The ELP SI Debate

As the industry continues to adapt following the implementation of MiFID II, Hayley McDowell sits down with Ben Springett from Jefferies, Matthew McLoughlin of Liontrust & Jonathan Finney from Citadel Securities, to discuss how the buy-side and brokers are interacting with market maker systematic internalisers.

Hayley McDowell: Almost a year on from MiFID II implementation, how have you found the systematic internaliser (SI) landscape evolving from an execution perspective?

Ben Springett [head of European electronic and program trading, Jefferies]: We have a more holistic view of how the execution landscape has evolved because it is not possible to determine the impact of the SI landscape in isolation. How you would use SIs is partly down to which avenues you have available to execute at a certain point in time (volume caps, order size etc.) as well as what other liquidity is available in the market at that time. However looking solely at SIs, it’s clear that there are two very distinct flavours.

Ben Springett, head of European electronic and program trading, Jefferies

First of all, you have the broker SIs which have replaced the broker crossing networks (BCNs). I would say that functionally we use broker SIs in the same way we used to use the BCNs, and I think the initial focus from operators has been on replicating historic matching capabilities in the new regulatory landscape, often times utilising a combination of an SI and a periodic auction on a member preference-basis. The second type of SI is the market maker, or electronic liquidity provider (ELP) driven SIs, which are certainly a new dawn for us on the European trading landscape, albeit the liquidity provider operators were active in multi-lateral venues previously. These single dealer pools have been prevalent in the US previously, and as they have come into Europe they have elegantly met the definition of an SI as the legislation would expect them to look.

We’ve had almost 12 months of ELP SI usage in some cases, and it’s clear that not all are created equal. There are different firms, offering products with different strategies behind them that you should view differently. When you go to an ELP SI you need to have a clear view of how they operate, and we have been able to build up a meaningful amount of data on the variety of ELP SIs that we currently use, which we use to inform our future usage. Part of that is initially determined by how the operator describes their model to us – maybe they propose to have low short-term impact because they have a longer holding period or a natural risk offset available to them from other strategies. In other cases you expect to have more short-term impact because they are very much a short-term market making operator, but until you see the data, you don’t really know.

So, we have been building a dataset across a number of providers and that has informed our current adoption levels along with consultations with clients. Over the course of the past few months we have seen an increase in the number of clients that are opting in to use these services, but Jefferies’ default position is to have them switched off for clients. I think that a level of maturity and understanding has been reached, and people have had the time to digest and we’ve also got more data to share with our clients.

Matthew McLoughlin [head of trading, Liontrust Asset Management]: We have a variety of fund types and trading strategies, so we have used a number of the new and pre-existing venue types so far this year. I have been pleased with the innovation that we have seen from both brokers and execution venues. In aggregate we are reasonably happy with our experience from liquidity and reversion perspectives, although we have been quite selective with the venues we have interacted with. We really tried to understand each venue in terms of what they offer and the model behind that before we switched them on for trading. In general the buy-side is now comfortable using conditional or large-in-scale (LIS) order books and, now that we have more information about them, periodic auctions for orders below LIS too. Systematic Internalisers, however, are very different beasts and it is taking much longer for the buy side to get comfortable interacting with them.

Bank and ELP type SIs are obviously very different from each other and within each of those categories there are also distinct differences. Within the ELP category you have SIs which make markets off the back of their own strategies, others that make markets off the back of their exchange-traded fund (ETF) business, and then the more high-frequency trading (HFT) type SIs that don’t want to hold their risk for long at all and look to unwind reasonably quickly. You’ll see different prices, reversion rates and different sizes from each of those.

