An Introduction to Portfolio Rebalancing Strategies

Hossein Kazemi, Ph.D., CFA

Isenberg School of Management
The CAIA Association
November 2022
Version 6

1 Introduction
This note provides an introduction to various strategies for rebalancing portfolios
through time. The objective is to introduce the most common approaches and discuss
their implications for a portfolio’s risk-return properties. To keep the discussion
simple, this note follows the tradition of considering a simple portfolio consisting of
only two asset classes: a risky asset (equity) and a safe asset (cash or short-term
treasuries). Much of our discussions and conclusions apply to portfolios consisting of
more asset classes. Also, the framework provided here will allow the reader to expand
and test the results in a multi-asset framework.
While there are many ways that a portfolio – even a simple one – can be rebalanced
through time, in this note we focus on the following well-known approaches:
ˆ Buy-and-Hold (BH): This strategy could be used as a benchmark because it
involves no rebalancing. The initial allocations are left untouched.
ˆ Constant-Mix (CM): This strategy requires the investor to reset the asset weights
to a pre-specified level periodically.
ˆ Constant Proportion Portfolio Insurance (CPPI): This strategy calls for rebal-
ancing so that the portfolio’s value does not fall below a pre-specified floor.
ˆ Option-Based Portfolio Insurance (OBPI): This strategy uses options or repli-
cates options’ payoffs to protect the value of a portfolio. Purchasing options
is a more straightforward method while the replication approach can be used
when traded options are unavailable.
There are some variations of these strategies, and they will be briefly discussed.
For example, Time-Invariant Portfolio Protection (TIPP) involves a slight adjustment
to CPPI.
The rest of this note is organized as follows. Section 2 will introduce the notation
and provide a general conceptual framework for understanding portfolio rebalanc-
ing strategies. Section 3 will use this framework to present and discuss the above
strategies and their various extensions. In addition, Section 3 will provide extensive
numerical examples demonstrating how each strategy can be implemented. Section
4 shows how under some circumstances the models presented in Section 3 can be
implemented if the portfolio includes some illiquid assets. Section 5 provides some
concluding comments.

1
2 Portfolio Rebalancing
This section first discusses a general framework for understanding the rebalancing of
a portfolio. In addition, we discuss the factors that may drive that rebalancing.

2.1 Framework
This section provides a general framework for developing and understanding rebal-
ancing strategies. As mentioned in the previous section, we limit our discussion to a
portfolio consisting of a risky asset (equity) and a riskless asset (treasuries).
The portfolio’s initial value at time t is denoted by Vt . This initial value consists of
a combination of equity and treasuries. The per-unit prices of these two asset classes
at time t are denoted by St and Bt . Therefore, the initial value of the portfolio is
given by
V0 = N0 × S0 + M0 × B0
where Nt and Mt are the numbers of units of equity and treasuries held in the portfolio
at time t = 0. The value of the portfolio at time t will be

Vt = Nt × St + Mt × Bt

Note that the number of units of equity or treasuries held at the beginning of time t
must have been decided at the end of time t − 1. The investor enters period t with
shares purchased in the previous period. Thus, the value of the portfolio at time t
before and after rebalancing is given by:

 Nt−1 × St + Mt−1 × Bt Before Rebalancing
Vt =
Nt × St + Mt × Bt After Rebalancing


This means the investor will change the number of shares of equity from Nt−1 to Nt
while the number of units of treasuries changes from Mt−1 to Mt .
Example: Suppose the market value of a portfolio at the beginning of time 1
consists of 10 shares of equity and 20 units of treasuries. Each equity share is $15,
while each unit of treasuries is $5. The total value is

V1 = 10 × 15 + 20 × 5 = $250

The portfolio manager intends to acquire 1 additional share of the equity. How many
units of treasuries should be sold? After rebalancing, the value of the portfolio is still
$250. Therefore,

250 = (10 + 1) × 15 + (20 + x) × 5 → x = −3

The portfolio manager has to sell 3 units of treasuries. In this example, 1 × 15 +

(−3) × 5 = 0. The only way equity can be purchased is to sell some treasuries. End
of Example

2
 Using the results of the above example, we conclude that
(Nt − Nt−1 ) × St + (Mt − Mt−1 ) × Bt = 0
Any rebalancing must be self-financing: the only way to increase investment in one
asset class is to reduce holdings in another asset class. Portfolio rebalancing strategies
consist of a set of rules about how the number of shares being held, Nt and Mt , should
change, bearing in mind that the strategy has to be self-financing. One such set of
rules could be to keep Nt and Mt constant through time, which creates the Buy-and-
Hold strategy. Another rule could require (Nt × St ) /Vt and (Mt × Bt ) /Vt to remain
constant and sum to one, which creates the Constant-Mix strategy. As we will see
below, each rebalancing strategy is a statement about how Nt and Mt should change
over time.
Example: Continuing with the previous example, suppose the current portfolio
consists of 11 shares of equity and 17 units of treasuries. For the next period, equity
earns 10%, and treasuries earn 1%. How will the weights of the two assets change?
First, we calculate the current weights:
11 × 15
wt = = 66%
250
The weight of treasuries is 34%. After changes in share prices, the portfolio will be
valued at
11 × 15 × (1 + 0.10) + 17 × 5 × (1 + 0.01) = $267.35
Therefore, the new weights will be
11 × 15 × (1 + 0.10)
wt+1 = = 67.9%
267.35
and the weight of treasuries changes to 32.1%. Because equity earned more than
treasuries, we can see that its share in the portfolio increased. End of Example

2.2 Drivers of Rebalancing

We saw that because different asset classes earn different returns, their portfolio
weights will change, which may require the portfolio manager to rebalance the port-
folio. What else could prompt a portfolio manager to rebalance the portfolio? The
answer to the question critically depends on the portfolio manager’s mandate. For
example, the rebalancing rule for a Constant-Mix (CM) portfolio is well defined, and
it depends entirely on the realized returns of the investments. On the other hand, the
rebalancing rule for an actively managed portfolio may depend on realized returns
and the manager’s expectations of future returns in a very complex manner. The
rebalancing strategies discussed in this note will depend on variables whose values
are known at the time of rebalancing. For example, past returns or the investment
time horizon will be important drivers of the rebalancing strategies. Also, estimates
of unknown variables such as volatility could be used in some specific rebalancing
strategies. However, factors representing a portfolio manager’s expectations of future
events such as corporate earnings or interest rate changes do not enter our calcula-
tions. These are mostly related to active investment strategies.

3
3 Rebalancing Strategies
This section discusses the basic features of these rebalancing strategies:

ˆ Buy-and-Hold (BH)

ˆ Constant-Mix (CM)

ˆ Constant Proportion Portfolio Insurance (CPPI)

ˆ Option-Based Portfolio Insurance (OBPI)

Researchers have developed various extensions of some of these strategies. We will

explore some of the most important variations of these strategies. Several numerical
examples are presented to demonstrate the applications of the strategies.

