The Gateway to Algorithmic and Automated Trading

An AXA to grind

Published in Automated Trader Magazine Issue 28 Q1 2013

You might think Christophe Roupie already has enough on his plate. He runs the trading operation for one of the biggest asset managers in the world, with a 55-strong team in London, Paris and Connecticut and he has plans for a presence in Hong Kong. But he is also keen to spend more time in Brussels. As he tells Automated Trader Editor Adam Cox, the buy side can do a lot more to work with regulators and ensure more sensible rules are put in place.

Adam: What would you say is the overall philosophy in terms of your trading strategies?

Christophe: We're here to execute our fund managers' and clients' investment decisions by going to the financial markets and trading their orders. Therefore, and I know it sounds very simple and very basic, our main objective is to minimise market impact whilst providing best-in-class execution in a sound and risk-controlled operational framework. Our trading strategy is driven by our multi-expert asset management model, meaning that market intelligence and brokers' interests are concentrated within trading before being directed to the relevant investment teams on a global scale.

We have also developed tools and invested massively in systems which allow us to stream real-time information to our fund managers and clients across all asset types. We always try to be the most efficient conduit for liquidity and market information to our clients.

We trade in all currencies and markets - equities, ETFs, sovereign, credit, emerging and high-yield bonds, convertible bonds, FX, money-market and treasury instruments as well as listed and OTC derivatives (CDS, IRS, ILS). I am also in charge of securities financing, which encompasses repo and securities lending, giving me a fairly wide spectrum of product coverage.

Christophe Roupie, AXA

Christophe Roupie, AXA

Adam: When you say your clients, you mean the fund managers.

Christophe: Globally yes, our clients being the fund managers. But we also consider clients all the stakeholders within AXA IM or outside the company who are supporting our activities and ensuring we meet our objectives.

Adam: Maybe we can talk about the tools and systems you mentioned.

Christophe: Minerva, our order management system, is at the core of our execution process. In terms of risk and control, we want to ensure that when looking at an order from creation to execution and matching we have as much STP as possible. To keep it fairly high level, Minerva is linked to the decision-making tools of our fund managers and clients, meaning that all the orders that we receive follow a very strict process whereby they are initiated by the fund manager, pushed through a constraint server and then sent electronically to the traders. The traders will then choose the way they interact with the marketplace, either via voice or using FIX technology to release orders to brokers or directly to e-trading platforms.

For example 80% of the orders received in FX were traded electronically in 2012, and that equates to roughly one-third of our volumes. It represents a 15% increase in the number of tickets and a 42% uptake in volumes from the previous year. Overall, we traded over 1.5 trillion euros across all assets in 2012.

Adam: Okay, so the scale of your trading must be a big factor. Not many firms are initiating these kinds of flows. If you're aiming to reduce market impact, how do you go about it?

Christophe: The numbers I've given you so far are for FX, which is a very competitive market. A lot of the tickets can be processed via electronic platforms and the market impact is therefore negligible. What we trade electronically usually has limited market impact. Anything above a certain size would be traded via voice, or any currency which is less of a natural fit for e-trading. We still take the view that for important transactions - regardless of the asset type - slippage can be better managed when conducted by voice. Again here, this is the job of the traders to make that choice and understand where they can go to trade at the best price or obtain the best possible result for our client, and also keeping in mind that the overarching philosophy from a risk perspective is that all of our business is driven by counterparty risk and counterparty limits we must comply with.

Risk control is at the heart of our investment process and also at the heart of our execution process. I'll give a very good example: we don't do DMA. We don't think that we can actually fully control the risk which is embedded in having direct access to exchanges. The traders are our first level of control and they also rely on brokers to perform a second level of control or oversight. And I have to say that I feel more comfortable having that 'safe pair of eyes' approach rather than running the risk of an execution going wrong because of a fat finger which could have a massive market impact and of course damage our investment performance.

“We probably have hundreds of strategies at our disposal today... But the traders tend to stick to a few with which they feel extremely comfortable with. The names do give it away though, like Iceberg, Dagger, Sniper or Stealth!”

"We probably have hundreds of strategies at our disposal today... But the traders tend to stick to a few with which they feel extremely comfortable with. The names do give it away though, like Iceberg, Dagger, Sniper or Stealth!"

