Portfolio Strategy

North America

Market Commentary

AES Analysis

Stephen Casciano
+ 1 212 325 0776

Phil Mackintosh
+ 1 212 325 5263

Impact Model Stands Up to the Credit Crisis

Key points

07 December 2010

Our transaction cost model estimates the cost

of executing a trade using various market
metrics including liquidity, volatility, trade size
and aggression.
With the extreme volatility seen during the
credit crisis, we find that transaction costs
increased significantly but have now returned to
more normal levels.
We also show that the EDGE model captured
these changes effectively, maintaining accuracy
for US trade forecasts throughout the period.

A Turbulent Testing Period

Our EDGE Pretrade tool provides execution cost and risk
estimates for global equity trades (see A New EDGE in Impact
Cost). This model was calibrated using trades from our ExPRT
TCA system (see ExPRT for Dummies) from May, 2005 to Aug,
2008.
The calibration of this model showed that volume, volatility and bidask spreads were important predictors of a trades cost. In
hindsight, this was a relatively tranquil period in equity markets
(volatility mostly trading well below 20). Over the past 2 years
each of these factors moved significantly up.

Average spreads almost tripled, reaching the 20s for US

Equities.

Volatility increased almost eightfold into the 80s.

Volumes tripled from around 4bn to over 12bn shares/day.

Our latest ExPRT data shows that average transaction costs

increased in line with the spikes in these inputs. As these
factors return to more normal levels we investigate how the
EDGE model performed for US equities during these unusual
market conditions.

Exhibit 1: Markets were relatively tranquil during the calibration time period

Source: Credit Suisse Portfolio Strategy

Portfolio Strategy
Exhibit 2: Execution Cost: Data shows shortfall
changes as size and participation change. We built
this feature into our impact cost model, and analyse
the models recent results in the same way.

Did Our Edge Model Pass the Test?

35

Sa m e -Siz e T ra de s
(S ho rtf a ll a s ag re ssio n ch a n g es)

30
A ve ra ge
S h o rtf al l

25

(b p s )

20
10
5

20

As a result, we built this feature into our impact cost model,

allowing users to adjust aggression and see how our estimated
costs and risk change. We also analyse the models recent
performance in the same way below.

15

Credit Crisis Period: Super Stress Test

10

The credit crisis provides a rigorous stress test scenario for

the Edge model. In the following pages we:

Plot real trade costs versus what the model would have
predicted over time for Inline and VWAP strategies.

Across buckets of orders with consistent size (as a % of ADV)

and aggression (as a % of volume achieved by the order).

0
0 .1

Tra d e
Si ze

In our initial calibration of the EDGE cost model we found that

shortfall changed as size and participation changed in addition to
volatility, spreads and volumes (see our report Estimating
Execution Costs).

2.5

7. 5

(A DV %)

15
20
25
10
A gg re ssio n ( %)

30

40

Source: Credit Suisse Portfolio Strategy

Exhibit 3: Execution risk is caused by factors outside the

control of the trader such as market news and other
trading activity. As a consequence, there is uncertainty
around how much actual shortfall a trade will incur.
16%

Shortfall can even be

positive (in favor of
the trader) due to
external factors.

14%
12%
10%
8%

Model Holds Up Well

What we find is that the model performed remarkably well,
forecasting the actual changes in costs consistently across
almost all buckets of our trade matrix large to small and
passive to aggressive.
As a side note, we also see evidence of the different
personalities of these two algos. Specifically, we see
evidence to support the fact that Inline works best in rangebound markets, buying dips aggressively, while holding back
as stocks move away from the trader.
We have discussed the different personalities of more
opportunistic algos in these two recent reports: Trade
Smarter to Raise your Alpha Profile and In Search of the
"Perfect Algo"

6%
4%
2%
0%
300

200

100

100

200

300

Shortfall(bps)

Source: Credit Suisse Portfolio Strategy

Estimated Trade Costs in Edge

Run pre trade analysis on custom portfolios
to estimate transaction costs and analyze
portfolio characteristics.

Our website Edge (http://www.credit-suisse.com/edge)

contains many analytic tools for traders and PMs

See the following reports for more information:

EDGE Update: **NEW Portfolio Tools**
New Trade Planning Tools Online

Portfolio Strategy

VWAP Tests Positive

The 9 charts below are organised similarly to our EDGE
calibration results in Exhibit 2 above.

The top left bucket contains the smallest and most passive
orders.

The bottom right bucket contains the largest and most

aggressive orders.

From the results we see that the model:

Has fairly reliably predicted the initial increase and subsequent

decrease in execution costs.

Slightly underestimates cost for passive trades.

Slightly overestimates cost for aggressive trades, most notably

for smaller trade sizes

Seems better at capturing cost and variations in observed

impact with bigger, more aggressive trades. This is important
since these trades will contribute the most to P&L.

