Sei sulla pagina 1di 26

Propeller-Head Ubiquity:

The Next Generation of Quantitative Trading











E. Paul Rowady, Jr. | V12:039 | June 2014 | www.tabbgroup.com

Data and
Analytics

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 2

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Vision
TABB Group has been closely studying and commenting on the evolution of quantitative
trading for many years. And though the fruits of quantitative research methods had
already permeated a vast spectrum of trading strategies to varying degrees,
depending on the level of workflow automation, among other factors just a few years
ago, high-turnover versions of quant strategies attracted the most attention. They still
do. This is primarily because of their awe-inspiring departure from anything related to
human scale as well as tales and growing evidence of their staggering risk-adjusted
returns (and absolute profits). Paradoxically, extreme success always defies attempts
to remain hidden, providing an attractive scaffolding for pundits, media and students of
the markets on which to build their narrative.

In June 2011, TABB Group published Quantitative Research: The World After High-
Speed Saturation. Our hypothesis at that time was that the latency arms race would
result in fewer players capturing more theoretical capacity in core markets, asset
classes and products the tip of the spear thereby forcing all high-speed quant
players (that were committed to remaining in the game) into new or additional regions,
asset classes, products, and/or slower turnover frequencies. In other words, ultrahigh-
speed trading in US equities had become saturated, and players would need to
diversify. Heres the excerpt:

Without so much as a whimper, the whole game has changed right under our noses.
Like a Rose Bowl parade marching band taking that hard turn onto Colorado Boulevard,
the quant game is set to head in a new direction, except without the hoopla. Thats the
nature of secretive sports: Theres no trophy presentation at the end of the season. And
theres no memo about the lockout. The more alert players move on, never disengaging
from stealth mode.

Of course, if last years [2010] Securities Industry and Financial Markets Association
(SIFMA) technology conference was any indication, the saturation of latency eradication
systems was palpable, indicative of the changes we face today. A handful of firms
wearing both sell-side and prop jerseysare successfully defending the limited capacity
in high-frequency and micro-market structure arbitrage strategies in the US. With the
prohibitive costs of incremental speed and dwindling volumes, punctuated by regulatory
zeal, the average quant team must face up to the inevitable: change or die. Alpha is
elusive, ephemeral and enigmatic. There is no guarantee of persistence.

For many of the players, the required measure of change will be too much to handle.
Over the past year, two prominent option-focused prop shops have yielded to the
superior speed and prowess of their competitors and taken their balls home. In one of
these cases, you couldnt possibly have thrown more bleeding-edge, out-of-this-world
low-latency technology at the problem. The impediments were beyond hardware, but
more likely related to a lack of quantitative creativity. Undoubtedly, others will follow.

For those willing to brave the journey ahead, there are plenty of juicy opportunities to
explore, some of which will certainly surpass the capacity and profit potential of current
microstructure arbitrage trades. But hunting for new sources of alpha requires new gear.
Surrounding and feeding this new gear is a requirement for new skills. Quants need to
cultivate a new zest for solving even messier data challenges; they also need to develop
a passion for the financial equivalent of a deep anthropological dig.

The need for speed will never end. Only the disproportionate focus of the recent past will
be reallocated. Cross-asset, cross-regional, multi-temporal, asymmetric versus
symmetric trades, and even enhanced front-to-back automation will all benefit from this
reallocation. Welcome to the age of multidimensional alpha.

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 3

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014

In March 2014, we conducted a survey to help us update our understanding of the
pulse of the quantitative trading landscape. The results of that survey can be found in
this report. Based on our understanding of quantitative history and coupled with
these latest survey results TABB Group is updating its vision and hypothesis for the
quant landscape:

First, saturation has spread beyond US equities into other regions, products and asset
classes. Today, fewer players are capturing more of the available theoretical capacity
for global high-speed alpha. At the same time, that theoretical capacity has been
shrinking.

Second, the future for quantitative methods has never been brighter. Our multi-
dimensional alpha concept is still intact. But the bridge from here to there as stated
in earlier research will require new skills and the digestion of a lot more data.

Lastly, quantitative methods will permeate more of the traditional investment strategy
spectrum and front-to-back office workflows to the point of propeller-head ubiquity.
Oddly enough, ubiquity will shroud quants in a cloak of invisibility. Automated methods
will become indistinguishable from everything else, since automated methods are
becoming an increasing large component of the whole. The only question now is:
When?


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 4

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Table of Contents

VISION................................................................................................................................................................2
TABLEOFCONTENTS...........................................................................................................................................4
INTRODUCTION..................................................................................................................................................5
2014QUANTSURVEY.......................................................................................................................................6
QUANTITATIVETHEORETICALCAPACITYFRAMEWORK..............................................................................................6
INGREDIENTSFORSPECIALSAUCE.......................................................................................................................8
LOWHANGINGFRUIT.......................................................................................................................................12
ASSETCLASSDIVERSIFICATION..........................................................................................................................12
PRODUCTCLASSDIVERSIFICATION......................................................................................................................13
REGIONALDIVERSIFICATION..............................................................................................................................14
HIGHHANGINGFRUIT.......................................................................................................................................16
ARAREGLIMPSE...............................................................................................................................................19
CONCLUSION....................................................................................................................................................23
ENDOFAGENERATION....................................................................................................................................23
BEGINNINGS..................................................................................................................................................24
ABOUT..............................................................................................................................................................25
TABBGROUP................................................................................................................................................25
THEAUTHOR.................................................................................................................................................25



2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 5

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Introduction
Although the metaphor might not be entirely accurate, we cant help but think of the
Roadrunner and Wile E. Coyote with the quants starring as the Roadrunner and the
rest of us observers, for better or worse, as Wile E. Since mid-2011, the lather over
high-frequency trading (HFT) has only grown, especially recently with the publication of
Michael Lewiss Flash Boys. The chatter on TabbFORUM and its newly launched
cousin, QuantFORUM, also has provided a growing archive of fact-based research,
ravings of opinion leaders, and other commentary.