3.1 Buy-and-Hold
BH is the most straightforward strategy and is typically used as the benchmark
strategy. It requires the portfolio manager to do nothing. Using the notation used in
the previous section, we can write the initial value of the portfolio as:

V0 = N0 × S0 + M0 × B0

The equity and treasuries holdings (M0 and N0 ) will remain unchanged through time.
The weight of the two assets are w0 = (N0 × S0 ) /V0 and (1 − w0 ) = (M0 × B0 ) /V0 .
Thus, the value of the portfolio at a future date will be

Vt = N0 × St + M0 × Bt

Let’s write the above equation in terms of relative changes in the assets’ values. We
can see that the relative sensitivity of the portfolio to changes in each asset class will
depend on its initial weight:
Vt St Bt
= w0 × + (1 − w0 )
V0 S0 B0
Example: The initial weight of equity in a BH portfolio is 60%. The value of
equity increases by 130% over time. What is the relative increase in the portfolio’s
value (assume the return to treasuries is 0)?

Vt S0 × (1 + 1.30) B0
= 0.60 × + 0.40 × = 1.78
V0 S0 B0
Thus, the return on the portfolio is 78%. Notice that 78% = 130% × 60%. End of
Example
The relationship between the portfolio’s value and the equity’s value is displayed
in Exhibit 1. In this example, the initial share of the risky asset is assumed to be

4
60% of the total assets. The lowest value that the portfolio can assume is reached if
the value of equity goes to zero. In this case, the value of the portfolio will be

Vt = M0 × Bt

Here, we have assumed that the initial weight of the safe asset is 40% of $100. The
value of the safe asset is kept the same as the value of equity is adjusted.

Exhibit 1: Portfolio value under the Buy-and-Hold strategy as a function of the

equity value

3.1.1 Risk-Return Properties

We can see from the above graph that the portfolio has a floor equal to the value
of the riskless asset. This implies that the investor has no tolerance for risk once
the floor is hit. What happens to the portfolio’s return volatility as the value of the
equity increases or decreases? In a BH strategy, the weight of the risky asset gradually
increases (decreases) as the value of the risky asset increases (decreases). Therefore,
the return volatility of the portfolio will increase or decrease as well. Exhibit 2
displays the relationship between the value of the equity and the return volatility of
the portfolio. We assume that the initial weight of the equity is 60% and that equity

5
has 20% annual volatility.

Exhibit 2: Relationship between portfolio’s volatility and changes in equity

value under BH strategy

We can see that as the value of equity increases, the return volatility of the portfolio
increases and eventually approaches the volatility of the equity (20% in this case).
Therefore, one can conclude that the BH strategy could be optimal for an investor
with a decreasing degree of risk aversion when the portfolio’s value increases. In
particular, the investor’s measure of relative risk aversion is likely to be decreasing
because as equity and wealth increases, the investor is willing to allow the relative
size of the risky asset to increase.

3.2 Constant-Mix
The Constant-Mix (CM) strategy represents one of the most common rebalancing
strategies. It requires the portfolio manager to rebalance the portfolio so that the
weight of each asset class is equal to some pre-specified figure. In general, this strategy
would require the manager to sell the asset that has increased the most and buy the
asset that has increased the least. In this sense, it is a contrarian strategy (opposite
of momentum).
Suppose the pre-specified weights of equity and treasuries are w and (1 − w) . The

6
initial positions in equity and treasuries will be:
w × V0
N0 =
S0
(1 − w) × V0
M0 =
B0
After one period, the portfolio’s initial value will change from V0 to V1 . To bring back
the weight of each asset class to the original level, the portfolio manager will have to
sell or buy equity depending on whether it has increased or decreased in value. In
particular, the new positions will be
w × V1
N1 =
S1
(1 − w) × V1
M1 =
B1
We can see that if equity has increased more than other asset classes, then N1 < N0 ,
indicating that some equity has to be sold.
Example: Suppose the initial value of a 60/40 portfolio is $100, with each equity
and treasuries unit selling for $10. How many shares of each asset class should be
purchased?
0.60 × 100
N0 = =6
10
0.40 × 100
M0 = =4
10
Suppose the value of each share of equity increases by 20%. How many units of equity
and treasuries must be held after rebalancing (return on treasuries is zero)? Each
equity share will sell for $12, and, therefore, the portfolio’s value will be

V1 = 6 × 12 + 4 × 10 = $112

As a result, the rebalancing requires the portfolio manager to hold the following units
of equity and treasuries:
0.6 × 112
N1 = = 5.6
12
0.4 × 112
M1 = = 4.48
10
The portfolio manager has to sell 0.4 units of equity and purchase 0.48 units of
treasuries. Note that 0.4 × 12 = 0.48 × 10. The proceeds from the sale of equity will
be just enough to buy 0.48 units of treasuries. End of Example
Exhibit 3 displays the relationship between the value of equity and the value of

7
the assets for both BH and CM strategies.

Exhibit 3: Portfolio’s values under the BH and the CM strategies as functions of

equity values

In the case of the CM strategy, the portfolio’s value is a concave function of the value
of the equity while that relationship is linear for the BH strategy. The reason is
that the CM strategy requires the portfolio manager to sell equity when it rises and
buy equity when it declines. Exhibit 4 shows what happens when the value equity
fluctuates through time.

Equity Bill Buy-and-Hold Constant-Mix

Equity Bill Assets Equity Bill Assets
100 100 60 40 100 60 40 100
110 100 66 40 106 64 42 106
120 100 72 40 112 67 45 112
130 100 78 40 118 70 47 117
120 100 72 40 112 67 45 112
110 100 66 40 106 64 43 106
100 100 60 40 100 60 40 101
Exhibit 4: Performances of BH and CM strategies in a non-trending market

Notice that for the CM portfolio the weight of equity remains at 60% through this
example. On the other hand, when the value of equity increases initially from $100

8
to $130, the weight of risky asset increases from 60% to 78/118 = 66.1% under
BH strategy. The weight remains at 60% under CM strategy (70/117 = 60%). In
the above example, the equity value first increases and then decreases, ending at
the initial value of $100. This is a non-trending market. The BH portfolio follows
the same pattern ending with the same initial value. However, the CM portfolio is
worth about $101 at the end, outperforming the BH strategy. The reason is that
the CM strategy requires the manager to sell (buy) equity when its value increases
(decreases), resulting in a higher portfolio value when equity returns to its original
price. Therefore, the CM strategy tends to outperform the BH strategy in a volatile
but non-trending market similar to our example.
It is helpful to note that as the equity market trends up initially from $100 to
$130, both portfolios increase in value but the BH increases by a larger amount ($118
vs. $117). On the other hand, when the equity market trends down from $130 to
$100, the BH portfolio declines by a larger amount. The net effect is that over the
entire period when the market shows no trend, the CM portfolio outperforms the BH
by a small amount ($101 vs. $100).

