Morgan Stanley?s Pete Eggleston and Gio Pillitteri talk to Colin Lambert about algorithmic
execution and the firm?s new algo products.
Colin Lambert: Morgan Stanley has been offering algo execution
for some years now, what is new about the impending release?
Pete Eggleston: We’re launching a new suite of hybrid algo
products that are a collaboration between our QSI [Quantitative
Solutions and Innovations] Group, which I run, and the e-FX
trading business, which Gio runs.
We do have an existing suite of algo solutions that offer
strategies like TWAP, as well as MSFix – a VWAP proxy – and
MS Radar, an opportunistic cost control algo. The first new
hybrid product is called Morgan Stanley Seeker, it is designed to
opportunistically access liquidity efficiently and effectively. It is
a relatively simple algo that will execute as quickly as possible
while minimising market impact and maximising efficiency.
Later this year we will be adding other strategies that also take
elements of our existing suite of products, but we are planning
a limited release in terms of the number of strategies; we don’t
think it necessary to swamp clients with many different algo
CL: That’s an interesting approach because there are providers
out there that seem to roll out a new strategy every week.
PE: Anecdotal evidence from clients shows that no matter how
many algos are on offer, approximately 80% of those used in
FX are currently TWAP-based. We could roll out very complex
strategies – we have done bespoke work with clients before that
has resulted in very complex products, but in general our
clients are telling us they want a simple, transparent algo that
gives great performance. We are looking to offer five or six products to sit on top of our
liquidity infrastructure to help deliver superior performance.
Gio Pillitteri: Importantly it will be provable performance.
Seeker will earn or pay the spread depending upon market conditions – it is all about minimising transaction cost by
operating within the framework we have built that offers a
superior liquidity profile.
CL: You have mentioned the framework or infrastructure behind
the algos a couple of times – this is obviously a very important
part of the service?
PE: The ultimate aim is best execution for our clients and to
achieve that we need a robust, proven transaction cost framework
within which we can operate. To create that framework required
a collaboration between QSI and the e-trading teams because
each could offer a unique view of the market to help us build the
GP: Our work overlaps to quite a degree. We have done a lot of
work on execution advisory for our clients and that has brought
us closer to the QSI team. Our dealing desks have been using
solutions developed with QSI, the client rollout is the next phase.
Our objective is the best possible liquidity picture for our clients.
This is most definitely not a move towards having 20-plus algos
on offer, we have focused on the best possible construction of
liquidity. Any algo is a function of the quality of the liquidity
you are able to access, but it is not about hitting everything on
offer – you have to be smart about how you execute and you
have to understand the nature of markets. The algo should be a
proxy of a good spot trader and be able to judge liquidity and
activity, and react accordingly.
CL: So understanding the liquidity mirage is important.
GP: The liquidity mirage has been there for years and
fragmentation has exacerbated it. Only if you work closely with
the markets do you really understand the mirage though, and that
is core to the work we have been doing with QSI. Liquidity is vitally important and clients need advice, it is not
just about executing on the major pairs that are liquid throughout
most of the day, often our clients are looking to execute in non-
liquid markets or outside local market hours, conditions that can
lead to significant slippage. It is our job to minimise that
slippage for our clients and we have done a lot of work with
them on volume prediction and the deployment of algos that
offer that efficiency.
CL: How easy is it to create that liquidity picture to ensure you
hit the right places at the right time? After all, liquidity can be
transient across platforms.
GP: We are very selective about where we hit in the market and
we do a tremendous amount of post-trade analysis on our impact
on various venues. We can, for example, often run the same algo
without hitting a specific venue at the same time of day as the
original order was executed to help us empirically judge the
impact of our business on that venue. We have deployed a lot of resources to creating the best possible
liquidity framework and of course, as part of that you need
optionality so we are connected to a lot of venues, but it is
emphatically not about hitting all of them.
CL: It seems a big part of the new suite of algos is about
anonymity? There is increasing noise in the market about certain
market makers using predictive, or “sniffer” technology that
leads them to drive the market away from the order. How do you
protect your clients from that type of activity? Equally, some
clients are reluctant to let the bank desk know about the order –
is that the best way to operate?
PE: Information leakage is a big issue with our clients and in
discussions it comes up again and again. The answer has to be adapting the strategy to the client’s needs. and ensuring the
strategy interacts intelligently with varied liquidity sources,
including our own pool. It is the different and varied nature of the structure of the FX
market that makes it so tricky and at the same time so
fascinating. Some clients are treating FX as an asset class – big
investor type clients who are not trading linked to an underlying
securities transaction – while others are seeking to maximise the
return on their underlying investments through efficient FX
hedging. To your second question, going forward, I am not sure the pure
agency execution model will be sustainable in FX due to the
client demand to keep the order quiet. If you do not want to leave
a footprint the only way to do it effectively is by interacting with
a bank that can internalise flow, warehouse some risk and use
their extensive presence in the markets to help mask the source
of the order. Occasionally a client raises the issue of the trading desk not
seeing the order, and we ensure we have strict controls in place
to ensure anonymity. However, the majority of clients appear
relaxed about this issue and the key is the degree of trust in the
relationship, and the fact that post-trade TCA will demonstrate
that the execution was carried out as effectively as possible.. The
key is getting the balance right to ensure you maximise the
liquidity available without having market impact.