It’s important to understand that it’s not simply a case of “SIs are good” or “SIs are bad” and “an ELP SI is good” or “an ELP SI is bad”. You need to understand what their models are, what they offer, how they are different and which best suits your trading strategy. We met with all of the ELP SIs that we have switched on as well as ones that we decided we didn’t want to interact with due to them not providing the type of liquidity that we wanted to access. We tried to understand their models as best as an outsider ever can, because obviously there is a lot information that they don’t necessarily want to share with us. We obtained as much data as possible from the SIs themselves, but also from brokers who had already interacted with them both in their SI format, but also when they were operating within their BCNs last year. A difficult component to this is that if you just use broker data on SIs, then one ELP SI could look very good according to one broker’s data, but then look very different on another broker’s data due to the way each interact with their liquidity. That’s difficult for the buy-side to manage, and that’s why we try to get as much data from the SIs themselves as well so that we can question the brokers on their smart order router logic.

We have interacted with a number of SIs this year, both bank and ELP style. Now that we have our own data set we continue to analyse each liquidity source as this is a constantly developing picture. The data will ultimately decide which of the venues thrive and which cease to exist. The more data we get, the easier it will become for the buy side to select SIs to interact with.

BS [Jefferies]: I think it’s important to understand that this isn’t a static landscape. As one comes to an opinion based upon a selection of data, that doesn’t necessarily mean that particular experience will hold in the future. I think you see changes in individual participants or destinations that arise as the market evolves. In times of volatility, for example, you will see some changes in strategy and the providers may begin to model more on those that are interacting with them. So market participants are feeling each other out in a way to try and understand and then responding. With that in mind, it can be difficult to make static decisions around the use of SIs in general, or individual providers.

At Jefferies, we have scoring mechanisms that we apply to each trading venue we trade with – that can be ELP SIs, broker SIs, even lit order books – for a dynamic quantitative model that ranks each venue in real-time. When it comes to the time we need to trade, we know from our short-term experiences what to expect in our outcome. We have actually seen a number of the market maker pools that we trade on drop below some of the regular lit markets in terms of their relative priority. We aren’t triggered or incentivised by the fact that it’s free to trade on these venues for brokers, and we don’t have a cost logic in our routing process. When we are trading aggressively, we are only looking at where it is likely we will get the smallest amount of price impact, and we see quite big differences between the market maker destinations, which as mentioned we also compare with lit markets. A small element of price improvement from a market maker won’t be enough to tempt us, we really care about the impact that could manifest from that interaction on an ex-post basis.

Jonathan Finney, director, European business development, Citadel Securities

Jonathan Finney [director, European business development, Citadel Securities]: I would emphasise Ben’s point there. Not all brokers are the same and not all market makers are the same, so we see very different types of evaluation and results. What Jefferies is doing in collating and analysing data and evolving behaviour based on the results of this analysis, represents one end of the SI evaluation spectrum. With regards to how this landscape has evolved over the past year, overall volume does not appear to have changed dramatically. However, the composition of liquidity sources certainly has changed. To that point, SIs are not the only alternative source of liquidity but they represent the single greatest opportunity for evaluating equity market makers. Given the bilateral nature of the SI, ironically this has achieved the transparency that MiFID II was aiming for. Matthew at Liontrust, via Ben at Jefferies, knows that he is trading with Citadel Securities and he can evaluate us on our merits.

I think people have been surprised at how slow ELP SI growth has been but we are definitely seeing the buy- and sell-side becoming more comfortable interacting with market makers as they are able to access more data. I would point out that there are only four market makers currently reporting their monthly data to Rosenblatt Securities and TABB Group and having this data available supports the evaluation process. Market makers are very different. For example, Citadel Securities specialises in risk-warehousing positions, larger quote sizes, benign market impact and a diverse security universe. Having a robust evaluation framework is important to assessing these factors. 

MM [Liontrust]: On the issue of transparency, if we had decided not to interact with SIs, then I would actually be trading with them on lit exchanges anyway, possibly at worse prices, smaller size and potentially with a larger market impact. That is up for debate, but when I interact bilaterally, I know exactly who I am trading against so that I can hold that venue to account.

HM: What have you found to be the most important factors when evaluating ELP SIs?

