picking winner funds - nomura holdings · 2016-06-15 · picking winner funds joshua livnat, phd,...

Picking Winner Funds Joshua Livnat, PhD, QMA/NYU

Gavin Smith, PhD, QMA

Martin Tarlie, PhD, QMA Nomura: 10th Annual Global Quantitative

Investment Strategies Conference

June 9, 2016

This material is provided for educational purposes only.

This material represents the views and opinions of the authors

and are not necessarily the views of QMA.

Confidential- Not for further distribution.

About QMA

1

• QMA is a quantitative investment firm that applies disciplined, research-driven

approach that seeks to identify and capture alpha opportunities and combines factor

exposures to create diversified, risk aware strategies designed for long-term,

consistent performance.

• Founded in 1975, QMA manages portfolios for a worldwide institutional client base,

including corporate and public pension plans, endowments and foundations, multi-

employer pension plans, and sub-advisory accounts for other financial services

companies.

• Wholly owned, independently operated subsidiary of Prudential Financial, Inc.

QMA is the primary business name of Quantitative Management Associates LLC.

How do Fundamental and Quantitative

Investment Processes Differ?

2

Fundamental Approach Quantitative Approach

Universe • Subset of universe

• Limited by analyst coverage

• Large universe

• Covers all investable universe

Information

Sources

• Information collected from management

meetings, industry experts, sell-side analysts,

company filings etc.

• Quantifiable market, financial and expectational

data

• Systematic collection of information

Stock Selection

• Research and identify individual securities that

are expected to outperform

• Requires analyst interpretation of data

• Subject to analyst behavioral biases and

emotion

• Rank all stocks in universe based on

expectations of future performance

• Use attributes associated with future

performance for ranking

• Re-evaluate stocks when information changes

Portfolio

Construction

• Focus is on upside to target price

• Near term catalysts are considered

• Focus is on portfolio tilted towards attributes

associated with future performance

• Stock attractiveness is impacted by risk and

transaction costs

Final Portfolio

and Expected

Performance

• Concentrated portfolio

• Large exposures to sectors and styles

• Performance subject to large variations

• Diversified portfolio

• Unintended exposures are limited

• Performance is more consistent

A Common Investment Goal

Generate portfolio returns in excess of the market

Comparison of Median Excess Returns

Fundamental Managers vs. Quantitative Managers

3

Source: QMA, eVestment.

Data derived from eVestment US Large Cap Core Universe, including inactive products, with S&P 500 as the benchmark. For Median Quant Managers, the primary investment approach of quantitative

was selected. For Median Non-Quant Managers, the primary investment approach excludes quantitative. eVestment is an outside vendor whose software has been used to create this exhibit. QMA

pays a fee for this software. QMA has made efforts to confirm accuracy/reliability of the data provided by eVestment but we disclaim responsibility for its accuracy or completeness. Please see ‘Notes

to Disclosure’ page for Important Information including risk factors and disclosures. Past performance is not a guarantee or a reliable indicator of future results.

ROLLING THREE-YEAR MEDIAN GROSS EXCESS RETURNS VS. S&P 500

October 1994 – March 2016

Excess R

etu

rn (

%)

-4

-3

-2

-1

0

1

2

3

4

5

6

7

Ap

r-96

Au

g-9

6

Dec-9

6

Ap

r-97

Au

g-9

7

Dec-9

7

Ap

r-98

Au

g-9

8

Dec-9

8

Ap

r-99

Au

g-9

9

Dec-9

9

Ap

r-00

Au

g-0

0

Dec-0

0

Ap

r-01

Au

g-0

1

Dec-0

1

Ap

r-02

Au

g-0

2

Dec-0

2

Ap

r-03

Au

g-0

3

Dec-0

3

Ap

r-04

Au

g-0

4

Dec-0

4

Ap

r-05

Au

g-0

5

Dec-0

5

Ap

r-06

Au

g-0

6

Dec-0

6

Ap

r-07

Au

g-0

7

Dec-0

7

Ap

r-08

Au

g-0

8

Dec-0

8

Ap

r-09

Au

g-0

9

Dec-0

9

Ap

r-10

Au

g-1

0

Dec-1

0

Ap

r-11

Au

g-1

1

Dec-1

1

Ap

r-12

Au

g-1

2

Dec-1

2

Ap

r-13

Au

g-1

3

Dec-1

3

Ap

r-14

Au

g-1

4

Dec-1

4

Ap

r-15

Au

g-1

5

Dec-1

5

Median Quant Manager Median Non-Quant Manager

• Quants have a lower target alpha but generate more consistent returns. Fundamental managers have a higher

return potential but experience higher return variability.

