Inevitably, in this world of vested interests, people will
be quick to jump on today’s LMAX white paper on TCA as evidence the platform is
talking its own book. Let’s be clear, it is. But that isn’t the whole story as
far as I am concerned, for without opinions (for whatever reason they are
generated) we don’t have debate, and we don’t move on as an industry.

There is also the argument that what this paper does is
provide genuine empirical evidence around the cost of trading on last look
pricing – and as suspected by those of us (me included) who don’t like the
practice of last look, that cost is not insignificant.

My main takeaways from going through the paper twice (I need
at least another two cracks I reckon!) is that I hope it does trigger further
debate around TCA and the impact of last look, and also that other platform
providers either share their data or provide their own analysis.

LMAX is to be congratulated for doing this work – and
although they are busy with other things such as a spin off, rebranding and a
change of management, both Nex Markets and Thomson Reuters operate venues with
both firm and indicative (I know, I’m just being mischievous) pricing and as
such they could contribute tremendously to this debate.

Equally, on consideration, price improvement maybe doesn’t
get the attention it deserves when we study TCA. I would take minor issue with
the thesis put forward in the paper that price improvements should be available
on limit orders, to my mind if a participant wants to buy or sell at a level
they should do so – whether the market moves in their favour or not is
irrelevant. On market orders however, I agree, price improvement can and should
be taken into account.

A third takeaway is that the performance range of the six
LPs studied is varied enough to provide serious evidence of a schizophrenic FX
market! Names are obviously not available but when it comes to fill ratios
across both limit and market orders there is quite some variance – although not
as much as the ratio of negative slippage to price improvement.

It is well known that I don’t like last look and would like
to see the practice eradicated, but for the time being at least it exists and
as such the anonymous Bank 1 is to be commended for a 99.94% fill ratio on
market orders and 99.71% on limit orders. I would like 100% but if one assumes
the pricing is competitive from this LP then it’s doing a good job.

Likewise Non-Bank 2, with a 99.98% fill ratio on market
orders and 99.11% on limit orders – and doesn’t this just highlight how, in
spite of some peoples’ attempts to argue the opposite, that the two sectors
operate in a similar fashion.

I can just about tolerate Bank 2 with 99.63% and 99.80%
respectively, while Bank 3 and Non-Bank 3’s performance is underwhelming, but
not as much as Non-Bank 1’s (96.95% and 97.48%), whose data suggests we either
have a major liquidity recycler on our hands or someone still harbouring under
the “market maker” name while operating a latency arbitrage strategy.

Just to confuse the issue and highlight the wild variance,
the aforementioned – and highly praised – Bank 1 registers the small matter of
a 10.38 ratio of slippage to price improvement. Its closest competitor is Bank
2 at 3.94, which suggests Bank 1 will only deal with you if you are benign
(i.e. often wrong) or dealing in smalls!

Bank 3, remarkably, has a zero percent ratio – very much a
case of “here is my price, like it or lump it”, which I have to say is an
attitude that appeals very much to me!

So generally speaking I tend to agree with the findings of
the white paper, however (and we all knew that was coming!) there is one aspect
of it where I think it falls short and fails to take into account the serious
concerns of the buy side.

I accept that, as the report notes, firm liquidity means the
trader can execute with no pre-trade information leakage, but, as I have
repeatedly argued in this column, in a world where the average trade size is
hovering below the $1 million mark, executing even 20 million units can lead to
slippage. In other words, just as lethal to the execution quality is the
in-trade information leakage.

Like it or not, there are a bunch of LPs out there (bank and
non-bank) who are just as happy to fill the first trade and then trade with the
counterparty in expectation of further selling or buying. This creates slippage
and, as far as I can tell, breaks no securities laws.

For TCA to be meaningful, I strongly believe the larger
order has to be taken into account and rather, as suggested by the LMAX paper,
measuring slippage from the arrival of what is likely to be a “child” order and
the final fill, it should be measured from the moment the parent order is
initiated.

I am not for one second suggesting that executing larger
tickets on last look liquidity is better, rather I am pointing out that
slippage is a hydra – it has to be measured in more than one way, and one of
those ways is the cost of executing part of the order on a lit venue.

While firm liquidity undoubtedly helps kick start the
execution, it also has its drawbacks, and if we exist in a world in which only
firm liquidity is available the only way LPs can protect themselves against the
“machine gunners” in the market will be to quote on fewer venues and at wider
spreads, with the obvious impact of client execution quality.

Although one cannot ignore the argument that LMAX has an
interest in promoting firm liquidity amongst market participants, this paper
does contribute to what I sense is a more strenuous effort to understand the
impact of last look – and, if I am right, help rid the market of the practice.

Recently I sat down with Velador Associates, a consulting
firm that is looking deeply into the issue of cost of rejects. Its findings
were similar to those released today by LMAX in that there is a definite skew
in favour of negative slippage over price improvement (more on this in the next
issue of Profit & Loss, out on
May 25).

At face value this data looks scandalous, and indeed may be,
however some caution is required, for even the most benign of clients execute
in a moving market. Those that leave limit orders are the exception, but my
experience over the years was that even the most friendly clients, who were
watching the markets with a view to executing an order, would jump in when it
started moving.

In other words, as the LMAX analysts accept, a lot of
trading is done “with the wind” and in these circumstances, when last look is
available, it is inevitable that rejects will cost more than price
improvements.

So there are two challenges as I see it for the industry and
the authors of today’s report.

On the one hand, the industry needs to be brave enough to
share its data so we can have a proper, industry-wide, empirical analysis (I
understand commercial pressures may mean LMAX doesn’t get to write the report
but there are plenty out there who can analyse it independently).

On the other, following on from today’s report I would like
to see the follow up include more analysis of market impact relating to the
“parent” order (and I understand this also probably needs other players to join
the effort) and also an attempt to break down the cost of rejections according
to volatility levels prevalent at the time.

I fully embrace the need to broaden the debate on TCA
metrics and I agree with many of the findings in this report, as I have already
noted, but we should not, for one moment, think we can ignore the role played
by predatory traders in creating some of these rejection statistics.

True transparency in this area, I would argue, involves more
of the type of empirical work released today, overlaid with a reasonable
representation of the executing parties’ historical market impact and
behavioural patterns. As anyone who has been rejected can tell you, if they’re
honest, both parties often share responsibility.

Colin_lambert@profit-loss.com

Twitter @lamboPnL

Twitter @Profit_and_Loss

Colin Lambert

Share This

Share on facebook
Facebook
Share on google
Google+
Share on twitter
Twitter
Share on linkedin
LinkedIn
Share on reddit
Reddit

Related Posts in