Exhibit 4: VWAP real vs modeled shortfall over the past 2 years, bucketed by size and participation.
Medium Participation (15-25%)

Low Participation (0-15%)

Data

Impact(bps)

Impact(bps)

Model

60

25
20
15
10

90

Data

70

Model

30

50

30
20

10
0

50

140

100

30
20

80

Impact(bps)

40

100

100

Impact(bps)

Medium Trade Size

(1-5% ADV)

50

Data

Model

Model

60

150

Data

120

Data

70

Impact(bps)

10
10

10

80

30

30

90

Model

50

40

Data

70

Impact(bps)

Small Trade Size

(0-1% ADV)

35

High Participation (25+%)

80

40

60
40

Model

50

20
50

10

20

300

250

150
100

Model

Model

200

200
Impact(bps)

Impact(bps)

Large Trade Size

(5+% ADV)

250

200

100

Data
Data

Data
Model

150

300

Impact(bps)

250

100

300

150
100
50

50

50

0
50

Source: Credit Suisse Portfolio Strategy

Portfolio Strategy
Exhibit 5: Inline is designed to buy dips. Comparing the
Trading Alpha component of ExPRTs shortfall breakdown
illustrates this characteristic (However this metric ignores
costs of completion)

Inline Trades Tougher to Predict

It seems from the results that the impact model persistently
underestimates costs for passive Inline orders and overestimates for
aggressive orders. However this conclusion is incorrect. The result
instead highlights a key difference between the naive VWAP strategy
and the more opportunistic and intelligent Inline strategy.

Evidence of Inlines Personality

The Credit Suisse Inline strategy thinks like a reversion trader. It
gets more aggressive when it sees favourable moves in the stock
price, trying to lock in gains before the opportunity goes away.
Consequently, what we actually see in the results below (and in
Exhibit 5) is that when Inline is aggressive, its because its getting
favourable fills. Note that this does not mean Inline should be used
with more aggressive settings. The aggression we see is a result,
not the cause, of favourable price moves.
Source: Credit Suisse Portfolio Strategy

We have discussed the different personalities of more

opportunistic algos in these two recent reports: Trade
Smarter to Raise your Alpha Profile and In Search of the
"Perfect Algo"

Exhibit 6: Inine real vs modeled shortfall over the past 2 years, bucketed by size and participation.
Low Participation (0-15%)

40
Impact(bps)

Impact(bps)

Impact(bps)

Small Trade Size

(0-1% ADV)

Model

20

10

15

30

10

20

10

70
Data

50

60

20
0

300

30

20

300
250

Model

200

50

100
50

50

50

Data
Model

150
Impact(bps)

150
Impact(bps)

Impact(bps)

Large Trade Size

(5+% ADV)

Model

Data

200

100

60

10

200

150

Model

40

250

Data
250

40

20

40

Data

80
Impact(bps)

80

100

Model

Model

100

120

Data

60

Impact(bps)

120

Impact(bps)

Medium Trade Size

(1-5% ADV)

140

Data

50

Model

Model

15

High Participation (25+%)

60

Data

25

Data
20

Medium Participation (15-25%)

30

25

100
50
0
50

100
150
200

Source: Credit Suisse Portfolio Strategy

Portfolio Strategy
Exhibit 7: RMSE for each single-input linear model
helps show which inputs are most helpful in prediction
of transaction costs.
RootMeanSquaredError

ADV

Spread

Volatility

All

Source: Credit Suisse Portfolio Strategy

Combined Inputs = More Accuracy

As we said in the first section, the Edge model uses volume, volatility
and bid-ask spreads as the predictors of a trades cost. However,
Exhibit 1 shows that all three factors appear to be correlated to each
other. In this section we verify that although the inputs are
correlated, its not redundant to include all three in the model.
We can use root mean squared error (RMSE) to measure, on
average, the discrepancy between a models predictions and the
observed data (smaller RMSE corresponds to higher accuracy). By
creating single-factor models for each input, this gives us a general
idea of which component is the most meaningful in the Edge
models transaction cost prediction. The bars in Exhibit 7 show the
ability of each input to predict cost, measured by RMSE of the
individual linear models. Its clear that volatility is the best predictor,
but additional accuracy is achieved by adding the remaining inputs.

Combining Model Components

The Edge model is the sum of a spread cost and an impact cost, as
we show in Exhibit 8. Consequently, in Edge we never combine all
three inputs at once. Instead we combine spread and liquidity in
spread cost and volatility and liquidity in impact cost.
Exhibit 8: Together spread cost and impact cost make up expected transaction cost in the Edge model

Exhibit 9: RMSE shows predictive power of each Edge model component across size and aggression
Low Participation (0-15%)

Medium Participation (15-25%)

High Participation (25+%)

Small Trade Size

(0-1% ADV)

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

SpreadCost
Term

ImpactCost
Term

Complete
Model

Medium Trade Size

(1-5% ADV)

SpreadCost
Term

Large Trade Size

(5+% ADV)

Source: Credit Suisse Portfolio Strategy

Exhibit 9 shows RMSE

for comparing each term.
We find that:
For small trades
spread cost is the
most meaningful
component
For moderate
trades, combining
both terms
improves the
estimate a lot
For large trades
impact cost is the
most important
This makes sense, as
smaller trades dont
penetrate the order book
deep enough to move
prices.

