outweighing net benefits of filled orders.
However, a recent innovation of block venues is the ability to place conditional orders.Such orders allow participants to get the best of both worlds by placing an order on a Dark Pool and at the same time placing orders on other lit/dark markets. In this case, the 9BPS ‘delay cost’ may not be applicable as resting on Turquoise Plato Block DiscoveryTM does not restrict other trading options, so the 19 BPS improvement received on filled orders in the pool can simply be viewed as a ‘bonus’ (or a missed opportunity if you are not using the Pool!).
The Ugly?
No one likes getting gamed or leaking information.
Participants can detect signs of gaming by examining market prices around the time they fill.We can checkhow ‘fair’matched prices are at the time of execution (reference price gaming) and to what extent prices move in unfavourable directions after fills occur (information leakage).
We found that execution was generally clean with EBBO spread capture close to 50%, very few trades falling outside EBBO and just 19% market movement caused by fills.
Bringing it Together
For the orders we studied,the benefits gained by accessing the pool were about 19 BPS for every filled order and an average of 0.3 BPS over all orders (filled or unfilled) that were submitted,assuming that participants were taking advantage of trading alongside their resting dark orders. There was little evidence of gaming or information leakage around the time trades occurred. Over the 12 months of the study fill rates doubled.
Conclusion
MiFID 2 might lead to more block trading. If it does, it will be important for participants to measure the benefits (or not) of trading blocks. Our study is aimed at developing a toolkit summarising participation rates, IS costs, opportunity costs, fill toxicity and comparison to algorithmic trading that will help participants quantify these benefits.
Our initial analysis shows that when fills occur on Plato, the benefits are significant with the caveat that as fill rates are relatively low,it is important to understand any opportunity costs relating to resting orders and if possible, take advantage of conditional order types to minimise delay costs.
* LiquidMetrix is a division of Intelligent Financial Systems Limited, a Strategic Insight company.