Bsilding COMMENTARY 71

Belter

9rorrp
Perf ottnqnGe

JULY
WHAT DO THE GOOD DESKS DO?

2OO2

A frequent question asked by traders is "What can we do to improve"? In this Commentary we search for answers by looking at the "good" desks and delving into the data for telltale signs of how good desks

differentiate themselves from the run-of-the-mill.
By our measurements, good desks at growth firms enhance performance by 42bp,62bp for diversified firms and a whopping 171bp for value firms. No wonder CIO's and head traders are keen to improve their desksl

The Settcm Line: How DO Good Desks Add Value?
We were able to identify five key characteristics that differentiated the good desks:

factors

magnitude of assets being traded. Fortunately, we have a standard benchmark: the Plexus Average Execution Gainiloss or PAEG/LrM. PAEG/L is a measure of expected costs for trades of varying

. Most importantly, good desks know how

to

PRIORITIZE. They quickly identify the dangerous trades and focus on a strategy for doing them first. What makes a trade dangerous is to be on the wrong side of a liquidity imbalance, a situation in which delay can cause the value of the decision process to rapidly slip out of reach. As Plexus has said before, "do the hard trades first."

of difficulty. By definition, a desk that consistentlv scores well relative to the PAEG/L is a quality deik. To qualify, desks had to pass four heuristic tests:
.

. . .

.

They know how their managers think; in general and soecific to the individual trade. Qualities that make a value trade desk good differ from those for a growth desk, yet both share the advantage of style consistencv. Diversified desks that handle both growth (licjuidity demanding) and value (liquidity supplying) situations must deal with mixed signals, leading to tougher prioritization decisions. They attune the mix of brokers to the trading needs. To increase their clout. too desks have shrunk broker lists and concentrate on trading through brokers who understand and can deliver the quality of executions consistent with their investment style.

Thev had to rank in the too third of costs vs. PAEG/L in elch of the four quarteis in 2001; They had to rank in the top half in each quarter of 2000, and. New clients with only '1 year's data had to be in the top quartile for each of the past four quarters.

Desks with non-compatible characteristics were
eliminated.

Separate samples were determined for investment style: Growth, Diversified/Core, and Value. The
following table shows the sample characteristics:

.

Number of Desks by Category
Style
Desk Size

Market Capitalization
Large
Small

Growth

Diversified

Value

Large

Small

.
.

Nothing stays the same very long in today's markets, so top desks are constantly innovating and experimenting, trying new trade routes and refining desk processes.
The best desks have the confidence of their portfolio managers and are seldom encumbered by confusing orders or manager imposed price limits.

22

44

15

44

37

66

78*

*

15 desks were small cap only

What Do Typical Orders Look Like?
Managers who follow a growth strategy tend to seize upon favorable news and thus often trade in high momentum conditions, competing with other growth

The Good Desks?

WhatDoes "Good" Mean?
a metric. This is not as easy as it seems, because each desk has a workload derived from the unique nature of the underlying investment process and the
In order to identify a desk as a superior desk, we need

managers

for

available liquidity. Holdings

are

concentrated, resulting in larger orders, but since the names are 'hot,' liquidity is high. In contrast, orders from value managers tend to be a larger percentage of trade volume, yet often represent smaller orders in

out-of-favor stocks. Orders to diversified trade desks are Goldilocks, neither too large nor small, too illiquid, too strong nor weak, although individual orders can deviate sign ificantly.

Past experience indicates that PAEG/L is a good predictor of investment style: growth desks have the

can use momentum, the prime determinant

higher benchmark costs, other things equal. Thus we of PAEG/L, to categorize desks. We identify the top momentum quartile as growth, the bottom as value, and the remainder as diversified. The following tables detail the trading characteristics and costs by investment type over that past two quarters. Shares, percent volume, momentum and costs align with expectations. The following table summarizes all managers in the Plexus universe - not just the superior ones.