Beyond the ebbs and flows of the media circus, however, there are many more tangible
clues that have added up over the interim period that have provided rare glimpses into
the revenue breakdowns of leading proprietary quant shops and which further serve
to support or refute our original hypotheses: that those firms engaged in the highest-
turnover strategies, and primarily those focused on US equities, would need to look
elsewhere; to focus in other regions, asset classes, and products and ultimately,
across all of these, including slower turnover frequencies. We have called that new era
Multidimensional Alpha, and identifying where we are in that overarching narrative is
the goal of this report.

Since mid-2011, the regulators, too, seem to be more on their game when it comes
to high-speed trading in comparison to many other topics. The January 2013 rollout of
the Market Information Data and Analytics System (MIDAS) by the US Securities and
Exchange Commission (SEC) is quintessentially indicative of the idea that Wile E. may
finally be gaining ground on the Roadrunner (See Exhibit 1). Or, at least, that someone
is paying attention to TABB Groups ongoing recommendation to show more of the data
in the form of pictures!

Exhibit 1
MIDAS - Order Book Reconstruction Visualization Tool

Source: Securities and Exchange Commission

Unlike in the cartoons, TABB Group believes that Wile E. does periodically catch up to
the Roadrunner that the observers and analysts can eventually attain insight into how
certain micro-structure arbitrage strategies work. That part is really the story of alpha
decay and natural information leakage and explains why purveyors of these strategies

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 6

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
need a continuous flow of new ideas. What formerly lasted years or months, now may
last only weeks or even days. Because of this dynamic, our original prediction on the
growth of multidimensional alpha was right at least in part. Are we at the end of the
high-speed arms race? Could we have finally reached the theoretical capacity at the tip
of the spear? Could there be only one firm or a few firms capturing all of it? And if
so, whats next?
2014 Quant Survey
In March 2014, TABB Group conducted a survey of market participants targeting
mainly quants, but generally any informed observer of the evolution of quantitative
trading strategies. Exhibit 2 is a breakdown of firm types from this survey.

Exhibit 2
Survey Methodology Breakdown of Firm Types (N = 51)

Source: TABB Group

However, before we dive into the newest clues to the evolution of quantitative trading,
it would make sense to showcase the prism through which we are observing this
landscape. It turns out that, though broad in terms of the possibilities, the scope of
those possibilities is finite. Identifying sources of theoretical capacity is where most
quant strategy discovery begins.
Quantitative Theoretical Capacity Framework
At its base, this story is about automating that which formerly has been done manually.
We have made this claim ad nauseam over the years, but it still bears repeating. First,
tasks in a workflow become automated, and then significant components if not, the
entirety of the workflow become automated. Following full workflow automation is
refinement, optimization, and higher-performance automation. In some product
classes, we are clearly at the stage of highest performance automation; and where
microscopic shifts in market structure implementations can open and close alpha
capture opportunities in an instant.

Where do quants look for alpha? The optimal target exhibits a convergence of factors,
but generally comes down to two: liquidity and level of automation. (A third factor
complexity makes the scenario even more ideal, but we will get to that later.) Those
products that are liquid and highly automated represent a breeding ground for high

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 7

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Sharpe Ratio (SR) strategies. This means that smaller players can leverage a small
amount of capital for high returns with low risks (see Exhibit 3).

Exhibit 3
Quantitative Strategy Framework



Source: TABB Group

The fact that quant equity strategies can have little or no exposure at the end of a day
makes all the difference in the world; since holding positions overnight creates margin
requirements and an exceedingly more complex level of portfolio construction, going
home flat is a huge deal.

But as competition has heated up, and the arms race of latency-sensitive infrastructure
relative to the perceived theoretical alpha has shifted, quant firms have been seeking
new sources of performance. This means migrating beyond the high-capacity/highest
Sharpe Ratio products (blue zone, above) closer toward the fringes of capacity and SR
(green zone, above). Translated, this means new regions, products and asset classes.

This was the claim we made in mid-2011 and more specifically in mid-2013: that
quant firms would need to look in new product classes, asset classes, and regions for
quantitatively-harvested sources of alpha. Turns out, the Roadrunner already was a few
steps ahead of us.

In the next 3 sections of this report, we showcase and respond to the findings of
TABBs March 2014 quant survey. In the 4
th
section, we highlight publicly available data
from two leading high-speed trading firms that ultimately complement the survey
findings. The 5
th
section summarizes TABBs latest views on the state and future of
quantitative trading and methods.


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 8

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Ingredients for Special Sauce
The components of competitive advantage or special sauce, in the vernacular for
quantitative trading firms are the most intensely discussed and often misunderstood
topics on the global markets landscape. This is the part of the story where the survey
yielded some of the most fascinating results.