If we look at fixed income globally, which is a large part of our AUM and consequently represents a large share of our trading volumes, in 2012 we traded half of our tickets electronically for roughly the same split on volumes - that's a shift we've seen every single year since 2005, a gradual pickup in the volumes traded on platforms.

And on equities, our use of algorithmic strategies has steadily increased to reach 25% of our volumes today from 5% a few years back. We use a combination of low and high touch trading to manage market impact, including program trades and crossing networks. The rest is still executed by voice and as a matter of fact we very much encourage our traders to keep a high degree of interaction with their sales coverage. From an STP/FIX standpoint close to 100% of our orders are confirmed and allocated electronically.

Adam: Why the focus on the electronic communication if ultimately trading is voice-based?

Christophe: When we talk about controlling the operational risk around the activities, what is key is that as soon as the orders are traded by our brokers, we send the allocation electronically, meaning that the broker is actually able to see and to confirm the operation which has been traded, the size, price and details of the financial asset they're willing to buy and sell, and that information which is sent to brokers will also come back electronically in our front office systems, including those of fund managers.

Adam: So it just makes everything neater and easier to manage.

Christophe: Exactly. The process is quite complex because one or more custodians are also in the loop. But we want to have as much STP as technically possible, meaning that once you actually execute the trade even if you negotiate by voice, you want to release the economic terms of the transaction electronically to your broker, in order for the broker to confirm that the trade has been acknowledged, recognised and allocated, on a fund-by-fund basis before hitting the custodian. That information will be reconciled and filter back into our fund managers' system so they're actually able to see that the order has been duly executed and their holdings duly updated. For us it is critical to have as neat a process as possible and the level of confidence and comfort we get when the information is actually conveyed electronically and not via voice is very high.

Adam: You mentioned that algo trading on the equities side jumped from 5% to 25%.

Christophe: That's what we've experienced in AXA IM over the last two or three years.

Adam: That's obviously a big move. What's behind that?

Christophe: Minerva had been successfully implemented in all trading locations and across most asset classes including equities. We decided then to look at the benefits of using an execution management system - EMS - for equities. The reasoning behind using an EMS was to give us better visibility, better control and oversight of our executions via enhanced applications, and also to outsource some operational duties like broker network maintenance and API integration.

Our provider TradingScreen is a gateway to access numerous algorithmic strategies, as the core of its business is to offer a global network of brokers and their tools to its client base. I would say that's the main reason behind the increased use of algorithmic trading for my teams. It's the fact that we provided them with more tools via an EMS. It was a big change in the way we manage our order flow.

We probably have hundreds of strategies at our disposal today given that brokers are either rebranding similar techniques as part of their marketing exercise or building new ones, or even offering to design our own strategies. But the traders tend to stick to a few with which they feel extremely comfortable with. The names do give it away though, like Iceberg, Dagger, Sniper or Stealth!

At the end of the day we try to interact with dark pools in the best possible fashion to minimise market impact.

Adam: And the overriding goal is execution, not alpha generation?

Christophe: Once the order has been created and hits our systems it is "good to trade", and we must execute at the best possible price taking into account various best execution criteria including, for example, the likelihood of execution or likelihood of settlement. The fund manager is accountable to his client for fund performance and outperforming its benchmark; we of course must execute at a price or spread which is closest to the price or spread used by the fund manager when simulating the investment strategy. We want to preserve alpha and if we can help generate alpha we'd be more than happy to contribute to the performance of the funds. To this end we do use algorithms but we also use pre and post-trade analytics, for example transaction cost analysis, as I think most of our peers actually do, to benchmark our execution. For equities we use two benchmarks: one is interval VWAP, the other is a liquidity-adjusted implementation shortfall benchmark. For fixed income we have developed an in-house TCA tool which is also adjusted for market liquidity. Our FX offering is still being fine-tuned. But overall the results are positive and outperforming our benchmarks.

We have very active interaction with our fund managers via regular meetings and calls. We send trade ideas from our brokers and also make our own recommendations. We are not advising them on their strategies, but we highlight what we feel are good ideas and I guess that's something which is very difficult to translate into a process. But in terms of governance, we do discuss every month the trading flows and the activity with our key stakeholders and we'll be going through the numbers and analysing various KPIs agreed with them.

