Two numbers describe a portfolio: its expected return and its variance. The first is built from the shares like an ordinary average, the second is not: the covariance enters the variance.
This is about money spread over several investments at once, some in a deposit, some in shares. The set brings income and carries risk. Exam tasks ask for its average income and spread. The risk here is computed with the covariance, which shows how much two quantities move together.
Anna has 10 000 € and splits it into three parts:
2 000 € go into a deposit at 2.5% a year;
4 000 € buy shares in one firm;
4 000 € buy shares in another.
In a year the deposit surely pays its 2.5%. What the shares bring nobody knows: maybe a lot, maybe they lose value. Anna wants two numbers: how much she gets on average and how far the result can stray from it.
Both numbers depend on how the money is split. More in the deposit means less risk but also less average income. Two stocks that rarely fall together lower the risk without costing income. Below we compute both numbers.
Let the first stock bring 10.5% on average and the second 7.5% (these figures come back below). Then the whole set brings on average 0.2⋅2.5+0.4⋅10.5+0.4⋅7.5=7.7%, about 770 € a year. The second number is harder: the covariance of the two stocks goes into it. With this article's figures the result typically lands about 1100 € above or below 770 €.
Check yourself. A task at exam level: 30% of the money sits at the fixed rate r=0.02, the rest is split equally between two stocks with expected returns μ1=0.06 and μ2=0.10. What return does the whole portfolio expect?
Return: what each invested euro brings
Anna buys one share for 50 €. A year later the share is worth 52 €, and during that year the firm paid its owners 1 € per share. Anna earned 52−50+1=3 €.
To compare investments of different size, the earnings are divided by the amount invested. That is the return (German “Rendite”):
return for the year==price todayprice in a year−price today+payouts==5052−50+1==0.06
A return of 0.06 means: each invested euro brought 6 cents. Tasks write it as a decimal, 0.06 rather than 6%.
The deposit's return is known in advance: 0.025. Such a security is called safe or risk-free, and its return is written r. Nobody knows today what a share will cost in a year, so its return is a random variable, like the kiosk sales in the article “Covariance: how two quantities move together”.
For a stock, a year is either bad or good, with probability 21 each: in a bad year the return is −0.06 (the share lost value), in a good one 0.16. Then:
the expected return is μ=21⋅(−0.06)+21⋅0.16=0.05;
the variance of the return is σ2=21⋅(−0.11)2+21⋅0.112=0.0121.
The expected return says how much the share brings on average. The variance says how big the risk is. A safe security has variance 0. Tasks usually give μi and σi2 for each stock straight away, and the covariance σ12 for the pair.
The portfolio: shares and return
Anna's money sits in three places at once: in the deposit and in two stocks. All of it together is called a portfolio (German “Portfolio”). Instead of euros it is handier to count shares (German “Anteile”), that is, which part of all the money sits in each security:
a0=100002000=0.2 in the deposit;
a1=100004000=0.4 in the first stock;
a2=100004000=0.4 in the second.
The shares always add up to 1: a0+a1+a2=1. Let the stocks have μ1=0.105, μ2=0.075, σ12=0.045, σ22=0.015, covariance σ12=0.0075, and the deposit r=0.025. These numbers will come back more than once.
Deposit: 2000⋅0.025=50 €. First stock on average: 4000⋅0.105=420 €. Second: 4000⋅0.075=300 €.
Together 770 € on 10000 €, so the portfolio's return is 0.077 on average.
The portfolio return P is the returns of the securities weighted by the shares. The expected value adds up the same way:
The safe security does not swing, so it does not enter the variance. What is left is the variance of the combination a1⋅X1+a2⋅X2 from the article “Combinations aX + bY”, where X1 and X2 are the returns of the two shares and the shares act as weights:
One more form turns up in almost every task. Substitute a0=1−a1−a2:
E(P)==r++a1(μ1−−r)++a2(μ2−−r)
The differences μi−r are called excess returns (German “Überrenditen”): how much more than the deposit the stock brings on average. For Anna 0.025+0.4⋅0.08+0.4⋅0.05=0.077, the same number.
Expected return from the shares
Anna's shares were known, and the return came from substituting them. In the exam it is usually the other way round: the return of the whole portfolio is given and one share has to be found. This is how the task looks in the exam:
Exam-style task
Eine Wohnbaugenossenschaft hält 20% ihrer Rücklage als Festgeld F, das sicher 0.02 Rendite bringt. Der übrige Teil geht in zwei Fonds: W mit der erwarteten Rendite 0.06 und X mit 0.03. Der Vorstand verlangt für die ganze Rücklage eine erwartete Rendite von 0.043.
