I have always been a little ambivalent over the importance of
high frequency trading, especially in FX, and my relationship with the segment
has kind of reflected that over the years – I get the value it has brought to
the market but I have always been a little unsure over certain models’
years ago I was talking about HFT “eating itself” and given the
announcement of the deal last week between DRW and RGM Advisors, the feast seems
What I find interesting is how so much of the media
attention has been focused on consolidation driven by lower volatility in
markets. Yes, volatility is lower, but could these deals we are seeing actually
be the result of there being too many HFTs?
Looking at the FX market specifically, one or two firms have
reinvented themselves as genuine market makers and established a solid presence
but there are sundry others merely using last look as an enabler of their strategy,
which is, it could be argued, still an arb game.
I know I can’t help myself when it comes to the practice,
but several of the HFTs in FX that are claiming to be market makers are merely
trading, placing a bid/offer in the market at a small profit, and then only
accepting the original trade if the cover trade is done within the last look
This is not market making, it is arbitrage with a twist –
and the twist leaves a nasty taste in the mouth.
The original HFT game was latency arbitrage as we all know
and this was predicated upon only trading when you make money. The only
difference between the latency arb model and “market making” as described above
is one involved two aggressive trades while the other uses (and abuses I would
argue) a self-imposed artificial delay in accepting a trade.
I am not sure if I have shared this with readers before (I
can’t believe I haven’t), but to me the archetypical story that describes the
HFT approach to trading is something shared with me a while ago by a senior
manager at an HFT firm. The manager revealed the firm traded something in the
region of 40,000 times a day in FX but had a stop loss limit of $2500. He was
very proud to say that in the previous 12 months the firm had had to stop
The firm concerned was very obviously not holding risk and
in my opinion those “market makers” who only trade when they have a cover trade
at a profit are not either. I know they can technically say they are holding
risk for a few hundred milliseconds, but if they can then reject the trade what
risk is there? None at all of course.
Notwithstanding the fact that these firms can still make
money in FX, it remains a relatively small market for many of the major HFTs.
They came out of the exchange-traded world, and with FX remaining resolutely
OTC, they have largely headed back there. This means that the low volatility
and lower volumes in equities for example, are bound to have an impact.
What is often overlooked, however, is the impact of
competition. A little more than a decade ago, HFT was typified by the four guys
in a garage in Nowheresville USA. They had tech smarts, great connectivity,
were very nimble and knew next to nothing about a market’s characteristics. All
they knew was the market structure (i.e. the technology) could be exploited,
and they did so with great success (and into the bargain helped financial
markets generally upgrade their technology offering, for which HFT’s role
should rightly be recognised).
The problem, as I saw it six years or more ago, was that
there are only so many opportunities to “exploit inefficiencies” – which
remains one of the great PR BS lines of our time – in markets. There could, for
example, be 10 venues supporting trading in the same asset, but there were
suddenly hundreds of firms chasing that same opportunity. There was, and still
is, only room for one or two to get through the door marked “inefficiency”
before it slams shut.
Thus, the firms found themselves in a technology arms race
wherein they were spending more at the same time as their returns were
diminishing. Even with the evolved “market making” model, there are only so
many firms that can be top of book in order to be able to cover the trade
immediately for a profit.
I don’t see how a lack of volatility or even volume can be
blamed for the shrinking nature of the HFT world – whichever way you look at it
there are a limited number of opportunities in these markets and that means
only a limited number of firms can succeed.
It is not only that an HFT is competing against its
historical rivals either, for in the tech world the challenge can come from
anywhere – a group of people invent or discover a quicker way to communicate
and the last 10 years R&D spending counts for nothing because there are new
kids (often literally) on the block.
So last week’s DRW/RGM deal should not, and most probably is
not, a surprise. The rising costs allied to increased competition surely mean
this is merely the latest in a long line of deals and/or closures.
On a related topic, I am indebted to a friend who pointed
out to me last week a release from a firm called Vigilant, in which the latter
stated it would not be appealing January’s decision by a local UK council to
refuse it permission to build a 322 metre communications tower on the English coast.
Obviously this was the next step in the push to squeeze
extra microseconds out of the arbing…sorry I mean trading…process by providing
line of sight connectivity between the UK and continental Europe, which would,
naturally, reduce round trip times between London and the main European trading
Vigilant was one of two firms to apply to build a tower, the
other was a firm called New Line. Vigilant is, coincidentally, owned by DRW;
New Line is owned by KCG (now Virtu after those firms merged earlier this year)
and Jump Trading.
My point is that even here, where a firm is spending a huge
amount trying to gain an edge, that “edge” is already halved by the presence of
a rival tower. All it needed was a third tower, build by another firm, and the
law of diminishing returns would have been putting in overtime.
In FX terms, many of the original HFTs were thwarted by a
very effective ring fencing operation on the part of the broader market. They
were unable to get to the original flow, failed to adequately penetrate the
manual traders’ market and were left to deal with…well each other actually.
Funnily enough, HFT dealing with HFT has never been a particularly popular
pastime in those circles, hence they exited FX or went to market making and
using last look.
It strikes me the same thing is happening in equities
markets. Institutional clients are either finding better ways to execute (block
trades and dark pools with minimum size) or are just doing less volume; day
traders are fed up with being pipped by HFTs and are also withdrawing from the
market; and the ‘buy and hold’ mentality seems to be creeping back in – all of
which leaves HFTs to deal with other HFTs.
When the strategy was first popularised, HFTs were fond of
telling us that they were making markets “more efficient”, and they were right,
they did make price more efficient. The trouble is that is only a short term
job because once a few HFTs were at work the markets became efficient and from
that moment on it was all about speed and shaving milliseconds.
While all this was going on, we seem to have forgotten the
basic definition of a price maker in markets – and it is this that means my
ambivalence to HFT continues. A dealer using quote and cover tactics is not
performing a social good – a risk warehouser, on the other had, is.
HFT did make markets more efficient, and it also helped the
industry raise the bar significantly in technology terms. The trouble is, the
time when this was needed or desired has long gone.
At some stage, we will probably see the HFT industry
consolidate into a very small number of firms and they will probably go OK, but
they will have to live with the risk of being disrupted by a new generation the
way they themselves disrupted. Ironically, FX remains a beacon of light for
those firms that moved into the “market making” business some years ago, but
even here I have bad news for them.
As greater transparency around last look procedures becomes
the norm, they will no longer being able to hide behind asymmetric response
times and latency buffering – market participants will not stand for it and
they will trade elsewhere. This means pressure on their FX businesses as well
as equities (although no doubt there will be opportunities in some swaps
markets and other products as they electronify).
Actually I am not totally correct in saying no-one will
stand for latency buffering and asymmetric round trip times in FX markets –
HFTs will because they exploit the framework’s weaknesses. There again though I
have bad news because while we will not, in the purest sense, have HFTs trading
with other HFTs only, we will have non-risk holding market makers dealing with
other non-risk holding market makers.
Semantics I know, but if it’s not exactly HFT eating itself
, it’s a bunch of close relations tucking into their own.