3.2.1 Risk Return properties

We can see from the above graph that unlike the BH strategy, the CM portfolio does
not have a positive floor. This implies that the investor has some risk tolerance even
when the portfolio’s value is very low. What happens to the portfolio’s return volatil-
ity as the value of the equity increases or decreases? In a CM strategy, the weight of
the risky asset stays the same whether equity increases or decreases. Therefore, the
return volatility of the portfolio will stay the same. Exhibit 5 displays the relationship
between the equity value and the portfolio’s return volatility. We assume that the

9
initial weight of the equity is 60% and that equity has 20% annual volatility.

Exhibit 5: Portfolio Return Volatility under BH and CM strategies

We can see that in the case of the CM strategy, as the value of equity increases,
the portfolio’s return volatility remains constant at 12%. Note that under the CM
strategy the portfolio’s return volatility will be w × vole , where vole is the volatility
of the equity. The CM strategy appears to be optimal for an investor with constant
measure of relative risk aversion since the relative size of the risky asset and the
portfolio’s volatility remain constant as the value of equity changes.

3.3 Constant Proportion Portfolio Insurance

We saw that the BH strategy creates a positive floor for the portfolio, while under the
CM strategy the floor is zero. Investors who desire or require a positive floor for their
portfolio can also use the BH strategy. However, this is not a viable strategy if the
floor is close to the current value of the portfolio. For example, suppose an investor
wishes to set a minimum floor equal to 90% of the current value of the assets. In that
case, the BH strategy indicates that 90% of assets should be invested in the riskless
asset. The expected long-term rate of return on this portfolio will be too low in almost
all cases. Then, the question is how one could create a positive floor for a portfolio’s
value while taking enough risk to earn a reasonable rate of return in the long run?
The Constant Proportion Portfolio Insurance (CPPI) provides a rebalancing strategy
that under most circumstances ensures that the value of the portfolio does not fall

10
below a pre-specified floor while taking more risk (at least initially) than the BH
strategy.
To implement the basic CPPI strategy, we need two additional pieces of informa-
tion: the investor’s investment horizon and the amount of risk that the investor is
willing to tolerate. Also, similar to the CM strategy, the portfolio manager needs to
decide on the rebalancing frequency. In the standard CPPI strategy, the floor is set
equal to the present value of the minimum amount that the investor will require to
have at a future time. If the investment horizon is denoted by T and the minimum
amount required at this by AT , then the floor at time t is given by

Ft = e−r(T −t) × AT

where r is the riskless rate and Ft is the floor at time t. If the value of assets at time t
is given by Vt , then it is required that the floor at time t to be smaller than the value
of assets at time t. That is, we must have Ft < Vt . The difference between the floor
and the portfolio value is the cushion, Ct .

Ct = Vt − Ft

The next piece of information needed to implement a CPPI strategy is the amount
of risk that investor is willing to take. This risk-taking variable, is called the multiplier
and is denoted by m. In almost all cases, the value of m is greater than one and less
than 10. Given the choice of m, the equity and treasuries positions of time t are given
by (unlike previous sections, St and Bt represent the total amount invested in equity
and treasuries):

St = mt × Ct = m × (Vt − Ft )
Bt = Vt − St

Notice that as long as m > 1, the amount invested in the safe asset, Bt , will be less
than the floor, Ft , indicating that there is a risk that the portfolio’s value could drop
below the floor. This will be discussed in more details shortly.
After one period, the portfolio value will change as equity and treasuries prices
change. In addition, the value of the floor will change as well because of the passage
of time. Given these new values, the new optimal equity position can be calculated
using the same approach used at time t:

Ct+1 = (Vt+1 − Ft+1 )

St+1 = m × Ct+1
Bt+1 = Vt+1 − St+1

If the value of the cushion has increased, the portfolio manager will need to increase
the portfolio’s equity position.
So, how does m relate to the investor’s risk-taking capacity? In the CPPI strategy,
the floor is not violated as long as the decline in equity between rebalancing periods
is not larger that 1/m. This relationship is approximately correct when there are also

11
changes in values of the riskless position and the floor. The following numerical ex-
ample shows how CPPI is implemented and how the choice of m affects the strategy’s
risk.
To set up the example, let’s assume that V0 = $100, T = 1, AT = $90, r = 2%,
and m = 3. Given these figures, the floor is F0 = 90 × e−0.02 = $88.22. Next, we
determine the cushion and the optimal positions in the equity and the riskless asset:
C0 = 100 − 88.22 = $11.78 Cushion
S0 = 3 × 11.78 = $35.34 Equity position
B0 = 100 − 35.34 = $64.66 Bond position
Let’s assume that the portfolio is rebalanced every month. Also, let’s assume that
equity increases by 2% during the month. The new values of the two positions will
be:
S1 = 35.34 × e0.02 = $36.05 Equity before rebalancing
0.02
B1 = 64.66 × e 12 = $64.77 Bond before rebalancing
V1 = 36.05 + 64.77 = $100.82 New portfolio value
11
F1 = e−0.02× 12 × 90 = $88.36 New floor
C1 = 100.82 − 88.36 = $12.46 New cushion
Next comes the rebalancing of the portfolio. Given the new value of the cushion, we
use the multiplier to determine the new equity and bond positions:
S1 = 3 × 12.46 = $37.38 Equity after rebalancing
B1 = 100.82 − 37.38 = $63.44 Bond after rebalancing
The following observations are important:

ˆ The initial equity position is larger than what is required by the BH strategy
for the same floor level. In this example, the BH strategy would have required
$88.22 in the riskless asset and $11.78 in equity.

ˆ The strategy increases (decreases) the equity position following an increase (de-
crease) in the equity value. That is, CPPI is similar to a momentum strategy.
ˆ The floor will be violated if the equity value declines by more than 1/m. This
statement is exactly correct when the floor is constant (e.g., interest rate is
zero). In this case, the maximum decline that can be handled is 33% = 1/3.
To see this, suppose the value equity declines by 33% during the first month
(instead of an increase of 2%). We assume zero interest rate, implying a floor
of 90. The above figures will be:
C0 = 100 − 90 = $10
S0 = 3 × 10 = $30
B0 = 100 − 30 = 70

12
 After the 1/3 drop in equity, the results will be:
2
S1 = 30 × = 20 Before rebalancing
3
B1 = 70 Before rebalancing
V1 = 10 + 70 = $90 New portfolio value
F = 90 = Floor
V1 =F

In this case, the portfolio’s value is just equal to the floor. We can see that a
larger decline would have resulted in a violation of the floor.

Example: Continuing with the initial example, suppose the equity value declines
by 2% from time 0 to time 1. What will be the new equity and cushion?

S1 = 35.34 × e−0.02 = $34.64 Before rebalancing

0.02
B1 = 64.66 × e 12 = $64.77 Before rebalancing
V1 = 34.64 + 64.77 = $99.41 New portfolio value
C1 = 99.41 − 88.36 = $11.05 New cushion

Given the new cushion, the rebalanced equity position will be S1 = 3 × 11.05 = 33.15.
End of Example
Exhibit 6 displays the relationship between the equity and the portfolio values.