• Quant and fundamental managers are complimentary – both can outperform over time, however, excess returns

are not highly correlated.

Mean Annual Excess Return of Mutual Funds

4

As of 2014.

Source QMA, CRSP Survivorship Bias Free Mutual Fund Database.

This is shown for illustrative purposes only. Please see ‘Notes to Disclosure’ page for Important Information including risk factors and disclosures.

-82

-69

-56

-2

7

-90

-80

-70

-60

-50

-40

-30

-20

-10

0

10

20

Restricted Sample Momentum -Top Quartile FF Intercept t-Statistic TopQuartile

High Active Share - FFIntercept t-Statistic Top

Quartile

Diversified Momentum TopQuartile

Overview

Research Objectives:

• Superiority of diversified over concentrated funds.

• Compare diversified funds and funds with high active share.

• Can we find a better predictor of future fund performance?

Research Design:

• CRSP Survivorship-Bias Free database of mutual funds with

36 months of prior returns, active share and number of portfolio

positions 2002-2013.

• We examine the performance of these funds in the subsequent

year for various measures.

5

Acknowledgements

• Active share data obtained from Antti Petajisto’s website which

relates to his 2013 FAJ study.

• Recent active share data provided by Joseph Mezrich at Nomura

Securities.

6

Short Literature Survey

• Momentum and tracking error were shown to be predictive of

future fund returns.

• Active share and activeness of a fund.

• Regression of fund returns on FF factors to identify manager’s

skill.

• Sparse work on IR.

7

Future Performance of Funds

• We assume selection of funds takes place at the end of

December.

• We examine excess returns for the following year (t+1).

• Excess returns are cumulative fund returns minus cumulative

returns on the fund’s benchmark.

8

Data and Variables

• Data are from CRSP survivorship bias free database.

• Fund benchmark is from Petajisto’s active share data. If

unavailable, we use the benchmark from Nomura’s active share

data. Otherwise, we assume the S&P 500 Index.

• We also examine the results of regressing the fund returns on

the FF factors of market, value, size, momentum and short-term

reversals.

• All variables are weighted averages over all classes of shares,

where the weight is Total Net Assets (TNA).

9

Selection Criteria

• Funds are domestic equity funds with 36 months of prior returns.

• Eliminate index or sector funds.

• Eliminate funds with assets below $15 million.

• Monthly returns are from CRSP.

• Fund data (such as expense ratios, turnover ratios, number of

positions, etc.) are all from CRSP.

10

Distribution of Sample Funds

11

As of 2014.



0

500

1000

1500

2000

2500

196

61

96

71

96

81

96

91

97

01

97

11

97

21

97

31

97

41

97

51

97

61

97

71

97

81

97

91

98

01

98

11

98

21

98

31

98

41

98

51

98

61

98

71

98

81

98

91

99

01

99

11

99

21

99

31

99

41

99

51

99

61

99

71

99

81

99

92

00

02

00

12

00

22

00

32

00

42

00

52

00

62

00

72

00

82

00

92

01

02

01

12

01

22

01

3

Number of Funds

Variables Used to Select Funds

Momentum - The cumulative return on the fund in the past 36

months minus the cumulative return on the benchmark.

Tracking error – The standard deviation of the fund return minus

the benchmark return during the past 36 months.

Information Ratio – The average monthly excess return (over

benchmark) divided by tracking error.

12

• Active share from Petajisto, or from Nomura if unavailable.

• Intercept from FF regression.

• Intercept t-statistic from FF regression.

• R-square from FF regression. This is the only selection variable

where low values are preferred; most of the fund return cannot

be explained by the FF risk factors.

• Number of positions is from CRSP.

13

Variables Used to Select Funds

Number of Observations

14

0

10,000

20,000

30,000

40,000

50,000

60,000

All Funds Sample Funds Excluded Funds

As of 2014.



Fund Characteristics

15

-200

-150

-100

-50

0

50

100

150

Annual Return Expense Ratio Turnover Ratio

All Funds Sample Funds Excluded Funds

As of 2014.



Restricted Sample

Required active share data and number of positions from CRSP.

Data for 2002-2013.

As before, 36 months of prior returns.