First, as background, our line of questioning divided the sources of competitive
advantage framework into four components: human capital, software and processing,
data, and hardware and infrastructure. When asked to rank these components, survey
participants were clear in their opinions overall. Human capital translated by TABB as
developer, analytical and management skills is the most important ingredient for
special sauce in quantitative trading/methods; and, surprisingly, hardware and
infrastructure the stuff that is perceived to deliver the speed was ranked last.
Software and processing vs. data as competitive components were ranked almost
evenly (see Exhibit 4).

Exhibit 4
All Survey Participants - Sources of Competitive Advantage (n=27)


Source: TABB Group

These rankings were different than what we might have originally expected. However,
looking more deeply into the results, proprietary trading firms expressed differing
opinions from the averages which were more in-line with our original expectations
(see Exhibit 5, next page).
















2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 9

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Exhibit 5
Proprietary Trading Firms - Sources of Competitive Advantage (n=14)

Source: TABB Group

Prop firms ranked hardware and infrastructure as the strongest contributors to special
sauce, with data a somewhat distant last, on average. TABB Group believes that the
overall variance in results is indicative of the underlying variance in strategies. We also
believe that attribution of competitive advantages to hardware and infrastructure is
often a misinterpretation of the value of human capital; that human capital influences,
both positively and negatively, the inherent value of hardware, infrastructure, software,
and processing. This is true, also of derived data / analytics.

When these results are taken in conjunction with the results from the expected change
(1-2 years from now) in sources of competitive advantage (see Exhibit 6, next page)
and overlaid with our experience with quantitative trading strategies (both in practice
and observation), we have the following reactions:

Hardware: Hardware and infrastructure is a declining source of competitive
advantage for a vast majority of quant players. TABB Group has been calling this
the democratization of infrastructure for the past year or more. With nearly plug-
and-play level ultralow-latency connectivity offerings from the likes of CFN Services,
Verizon, BT Radianz and others representing the growing performance
competiveness of third-party solutions as well as other platform offerings such as
UBS NEO quant firms that compete largely on speed will find hardware, though
still incredibly important, a declining source of competitive advantage. Only the
largest, leading firms will be able to continue to afford to deploy and maintain
mainly proprietary infrastructure and, even then, those firms could be challenged
by further declines in capacity for trading strategies using core products, asset
classes and regions. Going forward, quantitative firms will 1) off-load more of their
infrastructure needs to managed services providers, and 2) simultaneously migrate
more infrastructure to one focused more on computational engines able to execute
more complex calculations at much higher speeds than today. These developments
will become part of the new special sauce, and where connectivity, co-location and
even cloud grow in the realm of competitive necessities.

Human Capital: TABB Group believes that human capital should always be ranked
highest relative to the other components of competitive advantage, independent of
trading strategy or stage of workflow. Human capital is the one source of special
sauce that is least susceptible to democratization. This is where we found the
strongest shift in opinions in the survey data. Furthermore, TABB believes that
certain players high ranking of hardware and infrastructure is actually a
misinterpretation of the value of their human capital. With proprietary infrastructure

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 10

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
remaining a key competitive advantage by a declining list of firms, persistence of
this special sauce can only happen by virtue of the skills and creativity of key
employees.

Data: Though we might have expected a different outcome from these results, it is
true that data or at least, raw market data is the most democratized of the four
components in our framework. For certain asset classes and products in the most
developed regions, both historical and real-time market data is incredibly clean.
Scrubbing unclean market data used to be a competitive advantage. Now it is not.
Today, everyone has access to the same clean market data despite the fact that
direct access to many sources of market data is expensive. Custom developed
security masters the reconciliation of security ID taxonomies across vendors and
platforms used to be a competitive advantage, too. Now it is not. Along the road
ahead, new strategies will require a mix of new and different data sources. We can
detect a sense of this from a slight uptick in the importance of data in the forward-
looking results of the survey (refer to Exhibit 6, next page). Though they will lack
the velocity in message rates of most market data, these new sources many of
them from fundamental, economic and social media realms, among others will
represent new complexities on the basis of their diversity. And, of course, reference
data especially legal entity identifiers (LEIs) that are critical for use in fixed
income and rates strategies are essential for mapping context in inherently illiquid
products and markets. Harvesting more signal from more data will be among the
leading themes for new strategy development in the coming era.

Software and Processing: Like data, the ranking of software and processing as a
competitive advantage for quant strategies fell somewhere in the middle. TABB
Group interprets this ambiguity as a statement about the relatively limited level of
processing that actually occurs in the highest-turnover strategies. TABB Group
expects software and processing to become more important as the data complexity,
highlighted above, explodes in the development of next-generation quant
strategies.

Coordinated Framework: TABB Group expects the ranking of competitive ingredients
for next-generation quant strategies will be as follows: human capital,
software/processing, data, hardware/infrastructure. Truth be told, in an optimized
environment these factors should be more equivalent.


















2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 11

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Exhibit 6
All Survey Participants Expected Change in Sources of Competitive Advantage (n=27)


Source: TABB Group




2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 12

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Low-Hanging Fruit
TABB Group has been closely monitoring the diversification of high-turnover and the
broader spectrum of quantitative trading strategies for many years. Our claim in
2011 that ultrahigh-speed trading in US equities was saturated is further borne out in
our most recent findings. In fact, this sentiment finally may have gone mainstream with
comments like dont bother joining the high-speed arms race coming out at recent
industry conferences.