Exhibit 6: The relationship between the value of assets and the value of
equity under BH and CPPI

13
It is important not to interpret this figure as demonstrating the superiority of CPPI
because the curve representing the CPPI portfolio lies above the BH portfolio for most
equity values. This Exhibit cannot display the dynamics of the portfolios through
time as equity value changes. This dynamic relationship will be shown below through
a numerical example.
The graph shows that the CPPI has a convex profile in relation to equity value
changes. The reason is that the strategy calls for an increase (decrease) in the equity
position following an increase (decrease) in the value of equities. In this regard, the
CPPI strategy is a momentum strategy. We saw that the CM strategy outperforms
the BH strategy in a volatile but non-trending market. As we will show below, the
opposite applies to the CPPI strategy. The CPPI will underperform the BH strategy
in a volatile and non-trending market. Exhibit 7 demonstrate this property through
a numerical example:

Equity Bill Buy-and-Hold CPPI

Equity Bill BH Assets Equity Bill CPPI Assets
100 100 60 40 100 80 20 100
110 100 66 40 106 88 20 108
120 100 72 40 112 105 12 117
130 100 78 40 118 123 3 126
120 100 72 40 112 122 -6 116
110 100 66 40 106 103 4 107
100 100 60 40 100 85 13 98
Exhibit 7: The performance of CPPI vs BH

In this example, we assume that the floor is 60, the CPPI multiplier is 2 and the
riskless rate is zero. We see that equity increases from 100 to 130 and then return
to 100. Thus, the terminal value of assets under BH is unchanged. The value of the
CPPI portfolio increases rapidly to 126, outperforming the BH portfolio. However,
the CPPI significantly underperforms the BH strategy as equity prices decline such
that the CPPI’s final value is lower than that of the BH portfolio. Also, note as
the value of equity increases, there is a point where a negative position in treasuries,
−$6, is required after rebalancing (i.e., leverage must be employed). In practice,
constraints may be imposed on the level of leverage used.

3.3.1 The Relationship between CPPI, BH and CM Strategies

Of the three strategies discussed so far, the CPPI is the more general approach and
one can easily show that the BH and the CM strategies are special cases of the CPPI.
First, let’s consider the relationship between the CPPI and the BH strategies. We
saw in the previous section that the key formulas for the CPPI are:

St = m × (Vt − Ft ) Equity Position

Bt = Vt − St Treasuries Position

14
If we set m = 1, then above strategy will become the BH strategy. To see this assume
m = 1, V0 = $100, S0 = $60 and F = B0 = $40. Also, to keep things simple, we
assume that the riskless rate is zero. As we can, these satisfy the above equations.
Suppose equity increases by 10%, which means the, V1 = 66 + 40 = $106. From here,
we have:

S1 = 1 × (106 − 40) = $66

B1 = 106 − 66 = $40

It is seen that there is no rebalancing, as the optimal equity position remains at $66
and the optimal treasuries position remains at $40. In this case, the CPPI will behave
exactly like a 60/40 BH strategy. Therefore, the CPPI with m = 1 behaves like the
BH strategy.
Second, to see how the CM strategy is a special case of the CPPI, let’s assume
that m = 0.6, V = $100, S0 = $60, B0 = $40 and F = 0. That is, the floor is set
equal to zero. Again, suppose equity increases by 10%, increasing the value of assets
to $106. From here, we have

S1 = 0.6 × (106 − 0) = $63.6 After rebalancing

B1 = 106 − 63.6 = $42.4 After rebalancing

The new equity position will be equal to 60% of total assets, and the optimal position
in treasuries will be 40% of total assets. Therefore, the CPPI with m < 1 and F = 0
behaves precisely like the CM strategy.
Whether the CPPI behaves like the BH strategy or the CM strategy critically
depends on the values of m and F . More specifically, we have these three cases:

ˆ m = 1 and F > 0: Similar to the BH strategy, the payoff will be a linear

function of the equity value. Also, it means that the final payoff from the
strategy is path-independent. That is, for a given value of the equity at time 0,
the portfolio’s value at time t will be the same regardless of the path taken by
equity values over time.
ˆ m > 1 and F > 0: This will be the classical CPPI strategy and the payoff
will be a convex function of the equity. The strategy will perform well in
trending markets and poorly in volatile but non-trending markets. This is a
path-dependent strategy. That is, for a given value of equity at time 0, the
portfolio’s value at time t will be different depending on the path taken by
equity values over time.
ˆ m < 1 and F = 0: Similar to the CM strategy, the payoff will be a concave
function of the equity. The strategy will work well in volatile but non-trending
markets. This is a path-dependent strategy.

Example: Suppose m = 2, V0 = $100, and F = 0. How do the optimal positions

evolve if equity increases by 10% and then declines by 10%? What type of strategy
is this?

15
 Given the above figures, the initial positions will be:

S0 = 2 × (100 − 0) = $200
B0 = 100 − 200 = −$100

Given the 10% increase in equity, the total assets will be V1 = 200 × (1 + 0.1) − 100 =
$120. Therefore, the optimal positions after rebalancing will be:

S1 = 2 × (120 − 0) = $240
B1 = 120 − 240 = −$120

With a 10% decline in the value of equity, the total asset changes to V2 = 240 ×
(1 − 0.1) − 120 = $96. The optimal positions will be

S2 = 2 × (96 − 0) = $192
B2 = 96 − 192 = −$96

The portfolio value is less than the initial value ($96 vs. $100). Also, the treasuries
position is negative indicating that leverage is employed. Finally, we saw that the 10%
move in the equity resulted in 2 × 10% move in the portfolio’s value. This strategy
is similar to that employed by leveraged exchange-traded funds (ETFs). We can also
see why levered ETFs do not perform well in volatile markets. They follow a CPPI
strategy with no floor. End of Example

3.3.2 Risk return properties

The CPPI strategy is a momentum strategy, purchasing (selling) the asset that has
increased (declined) during the rebalancing process. This leads the CPPI portfolio to
outperform the BH and CM portfolios in trending markets and underperform both
in volatile and non-trending markets. Exhibit 8 displays the volatility of the CPPI
portfolio for different levels of equity. In this exhibit, we assume that V0 = $100,
S0 = $60, F = $40, m = 2, and the volatility of equity is 20%. Unlike the BH
strategy, there is no upper limit to the volatility of the CPPI portfolio. Therefore,
the CPPI could be optimal for an investor who has increasing risk tolerance as the
value of the portfolio increases. On the other hand, the portfolio’s volatility declines
faster than the BH portfolio’s volatility as the equity value declines. In fact, once
the portfolio’s value hits the floor, the portfolio’s volatility goes to zero. Therefore,
the strategy appears to be optimal for an investor with a risk tolerance that declines

16
rather quickly as the portfolio value declines.

