There is one area of the Global Code of Conduct that continues to attract controversy, and, Colin Lambert says, we all know what it is…

Although the assessment is a little harsh given the type of
misconduct that led to the creation of the FX Global Code
of Conduct, it is hard not to understand where the head
of e-FX trading at a major bank in London is coming from when
they note, “The Code had one job – give us clarity on last look
– and it has failed miserably.”

There remains the odd voice still raising concerns about
Principle 11 and its apparent endorsement of pre-hedging,
however, Guy Debelle, chair of the FX Working Group that
created the Code, stresses this Principle is really about the
“demonstration effect”. He says, “Participants will need to
demonstrate to their clients that pre-hedging brought them a
better result. It is unrealistic of a client to leave a very large
order in the market and not expect a potentially sharp move, so
if the service provider believes – and can prove – that an
element of pre-hedging can smooth the execution, that is a
better outcome for all.”

That aside, the lightning rod for controversy has been, and
remains, last look. As we report in our interview with Debelle, sentiment is running in favour of removing the
word “likely” from the text, meaning the somewhat infamous
Principle 17 will stress that using information from the client
request in the last look window is “inconsistent with good
market practice”.

This will come as music to the ears of the critics of last look,
not least CEO of LMAX Exchange David Mercer who has been a
long term critic of the practice; however, Mercer believes there
are still issues with Principle 17, as well as how some of the
Code’s creators appear to disassociate pre-hedging entirely
from last look.

LMAX endorsed the Code in early April – becoming the first
platform to publicly do so – but did so with a caveat, stating
that its support came “after we have been assured that the
Code will be updated post-publication”.

Mercer confesses to being “disappointed” that last look is
not being written out of the market, but adds he wants to take
a “pragmatic” approach to the issue. “LMAX Exchange is
committed to best practice and transparency so it makes
sense that we adopt the Code because aside from the issues
of pre-hedging and last look, it provides clarity over what
constitutes best practice,” he explains. “I am disappointed that
Principle 17 still includes an apparent endorsement of last
look, but I think the Working Group is taking a step forward by
stressing that people have to convince them not to remove the
word ‘likely’. [The text states: “During the last look window,
trading activity that utilises the information from the client’s
trade request, including any related hedging activity, is likely
inconsistent with good market practice.] Hopefully we will get it
right further down the road.”

Although the creators of the Code stress that pre-hedging
has nothing to do with Principal 17, Mercer disagrees. “On
one hand we are told that pre-hedging has no part of Principal 17, but on the other the Code says trading activity
related to the client order is likely bad practice. Trading
activity in the last look window is pre-hedging, no matter
what people say – it may not be on every occasion, but more
often than not it will be.”

Although there is a growing body of opinion in the industry
that all the Code has to do is hold the fort for a couple of years
before last look becomes obsolete, Mercer warns against
complacency. “All that is really happening is we are
concertinaing the practice,” he says. “Five years ago we were
talking about primary venues updating every 100 milliseconds,
now we have one at 5ms and we are likely to soon have 5-
10ms updates across the board.

“People are saying last look is nearly gone, but the problem
is technology continues to improve, so much so that there are
market makers who can trade inside that window,” Mercer
continues. “So nothing has really changed, people can still
trade in advance of the order in the last look window – the
problem hasn’t gone away, it’s just got quicker.”

What is harder to credit in the current environment in which
so many profess to following their clients’ wishes is how market
makers generally support last look, but a growing number of
their clients do not.

While some of the more aggressive traders are fine
with the practice – they just want the tightest possible
price and will accept rejections – more and more asset
managers appear to be against last look, especially
those involved in mechanistic hedging. “I don’t know why
I am being rejected just because the market may have
moved a relatively small distance,” says a senior
executive at a global
asset manager. “It’s not
as though I am timing my
execution or spraying an
order around, so to get
the ‘NACK’ message in
my logs really irritates

“In an ideal world I would
shift my liquidity providers
around, but the sad fact is
they’re all using last look
and it’s generally to my
detriment,” the head of
execution adds.
Another senior manager
at a US-based asset
manager, adds, “Last look
paints a false picture of liquidity and that doesn’t
help my business. We
need to know that we can
hedge appropriately at the
time we want, without
impacting the market – we
don’t want any footprint. I
would rather a firm price
that is slightly wider – it’s
still in competition so it
won’t be that wide – than
a string of unrealistic
prices that I can’t, or don’t
want to, hit.”

