The arrival of the Markets in Financial Instruments Directive 2 (MiFID 2) will fundamentally reshape the European financial markets landscape. The motivation of the European Commission and regulatory bodies was the attempt to bring the benefits of the equities space that were imposed by MiFID to other asset classes and current OTC trading while minimising two negatives aspects: the market fragmentation and the increasing use of dark pool trading with an estimated trading volume of 9% in the EU and 12% in the US (Jaccard, 2016 - see Automated Trader, Issue 40, page 46).
This led to maybe one of the most controversial rules in MiFID 2, the double volume cap (DVC) mechanism for dark pools. With RTS 3 (MiFIR Article 5(9) and 22(4)), MiFID 2 aims to limit dark trading by introducing double volume caps on two of the pre-trade transparency waivers: the Reference Price Waiver (RPW) and the Negotiated Trade Waiver (NTW).
The cap on trading per venue limits the percentage of a single financial instrument that can be carried out on a single trading venue under these waivers to 4%. The cap on dark trading across venues limits the total amount of dark trading of financial instruments across the EU to 8%.
As clarified in the regulatory technical standards (RTS) (ESMA, 2015) the caps will be based on data gathered starting from January 2017 and will be updated on an individual stock basis over a twelve-month rolling window. If the relevant cap is breached, the trading of the respective stock will be suspended from either the respective dark pool or all dark pools, depending on which cap is triggered. The caps only apply to transactions executed on multilateral trading venues (such as RM, MTF, OTF) but not to OTC or SI transactions.
Impact of the Double Volume Caps - The Rosenblatt Study
DVCs are considered as a very 'clunky' cap mechanism and there is little information why 4% and 8% were chosen by the regulators and how this calibration was derived. Carlens and Higgins (2016 - see Automated Trader, Issue 41, pages 27-31) show that under certain circumstances and due to the rolling calculation window the DVC can even lead to a suspension of two consecutive six-month periods, totalling a one-year suspension.
In a recent paper by Rosenblatt Securities (Puaar, Schnack & Kemmsies, 2017), which caught the interest of the trading community, the authors presented a quantitative study on historical trade data to investigate the effect the caps will have on dark trading. In some of their key findings they showed some fairly dramatic numbers on the future of dark trading:
The 8% (EU-wide) cap was triggered for 74% of the stocks under consideration.
UK and Irish stocks were most affected with 89% for FTSE100 and 85% of Ireland's ISEQ.
The market-wide cap (8%) seems to be the dominant cap, as only two stocks which exceed the venue-specific cap (4%) did not breach the market-wide cap in addition. Also, 16% of the stocks under consideration had an EU-wide dark market share of larger than 16% and would therefore be subject to a back-to-back ban and thus be suspended for twelve months in total.
Figure 01:Market shares and estimated LIS trading volumes for selected European dark pools (by traded volume in EUR, April and May 2017)
Mitigation of the DVC by optimised LIS trading?
As outlined in Puaar et al. (2017), Appendix 1, some of the dark pools may have trade sizes that qualify for the large-in-scale (LIS) waiver. This is illustrated by an analysis of market share and estimated LIS trading volumes for a set of selected European dark pools (see Figure 01). Certain specialised LIS venues (e.g. Bats LIS) naturally execute under the LIS waiver. But a number of other dark pools show a certain percentage (as in case of SIX Swiss Exchange of almost remarkable 60%) of trades which would qualify for the LIS waiver but are currently executed under RPW or NTW.
These trades could therefore be executed without pre-trade transparency and without eating up capacity of the double volume caps. Puaar et al. (2017) decided to include all the trades that would be executed under RPW or NTW in their analysis to avoid underestimating the caps and thus deriving a conservative estimate of the impacted stocks.
Next, we investigate to what extend the using up of capacity under the DVCs and therefore the suspension of stocks from dark trading can be mitigated by optimised execution. In particular, we analyse where every trade which qualified for the LIS waiver would be executed on LIS instead of RPW or NTW.
Figure 02 shows the percentage of stocks breaching any of the two volume caps broken down by index. The stocks are assigned to indices provided by Bats Europe. If a particular stock is quoted on multiple exchanges, all listings are mapped to the given index. The calculation is based on a 365-calendar-day rolling window from July 2016 till July 2017.
The left graph of Figure 02 shows the percentage of the overall number of stocks that would have breached the DVCs (suspended stocks). We decompose it further into the percentage which would have been achieved by utilising all LIS waivers (where possible) and by executing everything under the RPW/NTW (the Rosenblatt approach). As an example, for the AT20 Index 45% of the stocks would be suspended from trading. With executing all possible large-in-scale trades under the LIS waiver, this number of stocks suspended from dark trading could be reduced to 30%. The overall percentage of stocks suspended from trading is fairly similar to the results in Puaar et al. (2017), confirming the analysis in general.
One of the biggest reductions can be achieved for the FR40, where the overall percentage of suspended stocks can be reduced from 67% to 21% with 'LIS optimisation'. Only a very small improvement can be achieved at the UK100 which also impressively shows the highest percentage of suspended stocks with about 93%, again confirming the results of Puaar et al. (2017).