Note the large percentage of trading in adverse momentum conditions for growth desks. We are surprised to see value managers' desks perform as well as growth desks for adverse orders. This may reflect order size: 33,000 share average for value desks vs. 92,000 shares for growth desks. When providing liquidity - favorable momentum - value desks are adept at slowing down to trade at better
prices. Trading delays generally lead to high adverse trading costs, but value desks offset slower execution with much lower brokerage costs.

Comparison of Trade Gharacteristics by Style
Shares (000)

Finally, value desks beat PAEG/L for large orders, which may reflect easier market conditions more than desk quality. Trading size is usually not a punishing issue unless momentum is present. In favorable conditions, value desks capture price concessions. Contrarian firms, as we would expect, are especially adept at this.

Volume Mkt Cap
Yo

Mom,
Yo

Cost
(bp)

PAEG/L (bp)

Value

How Do the Besf Desks Compare?
The next section features the top desks in each style. We looked for those desks that consistently show better than average trading costs relative to
PAEGi L.

Growth

56

39

37

2.7

-121

- ILJ

Diversified

38

37

42

07

-72

Value

30

50

38

-0,7

Best Desks by Style

The table below reveals that value desks outperform in all kinds of market conditions. Growth desks, in contrast, focus on liquidity and don't seem to know ouite what to do in the infreouent situations where momentum is in their favor.

# vs, Peers

Timing
bp

lmpact
bp

Total Cost bp

PAEG/L bp

Trading Quality in Market Gonditions
Adverse Momentum
Value Added bp

Growth

I of22

-27

-93

- 411 tLl

Favorable Momentum
Value Added bp (% of Dollars)

Large Trades over
100 K Shares Value Added bp
(% of Dollars) Diversified 6

of44

a1 -L I

4A

-96

(% of Dollars)

Growth

+32 (340k)

-83 (10%)

+1 (6e%)

Value

6 of 15

+31

+38

+51

-66

Diversified

+20 (19%)

-48 (13%)

+1 (56%)

Value

+33 (13%)

+35 (18%)

+77 (48olo)

The first column shows the number of desks by style that qualified as a 'best desk.' 40o/o of the value desks, 36% of growth, and only 13% of the diversified desks achieved a'best' ranking.

Where Top Desks Shrne
The table below shows the key data of this Commentary. lt contrasts the best desks with the
remaining desks of the same investment style. The table analyzes how the desks respond under the contrasting conditions of supplying and demanding liquidity. lt also shows how the desks perform under two measures of difficult trading conditions: orders exceeding a day's volume and orders exceeding 250,000 shares.

Top Diversified desks: Few Diversified
significantly outperform

or

desks

underperform the

PAEG/L benchmark. However, six desks stand out. Most quantitative clients fall in the diversified group, but none of the best diversified desks are ouants. Other key diversified attributes:

.

Where Top Desks Shine
Adverse Favorable

Larger than average orders, but below average trading activity in dollar terms. . Much higher quality trading in both adverse and favorable situations. Top diversified desks stand out when trading large, illiquid orders. . Two thirds of the superior desks measure and rank both their traders and brokers. Meaningful feedback results in changes to their processes; e.g. shifting from low imoact brokers to those that can deliver
SIZC.

0ver 1 day's
Volume

0ver 250K shs

Value Added bp

\/^,uc nuuuu h^ \/:lrLo Addod hn Value Added va ^ A,t,.t^.t ug
to" nf flnllarc\ /0/ ^{ n^il^,^\ /O
\

bp

(% of Dollars)

U

UUIId D/

/0/ ^{ n^il^,^\ | /0 ur uuildr)/

Top Value desks: This group of desks stands out relative to all desks in the Plexus universe, averaging +117 bo better than the PAEG/L benchmark. The best
value desks:

Growth Desks
Best
Atl

.
+18 (13%) -30 (18%) +47
(500/o)

'92

(36c0)

-81 (e%)
t

+32 { 34r:

-83 (10%)

+5 (467,)