Adam: So if you felt it was a good time to get more aggressive or more cautious, you would be poring over the data and discussing it directly with the fund managers and stakeholders?

Christophe: The ultimate decision to buy or sell is left with the fund manager. Once the order is coming through the order management system, we have full discretion in the way we execute that order. But of course it might differ between traders and fund managers; it's a very sensitive discussion sometimes, depending on the seniority of the fund manager and the seniority of the trader. But overall the discussions are always very positive and constructive.

I will illustrate this approach with our fixed income traders and the sovereign crisis in Europe. As one of the largest asset managers in the world, it is conceivable that this crisis impacted our funds and created some flows. This information was extremely sensitive and the potential for market impact was real. The traders knew where liquidity could be found or how brokers were positioned, and it was shared with the fund managers.

We had to drive and manage our execution in such a way that there was no leakage of information to the Street in order to minimise or absorb market impact. And that philosophy and approach applies across all asset types, even more so when markets are stressed.

Christophe Roupie, AXA

Christophe Roupie, AXA

Adam: So your goal is to get those traders the best tools so that they can perform, and then over time watch that and manage that.

Christophe: Exactly. Meaning that we did and still invest a fair amount of money in systems and once in place you have to have the appropriate reporting of those flows. I would say it's taken us a good two years to build an appropriate reporting tool across all asset types and that means we can easily report activity internally and externally or conduct analysis on the activity. For example, the information I have given you on FX, fixed income and equities is directly extracted from our front office reporting system. I think what is key to any business is transparency. Once you have a process in place you can control the flows in and out of this process and report on those flows or make strategic decisions based on them. I believe that's actually a key component to the regulatory discussions in Brussels - the importance of trade repositories whereby all information is centralised and can, for example, be used by regulators if need be.

Adam: You've been stressing that under your watch there's been this increase in electronification and there's a big focus on STP. How far along do you feel you are in process? You were quoted in 2010 as saying that STP wasn't quite possible for all the markets you traded in. Are you where you want to be, or how much more work is to be done on that?

Christophe: Back in 2005 when I joined AXA IM I was able to demonstrate the positive impact of STP and more importantly the benefits it gave us in terms of best execution and trading capacity - i.e. headcount. I think we're probably in the stage where we're fine-tuning our model. We are very comfortable with the fact that our existing execution process, including bespoke and in-house developments, can be duplicated in any location including very fragmented markets like Asia, where you still need to invest to stay competitive.

The next stage for us is to maximise what our EMS can offer us in terms of analytics and trading strategies across more asset classes. We need to explore and understand how far we can go down the route of STP. I think that 100% will never be reached, due to the fact that some brokers are not set up from an STP perspective, but what that represents in trade volumes is negligible. Again, STP has to be looked at across the life cycle of a transaction, from order creation to order execution, trade settlement and trade matching including fails management.

Adam: Can you talk about how much you have invested?

Christophe: The cost of supporting such a business from an IT perspective and the cost of investing into new systems and processes on a global scale over the last eight years runs well into millions. Not only have we reduced our operational risk, but we also impacted positively our execution performance, so I feel it was - and still is - well worth the investment.

Adam: You've been, one could say, constructively critical of the level of dialogue with regulators. At the same time, with the investment you've made in technology, you would seem to be well placed for adapting to some of the directives in terms of providing full reporting. I'm wondering if new regulation is affecting you less than other firms.

Christophe: As much as I can be critical sometimes of the lack of discussion and exchanges with Brussels, I have also been fairly critical of the fact that the buy side has been pretty absent from Brussels. To this end, and in order to be more proactive with European commissioners and politicians, we will try to secure more time with them in order to better explain what we do, and how their decisions can affect our businesses. As for local regulators, I've always had a very constructive dialogue with them, especially since MiFID kicked in, or with the Cassiopeia initiative on fixed income.

Regulation is supposed to provide a level playing field to protect the end investor. And that's far from being the case. I think some decisions which are taken in Brussels today might affect less the big powerhouses like AXA IM because of the investments we've already made in our infrastructures, and also due to the commercial weight we have on our brokers. For a smaller size institution, that's a different story. The investment needed to cover some regulatory obligations will clearly affect them, hence our duty to be more proactive where it matters for all.