Exhibit 8: Volatility of CPPI and BH portfolio as a function of equity values

The CPPI has at least two additional risks not shared by the BH and the CM strate-
gies. We discussed one of them when it was shown that there is a chance that the
CPPI’s floor is violated if the value of the risky asset declines by more than 1/m.
This is called the Gap Risk. The Gap Risk is higher if markets are highly volatile and
when the multiplier is large. For example, a multiplier of 20 will cause the CPPI port-
folio to violate its floor if there is a decline of 5% or more in the risky asset between
rebalancing periods. The other risk faced by the CPPI is the Absorption Risk. This
will happen when the portfolio hits the floor and liquidates its risky assets. Under the
classical CPPI approach, there is no provision for returning to equity markets, and
therefore the portfolio will not benefit if equity markets rally. On the other hand, the
BH or the CM strategies will liquidate the equity position only if the value of equity
declines to zero – an improbable event. Therefore, both benefit from a rally in equity
markets after an initial drop.

3.3.3 Variations on the Classical CPPI

We saw in the previous sections that by changing two key parameters of the CPPI, the
multiplier and the floor, we can create a BH or a CM strategy. Over the years, several
variations of the classical CPPI strategy have been suggested. These variations are
too many to be covered in this note. We discuss three well-known cases here.
Changing the Multiplier: As discussed above, increasing the multiplier in-
creases the volatility of the portfolio and the Gap Risk. One variation of the CPPI

17
strategy calls for the multiplier to be adjusted as market volatility changes. For in-
stance, the multiplier could be expressed as a decreasing function of the VIX or an
estimate of volatility using the previous n trading days. This will reduce both the
Gap Risk and the Absorption Risk. However, it is unclear if the portfolio’s overall
risk-adjusted performance will always improve. Given the path-dependency of the
portfolio’s performance, the impact on its overall performance will depend on the
dynamics of the risky asset’s price. In addition to realized volatility or VIX, some
variations use other risk measures such as the Value at Risk or Expected Shortfall.
Resetting the Floor: In the classical CPPI, the floor is typically set to the
present value of the minimum value that the investor wishes to have at a future time.
An alternative is to set the floor equal to a percentage of the current value of assets,
x. That is, the optimal equity position will be

St = m × (Vt − x × Vt )

This is a CPPI with no floor and depending on the value of m, it could turn into a
CM or levered strategy. An alternative is to set the floor equal to a percentage of the
previous high-water mark. That is, let Ht = max (Vs : for s = 0, 1, ..., t) denote the
previous high-water mark of the portfolio’s value. Then, the floor could be set to a
percentage of Ht , with the optimal equity position expressed as:

St = m × (Vt − x × Ht )

In some sources, this is referred to as the Time Invariant Portfolio Protection (TIPP)
strategy. The main advantage of this strategy is that it does not require the investor
to specify a time horizon. However, the overall risk-adjusted performance of the
strategy may not improve. Again, this being a path-dependent strategy, its actual
performance will critically depend on the dynamic of the equity prices.
Stop-Loss: This is an extreme case of the CPPI strategy. Similar to the classical
CPPI, a floor is specified. Then the entire portfolio is allocated to the risky asset with
the provision that once the portfolio hits the floor, the entire portfolio is invested in
the riskless asset. This would be similar to a CPPI strategy with no floor appearing in
the formula, but with a multiplier equal to one as long as Vt > Ft and zero otherwise.
That is, 
m = 1 if Vt > Ft
St = m × Vt
m = 0 if Vt ≤ Ft
In some versions of the Stop-Loss strategy, there is a rule for allocating to the risky
asset again when the equity value exceeds Ft by a pre-specified amount. The main
advantage of this strategy is that it requires no rebalancing, and, therefore, it is most
suitable if transactions costs are high or the risky asset is not liquid enough.

3.4 Option-Based Portfolio Insurance

The Option-Based Portfolio Insurance (OBPI) strategy involves using various combi-
nations of the risky assets, treasuries and traded or synthetic options. As the name

18
implies, the strategy uses options (traded or synthetic) to protect the portfolio’s down-
side risk. Like the CPPI strategy, the investor must choose the value of the floor,
which represents the minimum portfolio value that the investor wants to have at a
future date. The investor will have several choices in implementing the protection
strategy.
While the OBPI can be implemented through various equity, options and cash
combinations, the strategies will have similar payoff profiles. Exhibit 9 displays the
payoff profile of a put-protected portfolio consisting of long positions in equity and a
put option with strike price (floor) of 90% of the equity’s initial value.

Exhibit 9: Portfolio values under OBPI and BH strategies when equity

values change.

Similar to the CPPI strategy, the OBPI strategy has a convex payoff profile. Again,
it is important not to interpret this graph as indicating that the OBPI strategy will
outperform the BH strategy regardless of the equity values. For example, this profile
does not reflect initial cost of purchasing the put. Similar to the CPPI strategy,
the OBPI portfolio will underperform the BH portfolio in volatile but non-trending
markets.

Long positions in put options and the risky asset Under this strategy, the
portfolio will consist of an investment in the risky asset and a long position in a put
option. The put’s strike price will be the floor. Since the put option has an up-front
cost, the portfolio manager must consider this cost in the design of the protection
strategy and the put option’s position size. This strategy will work if there are
liquid options on the portfolio’s risky asset. The strategy could be too costly if the
volatility risk premium imbedded in the put options’ implied volatility is large. In

19
general, there is a risk premium imbedded in the implied volatility as historically
implied volatilities have exceeded realized volatilities. However, this risk premium
changes through time and, as expected, tends to increase during periods of increased
uncertainty and distress in financial markets.
The terminal payoff from a long position in a European put option is given by
max (K − ST , 0) , where K is the put option’s strike price and ST is the underlying
asset’s price at maturity, T. Suppose the underlying asset is combined with a put.
The payoff from this portfolio at time T will be:
 
ST if ST > K
VT = ST + max (K − ST , 0) =
K if ST ≤ K
Therefore, the total portfolio will be at least worth K. Noting that it costs to purchase
the put, the net change in the value of the portfolio will be
 
ST − S0 − p0 if ST > K
VT − V0 = ST + max (K − ST , 0) − S0 − p0 =
K − S0 − p0 if ST ≤ K
where p0 is the initial cost of the put option. Given this initial cost, in order to ensure
that the the portfolio’s value at time T is at least as large as K, the portfolio manager
will need to purchase a put option with the strike price higher than K, increasing the
cost of the put option.
Example: Suppose the value of a portfolio is V0 = $100, 000. The share price
of the risky asset is $100. The investor wants to invest in the risky asset but does
not want the portfolio’s value to fall below $90,000 after one year. To protect the
portfolio’s value, the investor is considering the purchase of one-year put options. The
following put prices for various strike prices are observed:
Put Option Prices at Various Strike Prices
Strike Prices 100 99 98 97 96 95 94 93 92 91 90
Option Prices 6.94 6.46 6.00 5.55 5.13 4.73 4.35 3.99 3.65 3.33 3.02

Which one-year put option should the investor choose?

