You are on page 1of 6

Measuring a Managers Trajectory

a (Very) Simple Approach


Nearly all portfolio analysis be it value-at-risk or performance attribution is carried out by looking at static
portfolios at given points in time. Yet when we limit ourselves to this approach, an important piece of information
is getting lost: how well a fund managers actual investment decisions are performing. By comparing the returns
of a dynamically changing portfolio to its theoretical, static returns, it is possible to uncover this key additional
information. This article explores this calculation, proposes a new metric, trajectory, to represent the impact
of a fund managers recent bets, and shows how this measure can be used to illuminate the managers overall performance.

Dan Blum
is executive director and product manager for JP Morgan Investment Information Services. He has previously
worked in product management at several large financial institutions, specializing in performance, attribution,
exposure and risk analysis, and has also been involved extensively in software design and database technology.
He holds a Bachelor of Arts degree from Brandeis University in Massachusetts.
INTRODUCTION

It is interesting to note that, with the abundance of highly


complex and sophisticated attribution models, few if any
consider one of the most basic types of analysis: looking
in aggregate at the outcomes of the actual decisions of
the investment manager. Performance algorithms effectively analyze the composition of portfolios but seldom isolate the impact of the changes to that
composition. What did the manager actually do in the
most recent quarter? Did the managers decisions help
or hurt?

It is perfectly possible, for example, that the manager


may have outperformed his benchmark in a given time
period in spite of his decisions rather than because of
them, simply because he started with an exceptionally
good portfolio. In other words, had he done absolutely
nothing, had he taken the quarter off, he would have outperformed the market by even more. Conversely, he
may have started the quarter with a particularly poor
portfolio, made excellent decisions, and still underperformed.

An underperforming manager whose recent decisions


have been favorable may be worth hanging onto. An
underperforming manager whose recent bets are also
underperforming would clearly be a candidate for defunding.
APPROACH

One might assume that it would be necessary to look at


individual trades to glean information about a managers
decisions. But it turns out that this is not necessarily
the case. For any given time period, the overall impact
of a managers decisions (with some caveats I will address below) can be ascertained simply as:
Portfolio Calculated Return - Portfolio Buy-Hold Return
Where:

Understanding the impact of manager decisions is key


to understanding his or her trajectory. Are things getting
better or worse? Are the decisions helping or hurting?

The Journal of Performance Measurement

-6-

Calculated return is the portfolio return for the period,


using whatever methodology IRR or TWR is preferred.

Buy-hold return (also known as static return) is the theoretical internal rate of return of the portfolio based on
its composition at the beginning of the time period, and

Summer 2015

held constant through the time period.

WHY THIS WORKS

Clearly, with a few exceptions, the difference between


the real portfolio and the theoretical portfolio, held constant at its beginning period composition, is due to the
trading activity of the manager. The difference between
the returns of these portfolios, therefore, can be considered to be the impact of the managers trades for a given
period.

For lack of a better term, I propose calling this measure

trajectory, the overall impact of the active decisions


for a given time period, which is separate and distinct
from whether or not the portfolio as a whole had a good
quarter.
To restate, for any given period:

Trajectory = Portfolio return


Portfolio buy-hold return

Since many performance applications can readily perform these calculations, the hope is this is not only intuitive and easy to grasp, but also easy to implement and

Table 1

Figure 1

Plotting of Managers A thru D by Excess Return and Trajectory

Summer 2015

-7-

The Journal of Performance Measurement

of some practical value.

as well as caveats around these applications.

APPLICATION

Now that we have the basic concept, we can use our new
measure, in conjunction with excess return, to divide
managers into convenient quadrants:
1.
2.
3.
4.

Underperforming but trajectory is positive


Overperforming and trajectory is positive
Underperforming and trajectory is negative
Overperforming but trajectory is negative

Table 1 shows how managers can easily be fit into these


quadrants based on three simple inputs: calculated return, buy-hold return, and benchmark return; and two
derived numbers: excess return and trajectory.