The difference between 2011 and now, however, is this: While US equities were
saturated where leading quant firms were often consuming the alpha of other slightly
slower quant firms other regions, asset classes and products were not nearly as
saturated. Some proverbial lower-hanging fruit includes the key dynamics of high
liquidity, high levels of workflow/trade automation, and lower competition perfect
targets with high theoretical capacity to extend quantitative methods from core
strategies and markets. The question now is whether any of these lower-hanging
opportunities still exists within the scope of ultrahigh-turnover strategies. Or has
saturation proliferated into other assets, products and regions? The answer to these
questions will have a strong impact on the next generation of quantitative research and
the development of new automated trading strategies.
Asset Class Diversification
Survey participants rankings of current and future comparisons by asset class support
TABB Group predictions from 2011 and 2013; that there continues to be a need for
diversification into additional asset classes, particularly fixed income (see Exhibit 7).
These results show that in the coming years, the industry expects declines in focus on
equities and other asset class (which we would bundle together as alternatives or
experimental assets, such as real estate or structured products).

Exhibit 7
Quantitative Strategy Rankings Current and Future Comparison by Asset Class

Source: TABB Group

Our survey finds clear expectations for increased focus on FX and fixed income, which
have the greatest potential (comparative theoretical capacity to equities
notwithstanding) among all asset classes. However, in the case of fixed income, we are

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 13

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
concerned that the pace of transformation to more automated trade workflows is slower
than expected or desired. The pace of adoption of electronic execution of a broad
spectrum of fixed income products (cash and derivatives) may not supply enough
opportunities when compared to the overall demand for new high-turnover strategies.
Product Class Diversification
Product class diversification is primarily focused on cash, futures and single-stock
options, where cash represents cash equities and equivalents, such as exchange-
traded funds (ETFs) and exchange-traded products (ETPs), but does not include cash
bonds and is presumed not to include spot FX (even though technically it is the ultimate
cash product) and where futures represent multi-asset exposures, including equity-
index, fixed income, FX and commodities. Since options on futures typically have low
liquidity, the options category here relates primarily, if not totally, to single-stock
options (see Exhibit 8).

As with expected asset class diversification (above), survey participants also expect the
focus on the more mature products in quant strategies (i.e., cash equities) and
experimental other products, such as swap futures, to decline, on average although
for different reasons. The issues in cash equities are well noted herein. Newer products,
including swap futures, are worthy of experimentation across the expanded spectrum of
rates products, but their current low level of liquidity and concentrated adoption may
not justify the investment right now. Cleared swaps are similar in terms of potential for
experimentation, but the timing might not yet be good and the average trade sizes
are still way too large for most quant shops.

The one exception in core cash equity strategies has been the intense interest in and
addition of ETFs and ETPs, which have highly similar characteristics to equities with the
added potential for index arbitrage-like overlays.

Exhibit 8
Quantitative Strategy Rankings: Current and Future Comparison by Product Class

Source: TABB Group

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 14

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014

As previously mentioned, single-stock options markets are not only fiercely
competitive, with a number of seasoned incumbents; they also represent uniquely
complex computational challenges. With the exception of index and ETF arbitrage
strategies, most other securities, including stocks and most futures, move
independently. Because of this, high-frequency quant strategies can be deployed
without regard to portfolio construction or portfolio risk overlays. (The primary risk
management tactic for high-turnover strategies with independent positions is a notional
exposure boundary for each position that is influenced by the expected liquidity
between a point in time and the end of the primary trading session for that day.)

On the other hand, options represent a constantly moving matrix of puts, calls, strikes
and expirations and all based on the relevant underlying movement. So in options
market-making, not only do strategies have the speed imperative; they also have a
computational imperative; and despite mythology about speed, the computational lift
for single stocks is nowhere near that of an options matrix.

As a result, it will be fascinating to see how well high-frequency quant firms born in the
equities arena enter and grow in the options arena. Other than fixed income (which is
transforming slowly), options may offer compelling potential, but represent markets
that are very difficult in which to compete and harvest alpha. TABB Group believes that
this is a space that could see some (M&A) transactions. So far, leading equity-centric
HFT quants have been able to migrate well into futures products, competing well with
futures-centric firms, such as Infinium Capital, Jump Trading, and Chopper, without the
need to partner. It remains to be seen whether the same dynamic will play out in
options. Maybe this is what Virtu is really after with its IPO a better currency for
transactions. That said, the jury is still out on how well New York-Chicago mergers
actually work in practice.
Regional Diversification
Regional diversification is a classic move to extend the theoretical capacity of core
alpha. The original options market making powerhouses of the 1970s and 1980s,
including OConnor & Associates, first took their special sauce from Chicago to New
York, Philadelphia and San Francisco. From there, they sought to OConnorize
derivatives markets in London, Tokyo, Sydney and elsewhere. Its a fairly well-worn
path.

So, too, have noted quant firms especially high-turnover quant firms marched from
Chicago, New York, Amsterdam and a few other locations to set up shop outside of
their home turf. Though often proving successful in highly developed markets
namely, US, (Western) Europe, and Japan liquidity and, therefore, theoretical
capacity falls off dramatically from there. So while there are numerous other markets
where core quant strategies will work well, the level of profitability may not justify the
technology investment in the current environment.

Survey responses suggest that there is movement of focus from developed markets to
developing and emerging markets (see Exhibit 9, next page). But these are longer-
term expeditions, many of which will require patience while markets, economies and
political frameworks evolve. Without strong core cash-cow strategies, some of these
regional extensions may not have the funding to stay the course particularly in those
scenarios where the regulatory rhetoric has turned inhospitable.



