Professional Documents
Culture Documents
VavNE H. \IecNsR n defining Best Execution for securities is not so long ago that it was rarely forurd in
is a senior advisor at iTG trading, the Best Execution Taskforce of investment management. A rnore colnmon
Solutions Nefwork in Los portfolio nvrnager rnight have
the CFA Institute (formerly the Associ- resporlse frorr_ a
Angeles, CA.
wwagner@itginc.com
ation of Investment Management beerr-and in one case actually was-"I pick
Research) was careful to refer to it as a process, stocks. The rest is just plumbing." We are not
Maxx Envanus not an outcome. To many, this is a disquieting likely to hear that response again, for three
1s--at defirution, for it does not lead to easy agree- reasons:
ITG Soiutioas, Inc., ment. Indeed, some have argued for a simpler
in
---.. definition which, even though krrown to be 1. The Securities and Exchange Comrnis-
Emdl
inferior, is at least objective. lJnfortunately, sion {SEC) considers Best Execution a
this simpler defirution runs the danger of sub- part ofthe fiduciary duty ofan invest-
stituting appearance for realiry. ment manager;
In fact, the definition for "best execu- 2. As a result, instirudonal and even retail
tion" is not that different from the "best im/estors are increasingly atnmed to em-
efrorts" definition of prudence. As Einstein cient, effective investment rnanagement
said, "It is important to keep things as simple operations; and,
as possible. But not more simple." 3. More to the point, ,pplyrt g a process
If we think atrout best execution as a analyses to the implementation ofinvest-
process, we are inevitably 1ed to the work of ment ideas is a proven means ofimproving
Peter Drucker. Drucker devoted much of his investment performance.
liG to the issue of productivity in knowledge
and service work and phrased it succinctly: Let's take a look at the implementation
"The first question in increasing productivity of investment ideas, taking a page from the
in knowledge and service work has to be: Drucker viewpoint: In the following sub-
What is the task? What do we try to accom- sections, we will identify and discuss these six
plish? Why do it at all?" Of course, the ques- steps:
tion of "why do it at all?" takes us back to the
prudency principle that investment ryutnagers 1. Establish goals: What. are we trFng to
have lived comfortably under for well over a achieve, and how wiil we define success?
century; that is, the primacy of the client's 2. Define the process: Who and what are
interests. involved in the process, and why? We
Whiie the application ofa Drucker-iike model the execulion process, then set up
discipline might seem completely obvious, it data collection.
IF BEsr ExscurroN rs A PRocEss. WHAT DoES TFrAT PRoCESS LooK LIKi? SUMMER 2007
process-the link berween the portfolio manager' trader, to trades) above may be not one event but a stream of
and broker-isn't as clear cut. Flowever, with the aid of events. Thus the amount of inforrnation to be processed
data readily available (we say that with slight tongue in can be quite voluminous for a busy desk.
cheek) from Order Management Systems (OIS), the exe-
cution process can be modeled. ANALYZE THE DA?\
More importantly, the eft"ort is worth it. Think in
terms of the batting coach who s'orks with baseball Analyzing the data is a complicated, but large$ auto-
players who've akeady made it to 'the show.' By con- mated, task. Some steps that need to be taken are:
stantly analyzing swings, the coaches heip talented athletes
improve their game even more. Above all, the batting 1. Check the data for egregious errors; for e.g., incor-
coach demonstrates the advantage of an independent, rect trade dates due to cancels/c<,rrrects or incorrect
objective view with no stake in "things are just fine the prices and commjssions.
',vay they are." 2. Create a time line rvhich computes the state of the
Now we're ready to follow Sagan's advice to count process at any time that the stafus changes:
arrd measure. The frst step in counting is to record and
a. At every point in time, how much was ordered?
retain. 'We are far less interested, 1br example, in whether
b. F{ow much r.vas released to the trader?
the trader thinks the broker did a goodjob on thds trade
How much was executable bv a trader?
than we are about the nuts and bolts ofhow the trade was
How much was released to the broker?
accomplished: what, when and why was the trade done?
'We How much was executed, and how much
need to know the followinE details:
outstandins?
1. At what time rvas the trade released to the desk? 3. Embed the time line data in the record of the trading
2. Was the order complete and executable at that point environment at the time of action to address ques-
in time? tions such as:
3. What instructions were attached to the trade by the
portfolio manager? a. 'What was the price trend at the time of execution?
4. Were compliance checks performed? Were there b. What:vas the availability of liquidity-the iimit
any flags? order book-at the time?
5. To which trader was this trade assigned? c. Check the data for plausibfiB; for e.g., was the
6. Which broker was selected, and why? execution price outside the BBO at the time of
7. What were the instructions to the broker? trade? Correct it as necessary.
8. When was the order released to the broker? Ifit was 4. Compute the cost of each individual trade. ffi/e dis-
parceled out over time, when and holv much? cuss cost methodologies in a later section.)