When asked if the asset
manager believes market
makers are jumping in front
of their order using last
look, the response it blunt. “Absolutely. Why else would they
reject someone who is non-directional?”

Quantifying the Impact

It is the rejecting of trades from what would generally be
considered non-toxic flow that prompted Casper Marney, CEO
of Velador Associates, a firm that provides execution analysis
tools and has recently been providing consulting expertise for
some of the recent financial litigation cases, to undertake
closer scrutiny of last look.

Marney sits on the ACI Committee for Professionalism (CFP)
working group that helped deliver the association’s take on last
look in its Model Code, and from recent press reports, he
believes the ACI version is stronger in its language than the
Global Code. “I feel that ACI has it right when it says that rate
last look is bad practice,” he says. “If recent reports regarding
last look in the Global Code are correct, then it would indicate
that the Global Code stresses the need for transparency, and
that is good, but it does nothing to stop what we consider to be
an abusive practice.”

Marney says Velador uses “real data from dozens of buy side
firms” to quantify the impact of last look on these customers’
business. “We conduct toxicity analysis on the flow as well and
the results are quite startling,” he explains. “Even for GUI
traders, whose flow is rarely, if ever, toxic, last look is having a
huge impact on their bottom line.”

Marney has showed Profit & Loss anonymised data from one
client as support for his argument. The data is from a GUI user and
indicates that the client has a rejection rate of 22.5% by volume and a very slightly lower
rejection rate by trade count.

Velador’s data offers
trade response by currency
pair and Marney says that
rejections are not limited to
a small number of markets.
“Rejections for this user are
spread across liquid and
emerging markets,” he
explains. “It shows a
systematic rejection regime,
especially by trade size.”

The data is used by
Velador to provide a
graphical representation of
the order handling process.
“If a process is normal, our
graphics show a slight skew with a long tail after execution,”
Marney explains. “Any artificial latency (i.e., buffering), shows
up as a spike on the chart. It makes it very easy to see where
dishonest latency bufferers are holding the order bank to do a
last look check and then reject it.”

Marney also points to a chart showing price improvement and
negative slippage for the client’s trades, noting, “It is universally
a net financial loss to the
client. This is often because,
as the data show, there is a
straight dollar loss from
them being rejected and resubmitting
the order at a
worse price,” he says. “The
graph we produce indicates
price improvement and
slippage for the client. A
skew to the right is an
improvement to the client, to
the left the liquidity provider
– and this chart, as does
just above every other, is
heavily skewed to the left.

“This shows that last look is not being used symmetrically
the way some LPs claim,” he adds.

When asked if the data doesn’t merely show the client,
through immediate re-submission, is behaving aggressively,
Marney is quick to disagree. “The LP is re-pricing with
knowledge of the original order,” he points out. “If the order
was re-submitted to another LP that argument could be made, but this data highlights the information is already in place. I
think the real question is why, if the LP has knowledge of the
original order, does it reject a second time?”

If recent reports are correct, one of Marney’s complaints
about how the Global Code handles last look is that by merely
promoting transparency, it enables LPs to hide behind a
disclosure notice. “We have analysed data since the New York
Department of Financial
Services fined Barclays for
being dishonest about how it
was using last look and
indicated it should be used
symmetrically, and it doesn’t
look any different,” he
argues. “There is barely any
change in the clients’
fortunes in a supposedly
symmetric world.

“I think we can define
symmetry better than the
industry currently does. It
should be a blend of how far
the market moves after a
reject; the balance between rejections in favour of either party;
the volume of rejections; and the P&L on rejections,” he
continues. “That provides a much clearer picture of what this
practice is actually costing clients.”

So while Marney welcomes the Global Code as a step
forward, he remains unconvinced, on the subject of last look
at least, that it goes far enough. “While I understand that the
Code mentions the need for
participants to make
available metrics that
facilitate transparency
around order handling, I
would argue we need to
directly question why they
are holding a trade? Why
are the reject messages
ambiguous and generic?
We need more clarity over
why clients are being
rejected, which is why I
would like to see stronger
language in the Code. There
is a sense that in its current
form it condones what we
consider to be a bad

Galen Stops

Share This

Share on facebook
Share on google
Share on twitter
Share on linkedin
Share on reddit

Related Posts in