PLOTTING OF MANAGERS A THROUGH D BY


EXCESS RETURN AND TRAJECTORY

Note how a simple visualization of trajectory and excess


return along an x/y axis can be highly informative.
While Manager C has outperformed Manager A, he has
a negative trajectory and, therefore, may still be considered a weaker candidate for retention.
RELATED APPROACHES

The general principle here is not novel. Cantaluppi has


proposed a similar method to distinguish asset class level
trading impacts (2013) and impacts of individual holdings and transactions (2014).

Specifically, he shows how the impact of an individual


transaction can be determined by comparing the result
to what would have happened had the sell not taken
place; i.e., had the portfolio remained static. He further
shows a way of aggregating these impacts into a general
turnover return at the asset class level, which represents the sum of the impacts of trading activity for a
given period.

What this article does then is three-fold: first it offers a


relatively simple calculation for analyzing total level
performance. It then shows how the calculation may
be applied to produce a new analytic, trajectory, which
offers specific insight into the contribution of current period decisions versus alpha due to the legacy portfolio.
Lastly, it discusses broader applications of this measure
The Journal of Performance Measurement

Ryan (2001) and Cherkasov (2014/2015) both consider


the insights into a portfolio that may be obtained by
backing out the effect of passive returns on beginning
weights. Ryan uses his return-neutralized weight
analysis to segregate portfolio return into active and
passive components, and to segregate active decisions
across the portfolio. Cherkasov shows the noise created
by passive portfolio drift when calculating attribution
analysis and, using the same frozen weights, provides
a means of removing this noise, resulting in arguably
purer attribution results. These are somewhat similar
techniques to ultimately very different ends.
SOME CAVEATS AND ADDITIONAL
CONSIDERATIONS

I mentioned that there are caveats and additional considerations around this calculation. The following discussion raises and addresses these.
1. Corporate actions, in this calculation, may be incorrectly reflected as active decisions

Without looking at actual transactions or reverting to


some other method, mergers, spinoffs, etc., could be erroneously treated as part of the managers decisions
using our basic calculation, since the new security is
presumed to have arrived due to active management.
The impact of this, for time horizons of a quarter or less,
is likely to be de minimis versus the impact of actual
trades in an active portfolio. In an ideal solution, one
would mathematically back out the contributions of
these from the trajectory measure by adjusting the staticreturn to include the returns of receiving or spun-off securities. The pragmatist may alternatively decide that
these are not impactful enough to change the overall
story and choose to ignore them. However, it should be
noted that their impact is cumulative over time, so measuring a long period trajectory (one year or greater for
example) should require one of two options:
a. Perform the extra step of backing out the impact of
corporate actions described above.

b. Calculate all longer period trajectories by geometrically linking monthly or quarterly trajectories.

-8-

2. Inflows and outflows force investment manager de-

Summer 2015

cisions, so not all manager decisions are pure bets

This is true. However, the manager still has discretion


over what to buy or sell, and this discretion should still
cause one to count buys or sells based on inflow and/or
outflows to be incorporated into the managers trajectory.
3. Security transfers to or from other portfolios

Similar to corporate actions, these could be treated erroneously as part of a managers decisions if not properly accounted for. Ideally, one would calculate their
return impact separately and back it out of trajectory by
adjusting the static return to include the transferred securities based on date of transfer. Whether this is essential in practice depends on volume and frequency of such
events.
4. Relevance to different asset classes

While this calculation will work in the great majority of


portfolios, in short-term fixed income portfolios it is
problematic, as the frequency of maturing securities, and
the need to reinvest the proceeds of those maturities,
make it difficult to disentangle what is a buy/sell decision based on market analysis versus what is a maturity
and reinvestment of proceeds. Attempting this analysis
against alternative investment portfolios such as private
equity would also be challenging, as the infrequency of
valuation makes it difficult to determine the near-term
impact of current period transaction activity.
5. Hedging and risk adjustment

Not all manager decisions are driven by a desire to increase alpha, and some would argue that they, therefore,
should not be judged as though they were. While this is
true, the same argument would also work against looking at any excess returns in the absence of other metrics.
To the extent that we judge managers based on returns
versus benchmarks, it is equally valid to consider their
trajectories. Whether or not to judge a manager by additional measures such as portfolio VaRs, betas, etc., is a
legitimate, but entirely separate, consideration.