2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 15

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Exhibit 9
Quantitative Strategy Rankings: Current and Future Comparison by Region


Source: TABB Group

As we look back across developments in asset classes, products and regions, we cant
help but wonder if the saturation we saw in 2011, primarily in US equities, has spread
across the full spectrum of high-turnover strategies. If high-turnover strategies for cash
equities, futures and options in developed markets are pretty much at the point where
leading HFT firms are consuming the alpha of other HFT firms, then that is likely a
signal that the pure speed game is just about played out.

And, for what its worth, when highly experienced quant executives turn from discovery
and capture of new alpha to lobbying the politicians like some of our friends at the
Modern Markets Initiative (MMI) and KOR Trading we take it as a strong signal that
these sources of trading opportunities are nearing the tip of the spear.


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 16

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
High-Hanging Fruit
Shifts in quantitative trading strategies can be detected by shifts in the types of data
being consumed. The highest-speed strategies use only raw (and consolidated) market
data and relatively simple analytics, such as if-then, VWAP (volume-weighted average
price), and the occasional anti-gaming tactic. Theres no time to use anything else.

Beyond the bleeding edges of speed, however, the slower the turnover, the more other
types and sources of data are required. These sources can be broadly bucketed into
categories such as market data, reference data, fundamental data, commentary
(including traditional news, industry research, and social media), and the derived data
or analytics that flow off the back of all that raw data. The ranking of data
categories reflects the amalgamation of quantitative strategy needs (see Exhibit 10). In
the survey taxonomy, we assume that derived data, analytics and proprietary data are
largely the same, even though proprietary data might also include raw categories such
as trades, quotes, client data, or infrastructure monitoring and tuning output. Finally,
given the number of respondents that ticked other, we sense that there could have
been some confusion with this taxonomy, and therefore we have a good excuse to do
more surveying around data.

Exhibit 10
Rank of Data Source Types (n=27)



Source: TABB Group

TABB Group has the following reactions and comments on these findings:

Market Data: All strategies are fed in one way or another by market data
(independent of latency requirements), so it is no surprise that this category comes
out on top. It should always come out on top, with the possible exception of those
strategies including exotic products with rarely observable prices and that are
typically marked to models. Two themes that resonate the strongest today for
market data are: 1) breadth of coverage across regions, asset classes and products

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 17

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
since new strategies will be more multidimensional than ever before and 2)
something we are calling the long tail of hybridization, which is the long-term or
perpetual management of combined in-house and managed services/cloud
infrastructures for market data.

Reference Data: As we have detailed in recent research (Fixed Income Market Data:
Growth of Context, Rate of Triangulation, January 2014) reference data continues
to grow in importance, particularly as the pantheon of trading strategies becomes
more global and unified across traditionally siloed regions, assets and products.
Reference data is what creates these linkages, and as a result, the visceral nature
of context. Though security identifiers via security masters are among the
most developed of the reference data categories, legal entity identifiers (LEIs),
unique product identifiers (UPIs), and unique trade identifiers (UTIs) are where new
developments and new capabilities need to be focused given our outlook for new
quantitative strategy development.

Fundamental, Commentary and Big Data: Few quant firms have ever had to care
about fundamentals, commentary, or anything related to big data (other than the
explosion in message rates through about 2011). One possible exception to this
would be low-latency, event-based strategies that leverage a data category known
as machine-readable news. As high-turnover, market micro-structure arbitrage and
other latency-sensitive strategies became popular in the early 2000s, a massive
bifurcation of the quant community resulted: those engaged in the latency arms
race and everyone else. This bifurcation was actually the beginning of the 3
rd

generation of the application of quantitative methods in capital markets; the 1
st

generation being born as a result of the first market data feeds in the 1980s and
the 2
nd
generation being born in the merger of signal generation and trade
execution or, the heyday of statistical arbitrage in the 90s. As the quant
community embarks upon the 4
th
generation of quant trading what TABB Group
might suggest is the Age of Multidimensional Alpha they will care much more
about all sources and categories outside of market data (and in addition to market
data). And in that transition a mission for those who choose to accept it the
combined infrastructure and data management complexities will reach new heights.
The benefits are likely to be greater theoretical strategy capacity (via slower
turnover), but with lower Sharpe Ratios.

As for big data and social media: Dont hold your breath. Though the topic is
outside the scope of this report and discussed more specifically in other TABB
Group research it should be reiterated here that capital markets big data is small
relative to many other industries. However, it is incumbent upon all trading firms to
adopt new data management techniques and high-performance infrastructures as if
it were the biggest, since trading firms need to consume big data better than most,
even if they dont necessarily produce it. On social media: It is worthy of
experimentation and, yes, there are occasionally strong trading signals therein
but given the low signal-to-noise ratio, TABB Group does not see social media
becoming a mainstream component of quantitative trading soon. Despite the mass
impact it is having on human communications, social media is not yet mature
enough to be a persistent and reliable source of trading intelligence across a broad
range of securities and regions.