16

-160

-140

-120

-100

-80

-60

-40

-20

0

All Funds Not in sample Sample Restricted Sample

As of 2013.



Mean Excess Return

Bps

Comparison of Best Quartile and Others

17

As of 2014.



Mean Annual Excess Return

-140

-120

-100

-80

-60

-40

-20

0

OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile OtherBest

Quartile

Momentum Tracking Error Information RatioFF Regression

InterceptFF Intercept t-

StatisticFF Regression

R-Square Active ShareNumber ofPositions

Top Performers

18

As of 2014.



Measure

Active

Share Difference Number of Positions Difference Total

Low High Significance Concentrated Diversified Significance

Momentum

% Beat 45 47 0.079 42 50 0.001

- -

N 476 1042 754 783 3031

Mean Excess

Return (BP) 24 45 0.083 88 7 0.0052 69 -

- - 10th % 590 900 980 580 800 - - -

Measure

Active

Share Difference Total

Low High Significance

FF Intercept

t - Statistic

N 511 969 3055

Mean Excess Return (BP) - 90 - 2 0.0067 - 56

10th % - 550 - 890 - 750

% Beat 35 47 0.0001

“Unconventionally Skilled” Managers

19

As of 2014.



Comparing Diversified Top Performers to “Unconventionally

Skilled”, High Active Share Managers Managers

20

As of 2014.



Top F-F Intercept

t-Statistic and Top

Active Share

Top Information Ratio

and Top Diversified

Difference

Significance

Panel A- All Funds

N 932 803

Mean 1 11 0.7574

% Beat 48 51 0.2060

Panel B- 5 Most

Uncorrelated Funds

N 12 12

Mean -8 36 0.7571

% Beat 58 42 0.4363

Interim Summary

Among “Unconventionally Skilled” managers, high Active Share is a

good predictor of future performance.

However, among top performers, Diversified funds outperform, and

have more consistent returns than Concentrated funds.

21

Can We Suggest a Better Predictor of

Fund Performance?

We saw that momentum is a good predictor of future returns.

We also show that consistency of returns is important for predicting

future returns.

The Information Ratio combines both.

However, it suffers from another issue.

22

Fund Choice

Which fund should you choose?

Fund A has an IR of 1. Fund B has an IR of 0.5.

23

Fund Choice

Which fund should you choose?

Fund A has an IR of 1. Fund B has an IR of 0.5.

Fund A has an alpha of 1BP and tracking error of 1BP.

Fund B has an alpha of 200BP and tracking error of 400BP.

You cannot pay your mortgage with IR.

24

Modified IR

The conventional IR is alpha (α) divided by tracking error (σ), i.e.

IR = α /σ

We modify IR by subtracting a desired alpha (α*) from the

numerator:

IR* = (α – α*) /σ

Note that to achieve a higher modified IR*, a fund is likely to have a

higher tracking error (σ).

Note that the desired alpha is a preference parameter by the

investor. Not to be confused with target alpha of the fund.

25

Modified IR – Pros and Cons

• Combines momentum and tracking error.

• Requires a minimum return.

• A passive index investment has an infinite negative IR if

there is a cost to the investor.

• Can be calculated from past returns and does not need holdings

data as in active share.

• However, it typically requires a long history of data to make

meaningful comparisons.

26

Modified IR

• We examined the performance of the modified IR measure,

where an investor specifies a desired alpha (above

benchmark).

• We simulated a desired annual Alpha of 0, 75, 150, 300, 450,

600, 750 and 900.

• We examined those funds that were both in the top quintile of

modified IR, and that exceeded their desired alpha in the prior

36 months.

27

Characteristics

28

As of 2014.



Number of Funds

Desired Annual Alpha

0

1,000

2,000

3,000

4,000

5,000

6,000

0 75 150 300 450 600 750 9000

50

100

150

200

250

300

350

400

0 75 150 300 450 600 750 900

Tracking Error (BP/year)

110

112

114

116

118

120

122

124

126

0 75 150 300 450 600 750 900

Expense Ratio (BP)

0%

20%

40%

60%

80%

100%

120%

0 75 150 300 450 600 750 900

Turnover Ratio

0

20

40

60

80

100

120

140

160

0 75 150 300 450 600 750 900

Number of Holdings


Desired Annual Alpha Desired Annual Alpha


But, Actual Subsequent Return

29

-66

-65

-64

-63

-62

-61

-60

-59

-58

-57

-56

0 75 150 300 450 600 750 900

Annual Excess Return (BP)

As of 2014.