I also recognise that the learning curve for many people in Brussels, especially policy makers, has been very steep, with technical discussions on HFT, pre-and post-trade transparency, large-in-scale waivers and so on.

To come back to your point on how regulation affects us, I feel that we are pretty well organised at AXA IM to face up to our regulatory obligations. Being a multi-expert asset manager the depth of expertise of all functions, including support function, is staggering. EMIR for example has drawn - and is drawing - on our internal resources and expertise on a daily basis while we still have to invest massively in other projects. The fact that the EMIR deadline has been pushed back by Brussels is a testimony to how complex this piece of legislation actually is.

Calibration and simplification will be key because the story will not be the same depending on the type of assets and the potential impact of each transaction in the market. We cannot apply the same thresholds to different assets. Fixed income today represents a very large part of any European-based asset manager's AUM and MiFID 2 is trying to encompass assets which were not looked at in the first instance. So all in all, it still is massive work to be done in a very short time frame.

So yes, realistically, we would like to spend more time in Brussels, and I think it would be good to try to explain better, or in simpler words, what we do on a day-to-day basis and how some regulation would actually affect our business.

Adam: Could you sum up what you'd most like to see on pre- and post-trade in terms of focus, and are the regulators listening?

Christophe: Some of the proposals are nothing but good sense. Of course, those decisions are not serving the interests of all market participants - it's impossible to come up with a rule that will be accepted by all. But I do feel that with the right calibration we should be able, for example on the pre-trade aspect, to get the right level of transparency and market-making obligations. And on the post-trade I think it is exactly the same situation. We are an institutional investor, we invest for the long-term and fully understand when the trade should be disclosed or not. We understand when the size of a transaction might impact the marketplace and hurt the interest of the liquidity provider who might have taken a proprietary risk. I am afraid it will be challenging to come up with a one-size-fits-all model.

We know the markets are still not as efficient as we would like them to be. The Libor scandal has painfully reminded us how some people are still using market information to maybe front-run or manipulate data and prices against the benefit of our investors. So I think that we need to try and have a more constructive dialogue.

And to answer the second part of your question, are they listening? To be honest, yes they are, but they are also hearing many different viewpoints and some lobbying has clearly taken its toll. Today, I am not quite sure which door I need to knock on in order to be listened to. At the same time I refuse to categorise brokers as the bad guys even though some have been pushing their own agendas without their clients' interests at heart.

Of course it would be ideal to get permanent representation in Brussels. The top 50 European buy sides institutions could easily fund an office in situ where policy makers or anyone interested in what we have to say would come and hear from the buy side practitioners how we manage our businesses. That would actually allow us to keep momentum, which I feel is missing today.

So we need to be a lot more engaged. Do we have time to do this? Well, we'll have to make time. I don't want regulation to be - I love this expression in English - death by a thousand cuts. We should try to simplify some of the discussions and also be able to have some testing periods, whereby you implement the policies, but if you feel that there's a gap or if it's not actually addressing the right issues then we should be able to review and analyse the impact and go back to the drawing board directly with the policy makers.

This, of course, is wishful thinking as policy makers are doing just that, making policies across a multitude of products and affecting millions of people - the financial markets being one of the most focused areas of late - and quite often for the right reasons. But it is not all so negative, and the challenge is to show the benefits of efficient markets to a wider audience.

Adam: So there's a lot of box ticking.

Christophe: That's the way I see it. They are ticking boxes in an orderly fashion as the scope they are covering is so wide, and of course they are reacting sometimes to the people in the street, to politicians, to the press, to some lobbying being quite indiscriminate when it comes to showing us as the bad guys. We must continue to show we do not use financial markets as a huge betting platform; rather we are acting with the best interest of our investors at heart.

Adam: One semi-serious question: have you ever thought that you're in the wrong job and that maybe you should be working from the regulatory side?

Christophe: To be honest with you I love what I do, I really enjoy my job at AXA IM, and I am happy to take up challenges. Quite funnily, one regulator has actually asked me following a panel discussion last year whether I would consider maybe jumping over. So I said, 'No, not for the moment.' I'm 47, so maybe when I am a lot wiser and a lot older … and when my mortgage is fully paid up.

Adam: Then you'll join the 'dark side'?

Christophe: Then I will spend more time with my family.