Suppose the investor considers the put option with strike price of $90. The entire
portfolio can be allocated in the following manner:
ˆ Purchase 970 units of the $90 strike price options . Total cost = 3.02 × 970 =
$2, 929
ˆ Purchase 970 shares of the risky asset. Total cost = 100 × 970 = $97, 000
ˆ Total cash left: 100000 − 2, 929 − 97, 000 = $71
Let’s find out what the portfolio’s value will be if the stock price is less than $90
in one year. Suppose the stock closes at 85:
VT = 970 × 85 + 970 × max (0, 90 − 85) + 71 = $87, 371
We can see that the portfolio’s value is below the floor. It is clear that a put option
with a higher strike price has to be purchased, which will increase the cost of the put
option. A little trial and error will show that the best strategy appears to be this:

20
 ˆ Purchase 958 units of the 94 strike price put options. Total cost = 4.35 × 958 =
$4, 167
ˆ Purchase 958 shares of the risky asset. Total cost = 958 × 100 = $95, 800

ˆ Total cash left: 100, 000 − 4, 167 − 95, 800 = $33

The portfolio’s value if the stock price closes at 85 will be:

VT = 958 × 85 + 958 × max (0, 94 − 85) + 33 = $90, 085

This strategy appears to work. Note that the higher cost of the option will reduce
the amount that can be invested in equity, reducing the potential benefits of future
increases in its price. Roughly speaking, we should start with the put option whose
strike price matches our floor, 90 in this case. Then add the option’s premium to the
strike price and examine the results of purchasing the put with the strike that is right
above the figure, 90 + 3.02 = $93.03. In this case, the $94 strike is likely to be the
best choice. End of Example
Example: Suppose the V0 = $100, 000 and the investor wishes to have at least
$90,000 after one year while investing in a risky asset. What should be the strike
price of the 1-year put option?
Suppose each share of the underlying asset is worth $100 and the 1-year put option
price is p0 (K), indicating that it depends on the strike price. To protect the portfolio,
the investor should purchase one put option for each share of the underlying asset.
Therefore, each share will cost ($100 + p0 (K)), indicating that the manager can only
buy 100, 000/ (100 + p0 (K)) units of the risky asset and the put option. As we saw
before, at expiration the value of a portfolio consisting of N shares of the risky asset
along with long positions in N put options with strike price of K will be:
 
N × ST if ST > K
VT = N × ST + N × max (K − ST , 0) =
N × K if ST ≤ K

In the present example N is given by 100, 000/ (100 + p0 (K)) . Therefore, when ST <
K, the portfolio’s value will be:
 
100, 000
VT = ×K
100 + p0 (K)
Note that under this scenario, we want the value of the portfolio to be at least $90,000.
Therefore,
 
100, 000
90000 = ×K
100 + p0 (K)
K = (90% × 100) + (90% × p0 (K))

The strike price that we select must satisfy the above expression. Using the figures
from the previous example, we can see that strike price of $94 satisfies this expression.

94 ≈ (90% × 100) + (90% × 4.35)

21
On other hand, we can see that the strike price of $92 will not satisfy the above
expression
92 ̸= (90% × 100) + (90% × 3.65)
End of Example

Long positions in call options and the riskless asset Given the put-call parity,
a put-protected portfolio can be created using a combination of call options and cash.
For European options, the following put-call parity relationship should hold if there
are no arbitrage opportunities:

c (S, K, T ) + K × e−rT = p (S, K, T ) + S

where c is the call price and p is the put price. We can see that a combination of a
call plus some cash (the left-hand side of the equation) is the same as a combination
of a put and one share of the underlying asset.
Example: Suppose the premium for a one-year call with the strike price of $100
is $6. What should be the put price if the riskless rate is 2% per year?
Given the put-call parity, the put price will be:

p (100, 100, 1) = 6 + 100 × e−0.02 − 100 = $4.02

Note that the price of the at-the-money put is smaller than that of the at-the-money
call. The two would be equal if the riskless rate were zero. End of Example
Let’s look more closely at the mechanics of using call options to meet a minimum
floor at maturity. As before, the total value of our assets at time 0 is denoted by V0 .
The minimum value that we wish to protect at maturity is K, and the riskless rate
is r. Given our positions in the call option and the riskless asset, the initial value of
the portfolio can be expressed as:

V0 = N × c (S0 , K, 0) + B0
= N × c (S0 , K, 0) + N × K × e−rT + D

where N is the number of call options we have purchased, and D = B0 −N ×K ×e−rT

is any “extra” cash left after purchasing the options and setting aside the present value
of the protected amount. Of course, whatever is left after the purchase of the option is
invested in the riskless asset. The above presentation will be helpful when we examine
the portfolio’s value at maturity. At maturity, the portfolio will consist of the future
values of D and S0 plus the payoff from the put option.

VT = N × K + N × max (ST − K, 0) + D × erT


 D × erT + N0 × ST If ST > K
=
D × erT + N0 × K If ST ≤ K


22
Example: Suppose a portfolio consists of 100 shares of a risky asset with a price per
share of $10. The minimum portfolio value that is acceptable to us after one year is
$800. The one-year call option with the strike price of $8 sells for $2, and the riskless
rate is 2% per year. What is the optimal insurance policy?
First, we will sell all of our shares to create a protected portfolio using call options
and cash. Next, we buy 100 call options with a strike price of $8. The initial value
of the portfolio can be expressed as

V0 = $1000
= 100 × 2 + 100 × 8 × e−0.02 + 15.84

We spend $200 on options, invest $800 in treasury bills, and have $15.84 extra cash
left. Suppose after one year the price of the risky asset is $7.5. What will be the
value of the portfolio? The call option will expire out of the money.

VT = 100 × 8 + 15.84 × e0.02 = $816.16

We can see that because of the “extra” cash, the portfolio’s value exceeds $800.
Alternatively, the investor could have invested the “extra” cash of $15.84 in the
stock. End of Example
Example: Suppose in the previous example, the price of each call option is $2.5.
What will be the value of the portfolio at maturity?

V0 = $1000
= 100 × 2.5 + 100 × 8 × e−0.02 − 34.2

In this case, the “extra” cash left after setting aside the present value of the protected
amount is −$34.2. The value of this portfolio after one year when ST = 7.5 will be:

VT = 100 × 8 − 34.2 × e0.02 = $765

Given the high price of the call option, we cannot guarantee that our portfolio will be
worth at least $800. Basically, the cost of insurance is too high in this market. End
of Example
As we have seen from the previous examples, one may not be able to set the
minimum value of the portfolio exactly equal to the desired floor when options are
employed. As long as, several options with strike prices close to our floor are traded,
we should be able to come close to meeting the desired floor.

Option replication In this case, the portfolio manager can purchase protection
by replicating the options’ payoffs through dynamic trading. This approach will be
somewhat similar to the CPPI except that the multiplier will change in response to
changes in the price of the risky asset, the volatility of the risky asset, and time
decay. If the underlying asset can be traded in liquid markets, this approach could be
preferred, especially when substantial volatility premium is embedded in the options’
implied volatilities.