Figure 02: Percentage of stocks per index breaching the double volume caps
In our own analysis, the volumes traded on SIX Swiss Exchange and Oslo Stock Exchange are excluded. However, all the instruments that have a primary market within the EU but are traded via SIX Swiss Exchange Liquidnet Service (SLS) are included in the calculation.
We can conclude that for the highly affected indices like the UK100, there is little improvement by LIS optimisation. Nevertheless, a significant percentage of the overall trading can still be executed without pre-trade transparency under the LIS threshold.
Table 02 demonstrates the usage of the DVC, both LIS- optimised and non-optimised (meaning most conservative) for the top 20 most traded stocks (based on traded volume in EUR). Again, an efficient usage of the LIS waiver can lead to a successful mitigation of the impact of the DVC. For example, if all trades of BNP Paribas which qualify for LIS were executed under the LIS waiver the cap usage could be reduced from 102.5% (suspended from dark trading) to 84.83% (still eligible for trading on dark venues).
Table 02: Percentage of usage of the DVC for the 20 top traded single stocks (by traded volume in EUR)
Conclusion: Potential Growth of LIS Trading Facilities?
Together with the Rosenblatt analysis our results confirm that the double volume caps will be heavily restricting the current usage of dark pool trading. Nevertheless, the regulatory bodies were aware of the need to protect large orders from adverse market impact and to avoid abrupt price movements, which can cause market distortion (which was the original intention of the dark pools). Large transactions which exceed the LIS thresholds can still be executed under the LIS pre-transparency waiver. These LIS transactions will not affect the DVCs on the RPW and NTW while still keeping potential market impact low.
Our analysis shows that a certain percentage of the trades currently executed in the dark would qualify for LIS execution. This differs widely from one index to another depending on local trading behaviour. LIS optimised execution could further avoid using up capacity of the double volume caps on the dark pools. On the other hand, the recent trend for executing even small trade sizes in dark pools is clearly restricted by the DVCs. This is in line with the clear intent of the regulatory bodies to move as much trading back to the lit book as possible. Escaping the pre-trade transparency by moving the trading to OTC or SI will be only possible in specific cases.
The aggregation of client orders should not be used for the purpose of creating a total parent order size that falls above the LIS threshold. Nevertheless, it is to be expected that there will be further growth in specialised LIS trading facilities. Market participants then need to recognise LIS classified trades, track market share and navigate interactions across different liquidity pools. A swift and seamless integration of LIS trading into the existing execution strategies of a trading firm will require the analysis of large amounts of trade data and monitoring of thresholds over a variety of different liquidity pools and trading venues (see e.g. Figure 01).
Puaar, A., Schnack, J. & Kemmsies, A. (2017). Shedding light on MiFID II's dark volume caps. Trading Talk Market Structure Analysis: Rosenblatt Securities
Tyfield, S. (2016). The delay (or not) of MiFID2 and the release (or not) of the commission delegated acts if MiFID 2. Automated Trader. Issue 39 (Q2) 24-25
ESMA (2015), Draft Regulatory and Implementing Technical Standards MiFID II/ MiFIR (Final Draft). Paris. ESMA Europa
Jaccard, G. (2016) Shining light on new dark pool regulations. Automated Trader. Issue 40 (Q3) 46-46
Carlens, H. & Higgins, D. (2016). The future of dark liquidity in Europe. Automated Trader. Issue 41 (Q4) 27-31
Calculation period approach
All metrics included in the Double Volume Cap analysis are calculated over a rolling window of 365 calendar days.
Calculation of RPW and NTW traded values
Currently, trades that correspond to the Negotiated Trade Waiver can only be identified based on trade conditions for the following venues: Bats BXE, Bats CXE and Bats TDM. Identification is based on the MMT trade condition classification Trade Indicator=Negotiated Trade. All trades that are flagged with trade condition "market mechanism dark" are classified as Reference Price Waiver trades.
Total traded value across European Union venues
The trade categories included in the total traded value are:Lit (Central Limit Orderbook, Auctions, Lit Hidden Orders)
IOI Negotiation, Block Trades
Negotiated Trades (for venues: Bats BXE, Bats CXE, Bats TDM)
Trade categories not included in total traded value:
Off-Book (On-Exchange) other than:
IOI Negotiation or Block for all venues
Negotiated Trades for venues: Bats BXE, Bats CXE, Bats TDM
Exchange and products coverage
Volumes traded on SIX Swiss Exchange and Oslo Stock Exchange are excluded from the analysis as they are not part of European Union trading venues. However, all the instruments that have a primary market within the European Union and are traded via the SLS service (which is included in the SIX Swiss Exchange feed) should follow the European rules on suspension of trading and therefore SLS volume is included in the analysis for these products.
Additionally, all products that have the Istanbul Stock Exchange or the Warsaw Stock Exchange as primary market are excluded from the analysis.
Large-in-scale qualified traded value
Discovery of trades where the traded value in Euro is above the large-in-scale thresholds is based on Bats Europe reference data files available on the Bats Europe website: https://www.bats.com/europe/equities/support/reference_data
In order to assign the appropriate threshold, mapping between products is done based on LocalCode/Cusip and currency for MTFs. For standard exchanges thresholds are matched by ISIN and currency rule. If a particular product is quoted on multiple standard exchanges, we map all listings to a given threshold.