03 Track Records : Luck or Judgement? Introducing Hit

03
RESEARCH PAPER
WINTER 2008
Track Records : Luck or Judgement?
Introducing Hit Rates & Win Loss
Ratios
This paper introduces some common sense measures of skill that go
a long way to answering the perennial question about whether track
records are ‘down to luck or judgement’.
Fund management, as with most things in life, is made up of a mixture of
the good and the not so good, and what’s important is to understand the
balance between them.
To this end, this paper discusses two measurements of Fund Manager
skill: the Hit Rate and the Win Loss Ratio. Firstly, it introduces the Hit
Rate which shows whether the majority of decisions add value. Rather
surprisingly we have found that Fund Managers typically only get around
50% of their decisions right and that even good Managers only have Hit
Rates of 51%.
The typical Manager however, compensates for a mediocre Hit Rate by
generating good gains from the winners which offset the losses from
the poor decisions. This is represented by the Win Loss Ratio where the
industry average is 100% and a good ratio is around 110%. That means
the alpha from good decisions is 10% more than that lost from poorer
ones.
These measures of Hit Rates and Win Loss Ratios help establish how
a Manager generates alpha and whether the track record can be relied
upon as a useful indicator of skill.
MOVING ON FROM RELYING ON TRACK RECORDS
Track records are a poor guide
the future
All experience shows that track records are poor guides to the future as they say
little about the skills that generated the results, or if they are likely to be repeated in
the future. To use a sporting analogy, track records keep the score, but say nothing
about how well the game was played or if they are likely to win next time.
In fact, coaches and sportsmen accepted this years ago and now analyse every
aspect of their game in minute detail to understand the result and prepare for the
next game.
Turning to the world of Investment Management, the same detailed approach
entails scrutinizing every decision in terms of the buys, sells, long and short
portfolio positions. Identifying skill simply comes down to whether:
•
buys and long positions go on to perform well
•
sells and short positions then perform badly.
Having separately identified each decision and classified them into two groups winners and losers - we can then establish the Hit Rates and Win Loss Ratios from
this mass of data points.
01 / 03
INALYTICS’ DATABASE
Mandate group
The database used for this Research Paper is made up of 215 traditional ‘long
only’ portfolios and excludes hedge funds.
The distribution of the sample is as follows:
Number of Portfolios in Mandate Group
TOTAL
North America
UK
Pacific
Smaller Country Specific
No. of Portfolios
Japan
Global
Europe
Emerging
0
25 50 75 100 125 150 175 200 225 250
The combined market value of these portfolios is US$152bn.
The database is made up of the larger conscious decisions taken by Managers,
both overweight and underweight, and excludes the very small companies in the
benchmarks that are not owned as these tend to be unintended decisions.
HIT RATES & WIN LOSS RATIOS ACROSS THE INALYTICS DATABASE
The hit rate is only half the story
Hit Rates are based on the adage that “if you get six out of ten decisions correct
you’ll do a good job”. Only now is this being measured. The Hit Rate is simply
defined as the number of correct decisions as a percentage of the total number
of decisions. However, the Hit Rate is only half the story when assessing the
Manager’s skill. The Win Loss Ratio compares the alpha generated from good
decisions to the alpha lost from poor decisions.
The average Hit Rate and Win Loss Ratio for the portfolios in the Inalytics database
are a 49.6% Hit Rate and a Win Loss ratio of 102%. The average Manager’s Hit
Rate, or ability to identify winners and losers, at 49.6%, is no better than 50:50.
However we have found that where skill exists, it tends to be derived from the
Fund Manager’s ability to find the real winners as demonstrated by the average
Win Loss Ratio of 102%.
The Hit Rates are distributed as follows:
Frequency
35
30
25
20
15
10
5
0
< 46
46.0-46.4
46.5-46.9
47.0-47.4
47.5-47.9
48.0-48.4
48.5-48.9
49.0-49.4
49.5-49.9
50.0-50.4
50.5-50.9
51.0-51.4
51.5-51.9
52.0-52.4
52.5-52.9
53.0-53.4
> 53.5
Hit Rates
The means and standard deviations of the Hit Rates establish broad parameters
for identifying that a good Hit Rate is 51.3% and a great one 53%.
02/ 03
Carrying out the same exercise for the Win Loss Ratios:
Frequency
35
30
25
20
15
Win Loss Ratio
10
5
89-90
91-92
93-94
95-96
97-98
99-100
101-102
103-104
105-106
107-108
109-110
111-112
113-114
115-116
117-118
0
These figures imply that, at the portfolio level, a good Win Loss Ratio is around 110
and a “great” one is over 120.
MARKED DIFFERENCE BETWEEN THE OVERWEIGHTS & UNDERWEIGHTS
The aforementioned analysis is for the large active decisions taken for the
portfolio as a whole, but as Inalytics’ previous Research Papers 01 and 02
have demonstrated, there is a world of difference between the levels of skill for
overweight and underweight decisions.
This paper continues this theme by analysing the Hit Rates and Win Loss
Ratios in terms of these Overweight and Underweight decisions.
Hit Rates
Win Loss Ratio
Overweight
48.5%
113.9%
Underweight
50.6%
92.2%
The table illustrates that alpha derived by the Overweight decisions were
attributable to the ability of the Manager to correctly select “big” winners (when
they got it right, they got it right) despite their less than 50% Hit Rate.
By comparison, the Underweight decisions returned a higher Hit Rate, but a
considerably lower Win Loss Ratio. In other words, this Manager chose more
correct stocks to underweight (Hit Rate greater than 50%) but the incorrect
decisions more than offset the good work done.
CONCLUSION
As far as we are aware this is the first time the Endowment Effect, as introduced
in the Behavioural Finance literature and Inalytics Research Paper 01, has been
explained in terms of the Win Loss Ratio.
We have also found that a good Hit Rate for overweights is approximately 52%
while a great one is around 55.5%, while the comparable numbers for the Win
Loss Ratio are 125% and 140% respectively.
RICK DI MASCIO
CEO
MALCOLM SMITH
Head of Research
t +44 (0) 20 3675 2900
e rick@inalytics.com
t +44 (0) 20 3675 2905
e msmith@inalytics.com
12th Floor, Emerald House
15 Lansdowne Road
Croydon, Surrey UK
CR0 2BX
03 / 03