Percentage of Funds Exceeding Desired

Alpha – Subsequent Year

30

0%

5%

10%

15%

20%

25%

30%

35%

40%

45%

50%

0 75 150 300 450 600 750 900

% Beating Target

As of 2014.




Summary

• Selecting a desired alpha has implications for the type of funds

that will be selected based on their past performance, and

more important for their future performance.

• With higher desired alpha levels, top funds tend to be more

concentrated, have higher tracking error, incur higher

expenses, and have higher turnover.

• However, actual future performance is not necessarily better

with higher desired alpha.

31

Examination of Top Funds Given a

Desired Alpha Level

• In the following analyses, we again require both top quintile of

modified IR and exceeding the desired alpha in the prior 36

months.

• We then classify the funds according to their concentration

(number of positions).

• We examine the future performance of diversified and

concentrated funds.

• Data for 2002-2013.

32

Number of “Best” Funds

33

0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2,000

0 75 150 300 450 600 750 900

Fewer than 100 Positions More than 100 Positions


As of 2014.



Actual Subsequent Annual Returns

34

-200

-150

-100

-50

0

50

100

0 75 150 300 450 600 750 9000

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2,000

0 75 150 300 450 600 750 900



As of 2014.



Simulation Uncorrelated 5 Funds

Mean Subsequent Annual Return

35

-250

-200

-150

-100

-50

0

50

100

150

200

0 75 150 300 450 600 750* 900*0

200

400

600

800

1,000

1,200

1,400

1,600

1,800

2,000

0 75 150 300 450 600 750 900



As of 2014.

*Based on fewer then 12 years.



Replication with eVestment Data

• Similar results were obtained using the eVestment data during

1996 - 2013.

• IR is a good predictor of future excess returns.

• Top quintile of Modified IR funds that exceeded their desired

alpha in the prior 36 months had similar characteristics to those

observed for the CRSP database.

• Quantitative “winner” funds typically did better than Fundamental

“winner” funds.

36

Conclusions

• Having specified a desired alpha, select funds with past

modified IR that exceeded the desired alpha.

• The best funds among top past performers are not necessarily

those with high active share or the most concentrated.

• Diversified funds (with many positions) typically have better

future performance than concentrated funds among the top

past performers.

37

Notes to Disclosure

Important Information

These materials represent the views of the author(s) regarding the economic conditions, asset classes or financial instruments referenced herein. Certain information

contained herein has been obtained from sources that QMA believes to be reliable as of the date presented; however, QMA cannot guarantee the accuracy of such

information, assure its completeness, or warrant such information will not be changed. QMA is the primary business name for Quantitative Management Associates

LLC, an SEC-registered investment adviser that is organized as a New Jersey limited liability company. The information contained herein is current as of the date of

issuance (or such earlier date as referenced herein). The underlying assumptions and our views are subject to change without notice. QMA has no obligation to

update any or all of such information; nor do we make any express or implied warranties or representations as to the completeness or accuracy or accept

responsibility for errors.

Distribution of this information to any person other than the person to whom it was originally delivered and to such person's advisers is unauthorized, and any

reproduction of these materials, in whole or in part, or the divulgence of any of the contents hereof, without prior consent of QMA is prohibited. These materials are

not intended as an offer or solicitation with respect to the purchase or sale of any security or other financial instrument or any investment management

services and should not be used as the basis for any investment decision. No investment strategy or risk management technique can guarantee returns

or eliminate risk in any market environment. No liability whatsoever is accepted for any loss (whether direct, indirect, or consequential) that may arise

from any use of the information contained in or derived from this presentation. QMA and its affiliates may make investment decisions that are

inconsistent with the recommendations or views expressed herein, including for proprietary accounts of QMA or its affiliates.

The opinions herein do not take into account individual client circumstances, objectives, or needs and are not intended as recommendations of particular securities,

financial instruments or strategies to particular clients or prospects. No determination has been made regarding the suitability of any securities, financial instruments

or strategies for particular clients or prospects. For any securities or financial instruments mentioned herein, the recipient(s) of this material must make its own

independent decisions.

QMA is a wholly-owned subsidiary of PGIM, Inc.. and an indirect, wholly-owned subsidiary of Prudential Financial, Inc.

Copyright 2016 QMA. All rights reserved.

38

QMA 20160602-447

picking winner funds - nomura holdings · 2016-06-15 · picking winner funds joshua livnat, phd,...

Documents