23
 For the remainder of this section, we will focus on the put replication strategy as
it is relatively easy to implement if the investor already has a long position in the
underlying asset.
We need to use an option pricing model to replicate the option’s payoff. For exam-
ple, the celebrated Black-Scholes-Merton model shows that a European put option on
a non-dividend-paying stock can be replicated using a combination of a long position
in the riskless asset and a short position in the stock. That is,

p (St , K, T − t) = Mt − ∆t × St

Where Mt is the amount invested in the safe asset, and ∆t is the number of shares
of the risky asset that are shorted. The delta of the put option at time t is given
by −∆t . Both Mt and ∆t are time-varying and will depend on the underlying asset’s
volatility, among other things.
It is well-known that the Black-Scholes-Merton option pricing model is the limiting
version of the Binomial option pricing model. For this reason, we use the Binomial
model to show how option-like protection can be created using a combination of
positions in the risky asset and the riskless asset. Exhibit 10 shows a two-period
model of the risky asset, the value of the unprotected portfolio and the portfolio’s
value with the desired protection. The other assumptions are that we currently own
100 shares of stock, the riskless rate is zero, and we want our portfolio’s value not to
fall below 90% of its current value.

Stock Price Portfolio of 100 Shares Desired Value of the Portfolio

125 12,500 12,500
115 11,500 11,500
100 100 10,000 10,000 10,000 10,000
85 8,500 9,000
75 7,500 9,000
Exhibit 10: Values of the risky asset, the unprotected portfolio and the protected portfolio.

First, consider the upper part of the tree when the portfolio’s value is $11,500.
We need to form a portfolio so that its value will be $12,500 or $10,000 depending on
the value of the stock. It is clear that we can replicate those two values by holding
100 shares of the risk asset. That is, no position in the riskless asset is needed if the
stock price is $115 in period 1. Next, consider the portfolio’s composition when the
stock price is $85. We need to create a portfolio that will be worth $10,000 or $9,000
depending on the value of the risky asset being $100 or $75.

N12 × 100 + M12 = $10, 000

N12 × 75 + M12 = $9, 000

24
These two equations state that we need to hold N12 units of the stock and M12
units of the riskless asset such that the value of the portfolio will be $10,000 or
$9,000. The solutions are N12 = 40 and M12 = $6, 000. This portfolio will cost
40 × 85 + 6000 = $9400. Note that rather than holding 100 shares of the stock we
are holding only 40 shares. In other words, we have shorted 60 shares of the stock in
this case. Finally, we move to the first period. In this case we form a portfolio such
these two conditions are satisfied depending on what happens to equity values:

N0 × 115 + M0 = $11500
N0 × 85 + M0 = $9400

Note that when the stock price declines to 85, we need the portfolio’s value to be
$94,00, the amount needed to support our strategy in the second period. The solution
will be N0 = 70 and M0 = $3450. In this case, we have effectively shorted 30 shares.
The total cost of this portfolio will be 70 × 100 + 3450 = $10450. This exceeds our
initial capital by 450. That is, we need another $450 to protect our portfolio. The
results of the strategy are displayed in Exhibit 11

V3 = 12, 500
N11 = 100; M11 = 0; V11 = 11, 500
N0 = 70; M0 = 3, 450; V0 = 10, 450 V2 = 10, 000
N12 = 40; M12 = 6, 000; V12 = 9, 400
V1 = 9, 000
Exhibit 11: The values of the protected portfolio

If we were to calculate the price of a put option using the above figures, the price
would be $450, which is the extra amount needed to protect the portfolio through
dynamic trading.
Dynamic trading has some advantages and disadvantages compared to the pur-
chase of calls or puts. The advantages are:

ˆ It can be implemented when marketable options are not available.

ˆ It avoids counterparty risk if the investor has to use over-the-counter options.

ˆ It avoids paying the risk premium embedded in implied volatilities.

The disadvantages of dynamic rebalancing are:

ˆ It could be costly because frequent rebalancing increases transaction costs.

ˆ It is exposed to Gap Risk, which is avoided when options are used.

ˆ It is exposed to estimation risk because the underlying asset’s volatility needs

to be estimated.

25
3.4.1 Risk-Return Properties
The OBPI risk-return profile shares some features with the CPPI’s profile but also
differs from the CPPI in some important ways, especially when traded options are
used to purchase protection. We can see from Exhibit 11 that the OBPI portfolio’s
return volatility increases and approaches the volatility of the equity (20%) as equity
value increases. Also, as the equity value declines, the volatility declines rapidly,
reaching zero as the equity value drops below the floor (90%). However, compared
to the volatility of the CPPI portfolio, the volatility of the OBPI does not increase
above 20%. Therefore, the OBPI strategy appears to be optimal for an investor with
very little tolerance for risk as equity value approaches the floor but also does not
have unbounded tolerance toward risk as equity value increases.
The OBPI also shares some characteristics with the simple BH strategy. More
specifically, both have the same upper (20%) and lower (0) bounds for volatility.
However, as we can see, the volatility of the OBPI approaches zero much faster when
equity value declines.

Exhibit 12: Volatility of the OBPI, CPPI and BH Portfolios as Equity Value
Changes

As mentioned previously, one risk associated with the CPPI is that if the portfolio’s
value reaches the floor, the equity position is set to zero. The portfolio will no longer
participate in any subsequent recovery in equity values (the Absorption Risk). This
risk is avoided in the OBPI strategy. The reason is that unlike the CPPI approach,
the multiplier is not constant under the OBPI strategy. This leads to another poten-
tial advantage for the OBPI strategy: unlike the CPPI strategy, the OBPI is not a
momentum strategy, and, therefore, it will outperform the CPPI portfolio in a volatile
but non-trending market. Of course, the flip side is that the CPPI will outperform

26
in a rapidly rising equity market, but it will come with higher volatility (see Exhibit
12). Note that both the CPPI and the OBPI will underperform the BH strategy in
a volatile but non-trending market.

4 Dynamic Strategies with Illiquid Assets

The discussions so far have focused on portfolios consisting of liquid assets. It may
appear that none of the models presented here will have any use when the portfolio
consists of illiquid assets such as private equity, hedge funds, and real estate. This
section discusses circumstances under which most of these models can be applied to
illiquid assets. People familiar with hedging strategies using futures contracts will
find the following discussion very familiar. The sole requirement for applying these
models to illiquid assets is having access to a liquid asset or portfolio that is highly
correlated with the illiquid asset.
For example, we know that private equity and venture portfolios can be replicated
using a portfolio of publicly traded securities. The replication is far from perfect, and
the tracking error could be significant for individual funds. Large portfolios of these
assets, on the other hand, can be reasonably tracked using portfolios of publicly traded
securities. The same is true for hedge funds and real estate investments. Of course,
these replication strategies are unlikely to match the alphas of these investments,
especially for the top-tier managers.
Suppose a liquid asset or portfolio reasonably correlated with the illiquid asset is
available. We can then proceed with our discussion of the process of implementing
previous models in the presence of an illiquid asset. Similar to previous sections, we
assume that the portfolio consists of only two assets, cash and a risky asset, which is
illiquid in this case. The total value of the portfolio at time t is:

Vt = Bt + St

where Bt is the amount invested in the risk-free asset and St is the value of the illiquid
risky asset. The uncertain rate of return on this portfolio can be expressed as