MM [Liontrust]: Firstly I need to decide if their model makes fundamental sense and if it is adding liquidity that would be beneficial for me to interact with. I then look at price, size and market impact – those three factors are key. For the buy-side, I think market impact and size has become a lot more important this year. I’m not looking for a twentieth of a tick price improvement, I’d rather have larger size and lower market impact. It’s strange to say it as a trader who is addicted to getting the best price – but the best price is not always best execution for me. If I can get good liquidity and not impact the market, then I am usually a happy bunny. The importance of those two factors has definitely increased over the past couple of years for the buy side in my opinion. Fill rates are another factor to be aware of. I hear of SI fill rates between around 95-99%, but what happens when you don’t trade against them? That’s harder for the buy-side to monitor. One thing is for sure, with the proliferation of execution venues that we have experienced, it is more important than ever to make informed venue selection decisions.

BS [Jefferies]: Jefferies doesn’t have an execution process that tries to minimise the cost to us when doing business for our clients. So for us, the consistency of availability of liquidity is not a crucially important factor, but what is important is the size of liquidity when we need it. We actually don’t consider the marginal price improvements offered to us on the quotes we receive from market makers, and that’s because we are not looking for them to create an ability to synthetically jump to the front of the queue by a tenth of a tick-type price improvement. We completely ignore that. It doesn’t even compute in our system. We are sensitive to the size of liquidity that is being offered to us, and the price impact we will see as a result of that execution.

When we’re talking to the different providers, we are trying to understand factors like their holding periods, strategies and unwind models, and that has helped us to predict the availability of liquidity in size and what that price impact will look like based on what they are describing. By connecting to a variety providers, not ones that are very similar to each other, we also have a diverse range in terms of the universe of coverage. One provider covers more of the small and mid-cap range in continental Europe, an interesting area for us to explore, and other providers have their primary business in cash equities, ETFs or options, and the flow they have in cash is made in the unwind of hedges and those types of trades – but again, it’s a different profile. We’re looking for that complementary nature of the universe that comes together. But ultimately we go to providers and say, “show us more size – I don’t care if you’re not there as often, but show us more size”. Orr show us mid-point liquidity, because then it gets very interesting as we can connect that to a range of additional strategies that are taking advantage of mid-point liquidity elsewhere, from periodic auctions, dark MTFs and so on.

MM [Liontrust]: You have to give yourself liquidity options. We must remember that not all execution venues are suitable for every order type. No one venue type ticks all the boxes. SIs will admit themselves that they are not suitable for every order type, but they are there for when you need them. At every buy-side event I attend, the main problem we all come up with is finding liquidity, so having a variety of good quality liquidity options definitely helps.

JF [Citadel Securities]: The factors that Matt and Ben have covered there broadly fall into three categories: unique liquidity, quantity of liquidity and quality of liquidity. Unique liquidity comes up a lot in terms of what clients are looking for in an ELP SI, and this means offering liquidity across a large and diverse securities universe. This is an easy one to quantify because the information is available on the quote streams. With quantity of liquidity, it is via quote sizes. To Matt’s point, we are not going to be two-way in every single stock all the time in all sizes, but what we will do is offer a quantity of liquidity that matters to the buy-side. It all boils down to those three and making sure that brokers and the buy-side are aware of the different ELP SI models. Citadel Securities obviously only represents one type of ELP SI, but there are other types of SIs out there as well.

Regarding quality of liquidity, measuring the price impact following an SI trade remains the key focal point for many discussions. Ideally, the focus would be as an A-B test per market maker, but realistically this would not be practicable. Focusing on fill level analysis would help the buy side and sell-side determine which market makers unwind positions quickly and which ones risk warehouse positions. One point worth noting is that we are talking here from a Jefferies and Liontrust perspective, but brokers are evaluating these factors very differently depending on their interactions with market makers. It remains difficult for the buy-side to become aware of what a good evaluation and a bad evaluation actually looks like, so again, the more we can talk about it the more we can move into a standardised framework for evaluating these factors.  

HM: What are your thoughts around SI price improvements sub-tick above SMS and the European Commission’s recent take on the tick-size regime for SIs?