9. \X,tat venues were used, and which algorithms? 5. Contextualize the cost in ternx of a comparative
10. \X4ren were the pieces executed, and at what price? benchmark that addresses the question: What should
11. When was the order completedand/or abandoned? a trade of this nature cost?
12. How rn'eil did the idea work out as an investment? 6. Store the data in a manner that facilitates aggrega-
W-ere we able to pay the transaction costs and still tion into meaningfui sub-sets; for e.g.,
generate satisfactory returns?
13. To which accounts were the traded securities a. Controlling party: m:rnager by manager, trader
assigned? by trader, broker by broker, market by market
b. Buys vs. selis
Collecting all this detail sounds like a tail order fbr c. Order instructions; for e-g., if an aigorithm was
an active desk, but most ofthe information is readily avail- used, which one, and why?
able in a well-tuned Order Management System (OMS), d. Supply/demand imbalance: favorable, neutral or
especially one tightly wired to FIX connections. adverse?
In a large shop with multiple fimds, managers and e. Consumption of liquidity; percent ofvolume
traders, each ofthe data collection points (orders leading f. Company capitalization
-WHAT
IF BEsr ExEcurroN rs A PRocEss- DoEs rFLAT PRoCESS Loor Lx-e? SUMMER 2OO7
perforrnance will be affected. One very aggressrve man- us: often the solution gets adopted without consclous
ager chose to adopt the policy of waiting six weeks to a\Mareness. A TCA report shows improvement in a crit-
decide whether the stock should still be held in the port- ical area, but congratulations are met with a puzzled "We
folio, and found that it added vaiue by avoiding a panic didnt do anything differently'' What! going on here?
sale. Professionals, especially traders and athletes, are
Enc ourage out-of-the-box thinking. tJnless there accustomed to instant integration ofnew inlbrmation. A
is something totally dysf,rnctional in the orgarrization, we cornerback may get surprised by a receiver's move, and
are dealing with an organization that has already derived adjusts without thinking about it. A skier senses the snow
a solution for the most eft-ective use of people, perceptions, conditions and unconsciously adjusts. A trader hears an
comrrnrnications, processes, and technology. If we take idea and that idea immediatelv becomes his own.
those conditions as unchangeable, we cannot expect to
come to a significantly different solution. The operative REVIEIry AND REVISE
question is "what can be changed to make the process
more effective here?" "Try harder" is not an acceptable The final step is to start ail over again:
answer.
Make trading intelligence a priority. Keep an open 1. The success of solutions worked out in last cycle
charurel to nerv ideas from other traders, other rnanagers, need to be assessed and, ifnecessary, adjusted.
the industry press, and TCA providers. 2, Problems ignored in the last quarter may become
Ultimatel-v, all efGctive solurions will involr,'e change. candidates for new areas of improvement in this
It is the job of the CIO to assure everyone that the orga- quarter.
nization executes the changes that enhance the perfor- 3. New investment opportunities may require adapta-
mance outcomes. tion in the way they flow through the organization.
The proposed solutions should be written down in 4. New technologies, data, or communication faci-li-
the minutes of the meeting, and should be reviewed in ties may create opportunities to enh;mce effective-
detail when the next reoorts are available. ness in certain situations.
IMPLEMENT THE SOLUTIONS Confession titne: For years clierrts have asked me
whether they can expect to see the results in their per-
All the steps above are all necessary to the process, formance. Conersely, does TCA harre to be taken l*g"ly
but nothing happens until something happens. Keeping as an article offaith? Except in isolated and usually egre-
the organization on track to implement the soiutions and gious situations, it is difficult to detect any performance
assess their eft-ects is er,'eryone's responsibiliry. enhancement. There are simply too many moving parts.
While not incompatible with the compliance Best Execution processes can take some of the ran-
process, Best Execution is about capturing maximal invest- domness out of the result, and can reduce some of the
ment return. The problem is solved top-down, not slippage that leads to poor performance. But too many
bottom-up, from a set of prohibitions and reporting other sources ofvariation are perpetually out ofour con-
requirements that can only define a minimal acceptable trol. We need to take solace in the belief that better
level to keep the organization from crossing the line. But processes will lead to improved results.
a good compliance oficer builds the guardrails, then lets
the grand-prix individuals exercise their skills unham-
pered by over-zealously applied restrictions. To order reprints o-f this article, please contact Dewey Palmiei at
dpalmieri@iij ournals.eam or 2 1 2 - 2 2 4- 3 57 5
Having worked with many otgarizatrons to irnprove
execution processes, one curious thing has ah,vays struck