Derived Data, Analytics: Otherwise known as trading indicators, or signal
processing, what you do with the data the analytics has everything to do with
winning and losing. As quantitative trading firms evolve from pursuing the highest
turnover sources of alpha to slower turnover opportunities, their thinking will need

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 18

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
to migrate from more horizontal to more vertical. Or 2-dimensional to 3-
dimensional. Trading signals will have less to do with the history of independent
time series and more to do with comparisons across time series, similar to a
massive global, multi-asset matrix. Another way to think of this might simply be
independent versus relative value, except the new relative value will need to be on
an unprecedented scale, such as global relative value. In any case, analytics based
on correlations, liquidity, valuation, portfolio construction, and event-risk detection
will become more important than ever before. Much more on this and related topics
from TABB Group in future research.

These reactions are supported, in part, by additional findings from the survey. While
there are still those firms seeking more speed, the desire to slow things down is non-
trivial; 19.4% of respondents claimed their average turnover frequency (TOF) was
slowing. With this fact in mind, we would then expect portfolio construction to shift
from more asymmetric where positions are independent to more symmetric which
relies on portfolio construction and maintaining market neutralities since asymmetric
strategies most often rely on speed. This is also evident in the findings, where
symmetric strategies represented at least 29% of responses (see Exhibit 11).

Exhibits 11
Quantitative Strategies: Changes in Turnover Frequency and Portfolio Construction

Changes in Turnover Frequency Nature of Portfolio Construction


Source: TABB Group

We confess that there are some notable flaws in these findings, as well which have
more to do with survey construction. For instance, faster in the changes in TOF
chart above could mean several things, including slower strategies becoming faster,
but not at the same scale as the microsecond crowd. We confess this to emphasize that
understanding the quant landscape takes an exceedingly detailed understanding of tiny
clues; clues that cannot be extracted in a single, high-level survey and why
experiential overlays are critical to that level of understanding. To return to an earlier
metaphor, this is precisely why Wile E. has a tough time keeping up with the
Roadrunner.

Bottom line: It seems odd to say, but if the latest generation of high-turnover
strategies represented the equivalent of low-hanging fruit, then the next generation of
quant strategies will need to pick the fruit from much higher up in the proverbial tree.


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 19

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
A Rare Glimpse
It is extremely rare that the public gets a glimpse into the profitability, revenue
segmentation, or strategy breakdown of proprietary trading firms. In recent times, it is
not since the attempted IPO of Chicago proprietary options trading firm Peak 6 in
October 2001 that we have had such a look. However, in the past two years, the public
has been gifted unprecedented transparency into two of the worlds foremost
quantitative prop trading powerhouses: GETCO and Virtu. The first as a result of
GETCOs merger with Knight Capital, and the second as a result of Virtus ongoing IPO
trajectory (which has since been delayed. as of the publication of this report). Related
filings in both cases give us a taste for all three of the above: profitability, revenue
segmentation, and strategy breakdowns.

Both the words and the numbers in these filings tell a story that TABB Group has been
following for years. In the GETCO now KCG data (below), we can see how a high
dependence on Americas trading (most likely focused on US equities market making)
in the 2009 (and earlier) period led to regional diversification, mainly to Europe first
and then Asia-Pacific (APAC), which includes Singapore, Hong Kong, Japan, and
Australia (see Exhibit 12).

Exhibit 12
GETCO Form S-4 (12 Feb 2012): Trading Revenues (2007 Sep 2012, $millions)


Source: KCG, TABB Group

It is further evident from this data that overall profitability was highly correlated to
volatility first (2009-2011) and competition second (2012). With aggressive globally
coordinated central bank interventionism strategies ramping up in the post-global-
financial-crisis (GFC) period of late 2008 and early 2009, strategies that were
dependent on (higher) volatility suffered as global markets became awash with
volatility-crushing quantitative easing (QE). Note the positive impact on revenues in

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 20

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
2011, which TABB Group attributes to a correlation to the then looming threat of a US
default and the resulting spikes in volatility.

Furthermore, TABB Group believes that the nearly $300 million in revenue declines
between the 9 months ending September 2011 and September 2012 were due to
something more: competition. But we need not necessarily speculate on causes and
changes in tactics. Notable highlights from the GETCO/KCG filing deliver the answers:

Declines [in] revenues were impacted by lower overall market volumes,
reduced levels of volatility, increased competition and a higher percentage of
internalization in U.S. cash equity markets;

Headcount increases related to early investments in the Global Arbitrage, Mid-
Frequency and Options trading businesses; and

Colocation and data line expenses increased to $81.4 million in 2011 from
$55.2 million in 2010. The increase in colocation expense was primarily due to
the build out of new colocations in Mahwah, New Jersey, Basildon, United
Kingdom, and Tokyo, Japan, as well as higher vendor rates for data and
services.

Bottom line: TABB Group believes that the market ecosystem was no longer supporting
the same level of theoretical capacity for all high-turnover participants, and the
competition was becoming increasingly fierce.

Finally, although the reasons may have had more to do with operational retrenchment
in the wake of the GETCO-Knight merger in mid-2013 than the threat of competition,
we note that the KCG annual report for 2013 shows non-US revenues declining
dramatically, from 46.5% of total KCG revenues for 2011 (which is an almost exact
regional split of the GETCO-only numbers) to 18.9% of total revenues for 2013.
Whether for enhanced operational integrity, the threat of increasing competition or
some combination of these and other factors, TABB Group believes that there was a
retrenchment to the core regional strategy in 2013.