BS [Jefferies]: We do not factor marginal improvements in, it doesn’t mean anything for us. We believe it could potentially create the wrong incentive, and I think risks in efficient market structure. In Europe a few years ago when we saw the dawn of fragmentation in lit markets, we saw tick-size competition taking place between primary exchanges and alternative venues, but I think the industry did a very good job of self-policing that and coming to a harmonised tick-size regime. MiFID II of course has a harmonised tick-size requirement for trading venues, so the industry is under regulatory pressure to conform with the regime more widely. I think potentially antagonising the regulator by offering fractional tick price improvements is not very productive.

Matthew McLoughlin, head of trading, Liontrust Asset Management

MM [Liontrust]: From the buy-side’s perspective, a tenth or twentieth of a tick price improvement is not something we are looking for. The regulators have been looking at extending the tick size regime to cover SI flow, so price improvement could soon be a thing of the past. There is a danger that this gets extended to LIS venues and periodic auctions, which would prevent a large amount of mid-point trading. This is definitely not in the best interests of investors. I’m pretty comfortable with the tick size regime being expanded to include SIs, however, because it’s not a driver as to why I’m trading with them in the first place.

BS [Jefferies]: Yes, it’s very important as an industry that we protect being able to trade at mid-point. It’s a very valuable option for a range of different venues that use if it for a wide range of market participants, and is well established as a go-to option for many of the largest asset managers across the Continent.

HM: How do you see relationships between brokers, the buy-side and market makers evolving in future?

JF [Citadel Securities]: I would say the SI regime has achieved one of the key goals of MiFID II, that being increased transparency albeit on a bilateral basis. For the first time, the buy-side and sell-side can isolate, identify and ultimately evaluate each individual market maker on its own merits. Going forward, this provides a sort of three-way feedback mechanism where buy-side can evaluate their brokers and market makers by looking at aggregate SI statistics. This valuable tool will complement TCA and other capabilities for evaluating execution quality in equities and I am genuinely excited to see how this develops.

Over the course of next year, we are likely to see an increase in the number of market makers moving into the SI regime and an increase in SI volumes. Then, maybe towards the end of 2019, I believe we could see a drop-off in the number of market makers because the evaluation framework will be a lot stronger by that point, making it easier to identify and weed out weaker performers.

BS [Jefferies]: I would say that with MiFID II this year, we’ve seen a vast amount of technical and innovation work which will further the gap that exists between different providers. If I think about the sell-side, those with nimble technological architecture that can respond quickly to changes in the liquidity landscape or onboard new venues, are more likely to succeed than those who can’t keep up or their technology is too old and will fall by the wayside. I think that’s likely to continue as a trend over the course of the next few years. We’ve seen clients increasingly applying more quantitative and rigorous tests to their broker selection processes and their trade distribution processes for example.

In terms of the bilateral interactions that exist between the sell-side and the market maker, the transparency is important, but where I’d like to see more transparency is in the tape. The consolidated tape was a much desired aspect we wanted to be included within MiFID II, but as the regulation became more complex that didn’t happen. There’s a role to be played for the industry itself or the regulator to provide a consolidated tape solution, and what I’d like to see as part of that is more granularity around the SI prints that take place. When we trade against an SI we know with which provider the trade was done, and potentially the type of liquidity we have interacted with. We will pass this information back to our clients so they also have that transparency. What we lack though is the ability to identify from the tape generic SI prints when we aren’t involved in the trade. Which venue did they take place on? Is it genuine liquidity or more of a manual or technical trade reporting function? A greater level of granularity would be beneficial in identifying the relative location of liquidity, size of liquidity and the position of the market as a whole.

MM [Liontrust]: From a liquidity perspective, I would love for ‘perfect’ liquidity solutions to land in my lap, but in reality that’s never going to happen. The market structure is constantly evolving with new regulation and execution venues developing constantly. I think that without the full engagement of the buy-side, with the sell-side and the venues themselves, it will be much harder and take a lot longer to drive innovation in the right direction for the industry. Never before have I witnessed the level of collaboration between market participants than over the last couple of years. I only see that developing further in the years to come.