Diversified Desks
Best All

+77119%) +20
(1 996
t

-21(3ot;)
-48 (13%) Value Desks

+85 (270k)
-5

+69 (45%) +7 (33%)

(7ah)

Best
Atl

+72 (12'k) +33 (13%)

+'189 (19%)

+1

98 (2996)

+280 (30%)
+ 10Q /?Qo/^\

+35 (19%)

+86 (220h)

Trade nearly twice as much as the average desk, contrary to lower activity levels for top growth and diversified desks. . Trade relatively illiquid orders, but not necessarily larger-than-average orders. . Show very disciplined and predictable styles. They limit their broker relationshios and make extensive use of low impact ECN's. . Show relatively large inter-day timing gains, lf prices move in their favor, they often delay trading. Desoite this. none of the desks show sionificant Oppofiunity costs. Overall, growth and value desks appear to have an advantage. One key difference is how these desks select and use brokers. The top growth desks emphasize full service brokers, while the top value desks emphasize ECN's or brokers who are capable of filling large sized orders at a discount. The top
diversified desks have winnowed their broker lists to fit

Top Growth desks: There is no difference in dollars traded, nor in order profile compared to ordinary growth desks. Best desks show: . Low timing costs when accessing liquidity, regardless of size. . Attention to prioritization, ensuring that orders with high momentum are traded quickly. . An above average numbers of traders. . Few manager-imposed constraints. All desks contend with some level of broker direction, but most superior desks face little manager interference. . Undirected broker use is more concentrated among fewer brokers, with emphasis on full service brokers. . Early adaptation to new broker routes, recently emphasizing Bloomberg and LiquidNet, although history shows experimentation with failed systems.

with their needs, but most diversified desks still use more broker than averaoe.

Other lmpacts on Trading Quality:
We performed separate analyses to address
two

specific issues of interest to our clients: . Do desks who trade linked buys and sells operate at a disadvantage? The answer appears to be yes: linked trading desks lagged PAEG/L by 17 bp vs. a gain of '13 bp for all other desks - a 30 bp gap! We showed above that the universe of trading desks

adds value in adverse momentum and loses value in favorable situations. But the linked order desks lost value in both momentum conditions despite trading smaller, more liquid orders. The problem seems to be an unwillingness to step up in more difficult trading situations. The data shows that linked trades are prone to complete the easier trades that CAN be completed and compromise in the more difficult

Conclusions
Ed Deming, the quality control guru, once said "Quality is never an accident - it is always the result of high intention, sincere effort, intelligent direction and skillful execution. lt represents the wise choice among many alternatives."

trading situations that SHOULD be done
Commentary 56).

(See

. Do larger trade desk do a better job? Well, Niederhoffer says "To be best at something sometimes. Large desks have an advantage in consistently, you have to be economical, using the adverse momentum conditions due to their bought- least amount of input to accomplish the greatest and-paid-for access to full service brokers. Con- output." That's what the good desks do: identify the
versely, small desks show a significant advantage in favorable momentum. When providing liquidity,

So it is with trading. The purpose is to mobilize investment ideas with the least slippage. Victor

large desks prefer to capture !iquidity even if it requires a higher price. Small desks orders are smaller in dollars but in smaller-cap names: they exhibit less trend following or herding behavior, but
trade higher relative volumes. Thus small desks are able to effectively work large orders. Bottom line: apples-to-apples comparisons are tough, but neither shows a clear advantage.

critical points of potential slippage and focus their best resources where it matters the most.

Plexus News
At mid year, Plexus Group attained two significant milestones. We signed on our 200th client, and total retainer billings exceeded a $10 million level run rate for the first time.

Reprint any portion with credit given to:

grorrlt

1115A W. Otympic Blvd.,ll900 Los Angeles, CA900M PH : 31 0.31 2.5505 FAX: 31 0.31 2. 5506 wvw.plexusgroup.am
@

2002 Plexus GrouP, lnc.