Rt+1 = (1 − wt ) rf + wt rS,t+1

where wt = St /Vt , the weight of the risky asset in the portfolio. The rate of return on
the illiquid asset is related to the rate of return on a liquid futures contract according
to the following expression:

rS,t = rf + α + βrq,t + εt

Here, α is the alpha of the illiquid assets, β is the beta of the illiquid asset with
respect to the futures price, rq,t is the excess return on the futures contract, and εt is
the tracking error.
Regardless of the rebalancing strategy, the key is to adjust the weights so that they
correspond to the optimal weights given by the strategy. While the current weight of
the risky asset is given by wt , the optimal weight under the desired strategy is given

27
by kt . The exact value of kt will depend on the strategy being followed. For example,
under CM strategy, kt will be a constant, while under CPPI, kt will depend on the
parameters of the CPPI and will change through time. Regardless of the desired
strategy, the portfolio manager’s target return for this period is
T
Rt+1 = (1 − kt ) rf + kt rS,t+1

So, the question is, what is the correct position in a liquid replicating portfolio that
T
would make Rt+1 on average, approximately equal Rt+1 ? We will not be able to match
them perfectly because of the presence of the tracking error. If the portfolio manager
establishes a position in the futures contract such that its notional value relative to
the total value of the portfolio is Ft , then the rate of return on the portfolio will be

Rt+1 = (1 − wt ) rf + wt rS,t+1 + Ft rq,t+1

T
The optimal position in the futures contract is determined by setting Rt+1 = Rt+1 .A
little algebra, and some approximation shows that the optimal position is
 
α
Ft ≈ + β (kt − wt )
rq,t
The approximation is obtained by assuming that the tracking error is small. So, when
it is multiplied by the difference between the actual and optimal weights of the risky
asset, the result will be negligible. Let’s see how well our approximation formula
works.
Example: Suppose the optimal strategy that the portfolio manager wants to
follow is the CM strategy with the illiquid asset’s position equal to 60%. The initial
weights are indeed 60% and 40%. The risk-free rate is assumed to be 2%, the alpha
of the illiquid asset is 3% per year, its beta is 0.9 with respect to the futures contract,
and the tracking error is ±0.5% per year. Exhibit 13 displays the annual rate of
return on the illiquid asset and the futures contract.

Return on
Return on Tracking Illquid
Period Futures Error Asset
0
1 20% -0.50% 22.5%
2 20% -0.50% 22.5%
3 -20% 0.50% -12.5%
4 -20% 0.50% -12.5%
5 -25% -0.50% -18.0%
6 20% 0.50% 23.5%
Exhibit 13: Hypothetical Returns on
Futures Contract and the Illiquid
Assets

28
Given these figures, Exhibit 14 displays the portfolio’s evolution with the illiquid
asset, its annual total rate of return, and its total rate of return if the portfolio
manager could rebalance the portfolio.

Total
Illiquid Total Weight of Total Optimal
Period Bond Asset Value Illquid return Return
0 40.00 60.00 100.00 60%
1 40.80 73.50 114.30 64% 14.3% 14.3%
2 41.62 90.04 131.65 68% 15.2% 14.3%
3 42.45 78.78 121.23 65% -7.9% -6.7%
4 43.30 68.93 112.23 61% -7.4% -6.7%
5 44.16 56.53 100.69 56% -10.3% -10.0%
6 45.05 69.81 114.86 61% 14.1% 14.9%
Exhibit 14: Evolution of the illiquid portolio and the total return on
optimal strategy.

We can see that in period one the actual total return on the portfolio is equal to its
optimal total return as the initial weight of the risky asset is 60%. However, after the
first period, the optimal total rate of return diverges from the actual rate of return
as the weight of the risky asset moves away from the optimal weight of 60%. Exhibit
15 shows the optimal position in the futures contract and the total return on the
portfolio with the futures positions included.

Total
Optimal Total Return
Futures Optimal with
Period Position Return Futures
0
1 -4.5% 14.3%
2 -8.8% 14.3% 14.3%
3 -3.7% -6.7% -6.2%
4 -1.1% -6.7% -6.7%
5 3.0% -10.0% -10.0%
6 -0.8% 14.9% 14.7%
Exhibit 15: Optimal positions in futures
contracts and the resulting impact on
portfolio’s return.

The figures under Optimal Futures Position are calculated using the formula discussed
above. We can see that the total rate of return on the portfolio that includes the

29
futures contract is very close to the optimal rate of return that would have been
obtained under the CM strategy. The match is not perfect because of the tracking
error. End of Example
Example: A portfolio currently consists of $100 in bonds and $100 in illiquid
private equity before rebalancing. The portfolio manager follows a CPPI strategy
where the floor is $180 and the multiplier is 7.
ˆ Assuming that the private equity could be traded, what should be the portfolio’s
composition after rebalancing?
First, we calculate the cushion:
C =V −F
= 200 − 180 = 20
Next, using the multiplier, we calculate the optimal position in both asset classes
S =C ×M
= 20 × 7 = 140
B = 200 − 140 = 60
Therefore, if the risky asset were liquid, we needed to purchase $40 of the
risky asset while selling $40 of the risk-less asset. The optimal weights post
rebalancing will be k = 70% and (1 − k) = 30%.
ˆ While the private equity position is illiquid, its historical return has been highly
correlated with equity markets. Historically, it has had an annual alpha of 4%
with a beta of 1.2. The expected return on equity markets is 8%. What is the
optimal position in the futures contract that can help the portfolio manager
implement the CPPI in the presence of the illiquid position?
The optimal weight of the risky asset was shown to be 70%. The actual weight
is 50%. Given this information and the model discussed above, the optimal
futures position is:
 
α
Ft ≈ + β (kt − wt )
rq,t
 
0.04
≈ + 1.2 (0.70 − 0.50) = 34%
0.08
The portfolio manager will take a 34% long position in the futures contract.
ˆ Suppose the private equity position’s alpha has been close to zero in recent
years. What should be the optimal position in this case?
Ft ≈ β (kt − wt )
≈ 1.2 (0.70 − 0.50) = 24%
The optimal position in the futures contract is smaller now. More importantly,
we do not need an estimate of the expected return on the market.

30
5 Conclusion
In this note, we have reviewed the major portfolio rebalancing strategies. We showed
that no rebalancing strategy is uniformly better under various market conditions. The
choice of a rebalancing strategy critically depends on the objectives of the investor
or portfolio manager, market conditions, transaction costs, and even the tax status
of the asset owner.
A sound understanding of the basic strategies discussed here will be essential
before implementing more complex strategies for a multi-asset portfolio. Some of
the issues discussed in this note naturally extend to multi-asset portfolios. However,
there will be new challenges as well. For example, correlations among risky assets were
ignored in this note, while it will play an important role in designing and implementing
rebalancing strategies for multi-asset portfolios. Also, we ignored market conditions
where volatility, interest rates, and correlations among assets could change. More
complex models of asset returns and option pricing must be considered to account
for such dynamics. In these cases, simple analytical solutions will not be available.
Investors may need to use Monte Carlo simulation to understand the impact of various
rebalancing strategies on the risk-return profile of their portfolios.

31