Exhibit 13
Virtu S-1 (10 Mar 2014): Total Adjusted Net Trading Income (000s, USD)

Source: Virtu, TABB Group

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 21

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
But then, we turn to Virtu. The Virtu data, made public just this past March 2014,
continues the narrative (see Exhibit 13, previous page). This is an example of a firm
with a more diversified portfolio of strategies even if they are variants of high-
turnover, latency-sensitive strategies.

It is because of this diversification that revenues continued to increase for Virtu in the
post-GFC period, particularly in the past 3 years, when many high-frequency players
were suffering or exiting. Interestingly, with declining rates of growth in its core
Americas equities strategies and flat-ish growth in its global commodities futures
strategies, Virtu continued to more than make up for lost ground due to success in
areas like FX, non-US equities, options and fixed income (see Exhibit 14, below).

Exhibit 14
Virtu S-1 (10 Mar 2014): Total Adjusted Net Trading Income (000s, USD)

Source: Virtu, TABB Group

With growing competition and strategy maturity in global equities and futures
strategies, the target for further high-turnover strategy growth will need to be in fixed
income, including the ongoing Dodd-Frank-led transformation in the swaps market.
(But will this be enough to overcome declines in other products, assets, and regions?)

Also, bear in mind that this diversification is likely to be difficult in options strategies
which means single-stock options strategies. While Virtu may have discovered a unique
and persistent speed advantage across products, regions, and asset classes (an
incredible feat, representing the pinnacle of what TABB Group believes is its core value
proposition), it will be met with formidable competition from firms such as Wolverine,
Susquehanna, Ronin, IMC, Optiver and a few others.

As a result of all this, TABB Group believes that the timing of Virtus filing is indicative,
if not prophetic, of more persistent market shifts. Whether Virtu ever goes public (given
the current delay), there is some rather stark writing on the wall: Chances are that the
expected gains in high-turnover fixed income strategies will not make up for the
stagnation or declines in everything else for the foreseeable future.


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 22

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Virtu has taken diversification in high-turnover (market-making) quant strategies to
new levels. According to its recent S-1 filing, Virtu makes markets in more than
10,000 securities and other financial instruments on more than 210 unique exchanges,
markets and liquidity pools in 30 countries around the world (see Exhibit 15). Chances
are that this is the most comprehensive list of venues less the specific dark pools
where high-frequency trading is even possible on the face of the planet today.

Exhibit 15
Virtu S-1 (10 Mar 2014): Liquidity Venues



Source: Virtu, TABB Group



2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 23

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
Conclusion
This report has focused on high-turnover quantitative strategies. The main reason for
this is to leverage newly available intelligence to build a case for where TABB Group
believes the broader quant landscape is in its evolution today and why it may be in
transition to a new generation of indistinguishable quantitative strategies and methods.
We wish the bias was not like that, but the fact of the matter is and despite the level
of opacity and complexity embedded in many of its methods that the fascination with
speed has made this segment of trading strategies much more observable than our
interpretation of the strategies in the next generation. Its paradoxical, but with
growing propeller head ubiquity, quantitative methods will become infused into just
about all strategies and workflows and therefore, may in fact disappear from distinct
observation. The following is a summary of that hypothesis:
End of a Generation
Market microstructure arbitrage strategies otherwise blanketed less accurately by a
term that we have tried to avoid using here high-frequency trading (HFT) have
reached their full potential in the current global markets environment. The saturation
effect has spread from equities in developed markets to just about everything else.
Several factors could change this assessment for the better, including material upticks
in volumes and/or volatility in the most developed markets. This means large markets
with the most advanced high-performance market infrastructure. Smaller markets with
high-performance market infrastructure add incremental opportunity, of course. But
they dont tip the scales or change this assessment materially. Only the BRICs
represent large untapped potential here. But the level of economic, political, technical
and even cultural change necessary to make these markets into hospitable Petri dishes
for growth of high-speed quant trading means that the transformations have a very
long horizon. In other words, the BRICs dont represent a short-term bridge to the
near-term future success of quantitative trading.

Meanwhile, the headwinds are enormous, and growing. Believe it or not, echoes of the
GFC continue to reverberate in the bowels of G8 markets. Lets start with market
demographics: Diverse demographics of the pre-GPC period a requirement for a
healthy market ecosystem, including high-speed market-makers have been replaced
with an aggressive bacteria, otherwise known as coordinated central bank
interventionism, or by the more publicity-friendly term, quantitative easing (QE). (See,
for example, the trajectory of CNBC viewers to fact check on this one!) The goal of
these efforts is to squash any signs of volatility or deflationary movement that would
diminish (primarily) equity markets as a tool of public psychology management (aka,
the driver of consumption and spending). This is the part of the story that we did not
see all that clearly in early 2011. Moreover, TABB Group doesnt see a logical end to
this policy. (By the way, dont be fooled by the reduction in monthly asset purchases by
the Fed and its allies. They have their own quants now. And these quants know how to
use market jujitsu with less and less money to influence trajectories of market indices
and force the algo community to drive individual security correlations back into line.)

One of the symptoms of low volatility has been to destroy the performance of other
active trading strategies, thereby leaving primarily high-speed market makers, long-
term buy and hold investors, and the central banks (led by the US Fed). For purposes
of quantitative strategy analysis, the knock-on effects of these shifts in market
demographics and volatility have basically resulted in an accelerated competitive frenzy
known to many as the low-latency arms race where leading HFTs have essentially
been eating their own. This has led to the tip-of-the-spear phenomenon we

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 24

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
hypothesized about in mid-2011. (Note: As if exactly on cue, Infinium Capital
Management - of HFT and 600 West Chicago Avenue fame announced during the
development of this research that it is winding down.)