6. Time Horizon

As noted above, the longer the time scale of the trajecSummer 2015

tory period, the more likely it is that noise will be introduced into the numbers, owing to events such as corporate actions. To avoid this, trajectories over longer time
horizons should be calculated by linking shorter periods
(see note below regarding linking extended periods).

The question then becomes, what time horizon makes


the most sense for this measure? Managers may well
claim that measuring their bets in the first few months
after the bets are made is meaningless, as they are investing for the long term. Clearly, this could lead to an
extended, philosophical discussion. I would certainly
argue that the impacts of decisions even in the short term
cannot be dismissed. Losing 100 basis points is still losing 100% basis points whether it happens over five
years or one month. It is equally difficult to recover
from. And while the longer the time period may seem
more logical to the long-term-minded investor, the numbers will say less about the trajectory of the most recent
manager decisions.
7. Linking of returns to derive extended period trajectories

It is important to note that linking single-period trajectories geometrically will produce the same residuals as
the linking of attribution effects; i.e., the linked, extended period returns will not fully explain the difference between the extended period calculated return and
the extended period buy-hold return. Nonetheless, this
calculation is still extremely useful because, on a relative scale, extended period trajectories can still be compared with one another and ranked appropriately.
CONCLUSION

Past performance may not be a predictor of future performance, but we still must presume it is in some way
meaningful or we would not look at it. And given that
presumption, we might also presume that the aggregate
impact of recent manager decisions, distinct from the
actual returns, is particularly relevant and should be
given special weight. This is what trajectory is telling
us.

While traditional performance attribution is a useful


guide to the portfolio manager, it seldom provides actionable information for the asset owner. As it really
amounts to a decomposition of excess return, rather than

-9-

The Journal of Performance Measurement

a separate metric to consider along with excess return,


it seldom informs hiring and firing decisions. Trajectory,
on the other hand, may prove to be a useful tool, along
with extended period returns, excess returns, risk measures, betas, etc., for asset owners in assessing their managers.

In an environment where budgets are ever tighter, and


where asset owners are struggling to fund their liabilities
and meet their targets, we dont need more highly complex and often impenetrable ex-post analysis of portfolio
returns. We dont need esoteric algorithms that few systems can calculate, fewer individuals can understand,
and that provide little truly relevant information regarding the state of ones portfolio. What I am proposing is
an alternative to the move toward ever-increasing complexity - a straightforward way of assessing a manager
that is intended to be intuitive, actionable, and relatively
easy to implement.
REFERENCES

Cantaluppi, Laurent, Turnover Performance, The


Journal of Performance Measurement, Spring 2013.
Cantaluppi, Laurent, Contributions of Initial Holdings
and Transactions to Performance, The Journal of Performance Measurement, Summer 2014.

Ryan, Timothy, The Toolkit to Analyze a Pure Stockpicker, The Journal of Performance Measurement,
Spring 2012.

Cherkasov, Dmitry, New Look at Multi-Period Attribution: Solving Rebalancing Issue, The Journal of Performance Measurement, Winter 2014/2015.

The Journal of Performance Measurement

-10-

Summer 2015

Copyright of Journal of Performance Measurement is the property of Spaulding Group and its
content may not be copied or emailed to multiple sites or posted to a listserv without the
copyright holder's express written permission. However, users may print, download, or email
articles for individual use.

You might also like