CL: It sounds to me that you don’t see FX going to the equity-
like exchange model?
PE: We are in a period of flux in the FX market but no, I don’t
believe it will fully go the way of equities and that a hybrid
market structure will prevail. This is the structure that clients are
effectively driving the evolution of, not the banks. Clients are saying they want anonymity and lack of footprint and do not
typically want to execute in a fully lit all-to-all environment
where there are predators seeking to pick up on their order and
GP: Our algos are a different flavour to those typically used in
equities where the order is just routed to various venues
depending upon where the bid or offer is, and it very quickly
becomes public knowledge and can lead to slippage or a partial
fill. As Pete says, clients do not want to leave a footprint and
while it is hard to do that in FX due to the sheer size of the
market, if you hit every venue you make it easier for people to
spot the order – and that can create problems for you.
CL: What role does the sales desk play in this structure?
GP: The sales desk is a big part of the process because they
know the client best, they are closer to them and thus can better
understand their objectives. Working with the sales team, we can
then help build the right framework around the customer’s
PE: It is about making sure you have the right people in the right
place. We have ensured that clients have the right specialists at
the important touch points with Morgan Stanley. If they want to
talk to someone, the appropriate person is there for them. It’s an
approach that takes in aspects of QSI, sales and trading and it
GP: I also think that if you separate the three functions you are
creating unnecessary operational risk. A unified approach to
customer service works best.
CL: An area of interest to a lot of people at the moment is
benchmarking – both the public fixes, which appear to be
under some scrutiny at the moment, and the role algos can
play in delivering benchmarked execution. Where do you see
PE: Benchmarks are interesting because there is no “one size fits
all” solution – it depends very much on the client and what they
want to achieve. We have developed TCA Benchmarker, which is
an application that allows clients to build a tailored benchmark
specific to their needs. You have to be able to benchmark to a relevant mark that suits
the needs of the clients. For example, asset managers typically
like to construct benchmarks closely aligned with the
exchanges and local venues on which they are trading the
underlying investment. From there, they are able to deploy one
of our algo strategies that executes over a similar time horizon
to the underlying asset trades, which reduces the potential for
tracking error. This approach is also more efficient than trading on a single fix
that may not even be set during local trading hours. A lot can
happen between a local market closing in, for instance, Asia, and
the 4PM London fix that so many asset managers like to use.
The challenge for the client lies in the difficulty of constructing
the necessary liquidity profile as data can be thin on the
ground. They need to ensure there is sufficient local depth
before loading up the algo strategy, which is where we can help. If we can create a strategy that optimises the liquidity
available during local hours and continues to execute without
impact out of local hours, there are considerable savings to be
had for the client.
CL: Interest in algo execution, for whatever reason, appears to be
growing strongly. Is this what you are seeing and how do you
respond to that demand?
GP: We are seeing increased interest in algo execution from the
corporate sector and other clients that do not trade FX
professionally and are more interested in making their hedging
process as efficient as possible. We believe the key to meeting
this demand is making it a simple product at heart. Many clients
have to explain to the compliance committee what it is they want
to do, it makes that job a lot easier if the product or service they
are describing is simple to understand. The algos cannot be too complicated otherwise it becomes
harder for the customer to explain the benefits internally and
they will not get permission for their use. You cannot have a
quant team just develop tremendously sophisticated strategies;
that just confuses things. The bottom line is all about
accessing liquidity – nothing else – and that is a relatively
CL: There is also the need to create a product specific to FX.
PE: Our new suite of algos has been born out of a generic
demand from our clients for something that is simple to
understand and transparent. We have an advantage because we
are able to work with a central quant group at Morgan Stanley
that has been looking at these issues across markets for several
years and as such has a knowledge that we can tap into.
Overlaying that, however, you need to have some FX specificity
– which can only come with understanding the nuances of the
FX market and being able to utilise that knowledge in creating
products for the FX community. We didn’t want to simply take a list of algos from other
markets and offer them out to our clients, we wanted to take
some great ideas and use them in an FX specific way – Seeker
is the child of that knowledge transfer. We believe the need is
for a small family of algos that meet the needs of a variety of
clients, all of whom have individual objectives that are
transparent and simple to use.
CL: What about that small group of clients that want to be able
to interact with the algo? Those clients that see a quasi-agency
model taking their job away from them?
GP: There have been instances where clients have wanted to
suspend and amend the algos on the go, but increasingly that
interaction is coming earlier in the process. Our research tells us
that interacting with the algo during execution rarely – if ever –
improves execution efficiency and effectiveness. The time to play
with the strategy is before you execute, during conversations with
us. That is when the customer sets their objectives and identifies
their benchmark. Once the strategy is live we think it is better to let it run. You
need to trust the data, the transaction cost analysis – it is all
about the numbers.
PE: We have spent a lot of time over the last 18 months working
with over 300 large investors from different sectors to identify
how we can improve their execution efficiency. These people are
talking to us because they want best execution that is measurable
and transparent, and we are able to work with them to help them
achieve those goals by providing analytics, products and services
that address the full execution process, from pre-trade, actual
execution styles and methods and through to benchmarking and