Portfolio Strategy

Credit Suisse
Portfolio Strategy

Market Commentary Disclaimer

USA
Phil Mackintosh
Gaurav Mundra
Victor Lin
Ana Avramovic
Glenn DeSouza
Stephen Casciano

+1 212 325 5263

+1 212 325 4096
+1 212 325 5281
+1 212 325 2438
+1 212 325 5664
+1 212 325 0776

phil.mackintosh@credit-suisse.com
gaurav.mundra@credit-suisse.com
victor.lin@credit-suisse.com
ana.avramovic@credit-suisse.com
glenn.desouza@credit-suisse.com
stephen.casciano@credit-suisse.com

+44 20 7888 9637

+44 20 7888 0908
+44 20 7888 0082
+44 20 7888 5832

colin.goldin@credit-suisse.com
mark.k.buchanan@credit-suisse.com
marwan.abboud@credit-suisse.com
ashish.kumar.3@credit-suisse.com

Europe
Colin Goldin
Mark Buchanan
Marwan Abboud
Ashish Kumar

Please follow the attached hyperlink to an important disclosure:

http://www.creditsuisse.com/legal_terms/market_commentary_disclaimer.shtml
Structured securities, derivatives and options are complex instruments that are
not suitable for every investor, may involve a high degree of risk, and may be
appropriate investments only for sophisticated investors who are capable of
understanding and assuming the risks involved. Supporting documentation for
any claims, comparisons, recommendations, statistics or other technical data
will be supplied upon request. Any trade information is preliminary and not
intended as an official transaction confirmation. Use the following links to read
the Options Clearing Corporation's disclosure document:
http://www.cboe.com/LearnCenter/pdf/characteristicsandrisks.pdf
Because of the importance of tax considerations to many option transactions,
the investor considering options should consult with his/her tax advisor as to
how taxes affect the outcome of contemplated options transactions.

This material has been prepared by individual traders or sales personnel of Credit
Suisse and its affiliates ('CS') and not by the CS research department. It is not
investment research or a research recommendation, as it does not constitute
Asia
substantive research or analysis. It is provided for informational purposes, is
intended for your use only and does not constitute an invitation or offer to
Mindy Chen
+852 2101 7133 mindy.chen@credit-suisse.com
subscribe for or purchase any of the products or services mentioned. The
information provided is not intended to provide a sufficient basis on which to make an investment decision. It is intended only to provide observations and views of individual traders
or sales personnel, which may be different from, or inconsistent with, the observations and views of CS research department analysts, other CS traders or sales personnel, or the
proprietary positions of CS. Observations and views expressed herein may be changed by the trader or sales personnel at any time without notice. Trade report information is
preliminary and subject to our formal written confirmation.
CS may, from time to time, participate or invest in transactions with issuers of securities that participate in the markets referred to herein, perform services for or solicit business
from such issuers, and/or have a position or effect transactions in the securities or derivatives thereof. The most recent CS research on any company mentioned is at
http://www.csfb.com/researchandanalytics.
Backtested, hypothetical or simulated performance results have inherent limitations. Simulated results are achieved by the retroactive application of a backtested model itself
designed with the benefit of hindsight. The backtesting of performance differs from the actual account performance because the investment strategy may be adjusted at any time,
for any reason and can continue to be changed until desired or better performance results are achieved. Alternative modeling techniques or assumptions might produce
significantly different results and prove to be more appropriate. Past hypothetical backtest results are neither an indicator nor a guarantee of future returns. Actual results will vary
from the analysis.
Past performance should not be taken as an indication or guarantee of future performance, and no representation or warranty, expressed or implied is made regarding future
performance. The information set forth above has been obtained from or based upon sources believed by the trader or sales personnel to be reliable, but each of the trader or
sales personnel and CS does not represent or warrant its accuracy or completeness and is not responsible for losses or damages arising out of errors, omissions or changes in
market factors. This material does not purport to contain all of the information that an interested party may desire and, in fact, provides only a limited view of a particular market.
Copyright 2009 Credit Suisse Securities (USA) LLC, and its subsidiaries and affiliates (together "Credit Suisse"). All rights reserved. AES, Advanced Execution Services, INLINE and TWAP are registered trademarks of Credit Suisse. Various features of the Advanced Execution Services system are patent pending.