Without new and different players at the proverbial poker table, as many players
crapped out, these leading firms dwindling in numbers as they may be have needed
to diversify into new regions, asset classes and products. From the available data, we
can now observe unequivocally that this was already happening over the past 5 years.

Finally, notice that nothing has been said here yet about regulations (or the less
obvious fact that some of the greatest operators in quantitative methods have turned
to lobbying). Of course, this is a significant factor in the enormity of the headwinds
high-speed quants face. Suffice it to say, that factor is not positive.
Beginnings
Today, TABB Group believes that the highest-speed strategies are approaching the
limits of their native capacity in the current environment, particularly in cash equities
and futures-based strategies in equity-index and commodities. FX market-making
strategies across both futures and spot likely have room for additional growth,
particularly if wholesale blocks (in the spot market) can be accessed through liquidity
venues that are accommodative to low-latency market-making.

This leaves fixed income and options (primarily single-stock options) open for
interpretation. Evidence shows that leading quant market-making firms are heading in
this direction, but the revenue accumulation is coming off of a small base. TABB Group
believes that these market opportunities are unique in that there are pricing and
valuation components. Unlike equities, as previously mentioned, risk-taking in these
products and asset classes cannot be assessed independently. They are not horizontal,
but vertical. Relative-value techniques are critical to success here, and the HFT
community doesnt have those skills (yet).

The good if not, great news here is that when the high-performance infrastructure
of the current generation of HFTs (which we call Gen 3) is integrated with a global big
data vision, some exciting things could happen. Business combinations that merge
these two may make a lot of sense in the near term. Meanwhile, arguably the true
quant legends, including Renaissance Capital and D E Shaw, are already there and
have been there since early in the 2
nd
generation of the 1990s.

This 4
th
generation of quantitative research and trading is not a greenfield or even a
tilted battlefield. The pattern recognition and performance needs are much more
complex than in the high-frequency era. Firms singularly focused on speed, with less
diversification today, are either already gone, or in the process of making an impossible
transition. For those firms that can make the transition which means a global HPC
infrastructure capable of identifying and harvesting volatility spikes in a relative-value
manner anywhere, at any time, in any automated product workflow using magnitudes
more data than today may successfully become a long-standing member of the
multidimensional alpha era. But please note that there are legendary quant firms that
already have been playing there for, oh, about two decades.

For those firms that dont want to compete on this basis, enterprise risk analytics for
market, credit and operational risks; surveillance; and new software-defined
infrastructure (SDI) to combat inherent technical fragmentation in our business are use
cases that remain sorely in need of quantitative attention. Once that happens,
quantitative methods will be sufficiently ubiquitous to be indistinguishable.

2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 25

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014
About
TABB Group
TABB Group is a research and strategic advisory firm focused exclusively on capital
markets. Founded in 2003 and based on the methodology of first-person knowledge,
TABB Group analyzes and quantifies the investing value chain, from the fiduciary and
investment manager, to the broker, exchange and custodian. Our goal is to help senior
business leaders gain a truer understanding of financial markets issues and trends so
they can grow their businesses. TABB Group members are regularly cited in the press
and speak at industry conferences. In 2010, TABB launched TabbFORUM, the online
capital markets community for peer-to-peer contributed opinion and analysis covering
current industry issues, tracked daily by 23,000-plus professionals. In October 2013,
QuantFORUM, a new online channel for the global quantitative investing community,
went live. For more information about TABB Group, visit www.tabbgroup.com.
The Author
E. Paul Rowady, Jr.
Paul Rowady, Principal and Director, Data and Analytics Research, joined TABB Group in
2009. He has more than 24 years of capital markets, proprietary trading and hedge
fund experience, with a background in strategy research, risk analytics and technology
development. Paul also has specific expertise in derivatives, highly automated trading
systems, and numerous data management initiatives. He is a featured speaker at
capital markets, data and technology events; regularly quoted in national, financial and
industry media; and has provided live and taped commentary for CNBC, National Public
Radio, client media channels and others. With TABB, Pauls research and consulting
focus ranges from market data, text analytics, social media impacts and data
visualization to OTC derivatives reform, clearing and collateral management; and
includes authorship of reports such as Patterns in the Words: The Evolution of
Machine-Readable Data, Enhanced Risk Discovery: Exploration into the Unknown,
The Risk Analytics Library: Time for a Single Source of Truth, The New Global Risk
Transfer Market: Transformation and the Status Quo, Real-Time Market Data: Circus
of the Absurd, and Quantitative Research: The World After High-Speed Saturation.
Paul earned a Master of Management from the J. L. Kellogg Graduate School of
Management at Northwestern University and a B.S. in Business Administration from
Valparaiso University. He was also awarded a patent related to data visualization for
trading systems in 2008.


2014 The Tabb Group, LLC. All Rights Reserved. May not be reproduced by any means without express permission. | 26

Propeller-Head Ubiquity: The Next Generation of Quantitative Trading | July 2014










www.tabbgroup.com

New York
+ 1.646.722.7800
Westborough, MA
+ 1.508.836.2031
London
+ 44 (0) 203 207 9397

www.tabbgroup.com

New York
+ 1.646.722.7800

Westborough, MA
+ 1.508.836.2031

London
+ 44 (0) 203 207 9027

Potrebbero piacerti anche