Welchen Anteil der ganzen Rücklage muss die Genossenschaft in den Fonds W legen?
a)0.6250
b)0.4333
c)0.3895
d)0.2050
e)0.5000
The task gives itself away by three things: a safe part that “surely returns” a known rate, the expected returns of the other parts, and a return the whole portfolio must reach. One share is unknown. Write E(P) as a sum of “share times return” and solve one linear equation.
The numbers of the task: the reserve consists of the deposit F at 0.02, which holds 0.2 of the reserve, and two funds, W with expected return 0.06 and X with 0.03. The whole reserve is to expect 0.043. Here X is the name of a fund, not a random variable. The question asks for the share of the whole reserve in W.
shares of the parts: the deposit holds 0.2, the rest is split between two fundscontribution of a part: its share times its returnall contributions together equal the target returnthe share of fund W is 0.5, option (e)
Step 1 of 4Write down the shares
The deposit F holds 0.2. The rest, 1−0.2=0.8, is split between the funds: w in W and 0.8−w in X. The letter w is the share of the whole reserve, which is what the question asks for.
Each part gives “share times return”: 0.2⋅0.02=0.004, 0.06w and 0.03(0.8−w)=0.024−0.03w.
0.004+0.06w+0.024−0.03w=0.043, so 0.028+0.03w=0.043.
0.03w=0.015, w=0.5. Option (e). Check: X gets 0.3, and 0.004+0.03+0.009=0.043.
Ask yourself: why does X get 0.8−w and not 1−w?
A portfolio is invested like this: 25% at r=0.02, 50% in a security with μ=0.08 and 25% in a security with unknown return x. The portfolio expects 0.0645. What is x?
0.25⋅0.02=0.005, 0.5⋅0.08=0.04, the third part 0.25x.
0.005+0.04+0.25x=0.0645, so 0.25x=0.0195.
x=0.078.
How much to add. A portfolio of 10 000 € has return 0.05: everything is in a security with μ=0.05. How many euros z must be added in a security with μ=0.08 for the whole portfolio to expect 0.056?
After the purchase the portfolio holds 10000+z euros.
The old part gives 10000⋅0.05=500 €, the new one 0.08z: together 500+0.08z.
The income must be 0.056 of the new capital. Write down the equation and find z.
Your turn1 of 2
30 % of the money sits at a fixed r=0.03, the rest is split between securities A (μ=0.09) and B (μ=0.05). The portfolio is to expect 0.056. What share of all the money goes into A?
Traps. Which mistake each wrong number gives away:
Share of the rest, option (a), 0.6250: 0.5/0.8, the share of W within the funds. The question asks for the share of the whole reserve.
Deposit forgotten, option (b), 0.4333: the whole reserve split between W and X, 0.06w+0.03(1−w)=0.043. The deposit holds 0.2 and contributes its 0.004.
Plain average instead of the weighted one. If the left side of the equation is 30.02+0.06+0.03≈0.0367, the shares were lost: the left side must be the sum of “share times return”, and it equals 0.043.
This is not on the formula sheet. Remember: E(P)=a0⋅r+a1⋅μ1+a2⋅μ2, and the shares add up to 1.
Without a safe security: return and variance as functions of one share
The covariance did not enter the return, and the sums were easy. In the next task there is no safe security, and next to the return you write down the variance, where the covariance does appear:
Exam-style task
Ein Forstbetrieb liefert Holz zum Festpreis an ein Sägewerk; der Rest geht in Brennholz (A) und Christbäume (B). Die Anteile am Portfolio P sind a0 (Sägewerk, r=0.05), a1 (Zweig A) und a2 (Zweig B). Die Kovarianz der beiden Renditen beträgt σ12=−0.01. Für die Renditen gilt μ1=0.1, μ2=0.075, σ12=0.08 und σ22=0.065 (Index 1: Zweig A, Index 2: Zweig B).
Welche der folgenden Aussagen ist richtig, wenn a0=0 ist?
The task shows itself by a model of three securities and the condition a0=0. The model is given by six numbers: the rate r of the safe part, the expected returns μ1 and μ2, the variances σ12 and σ22, the covariance σ12. Without the safe security a1=1−a2, and both quantities become functions of one share a2: a line for E(P) and a parabola for V(P). In the task, firewood and Christmas trees are two securities, A and B, with the shares a1 and a2, and the safe part a0 is the delivery at a fixed price.
The numbers of the model: expected returns μ1=0.1 and μ2=0.075, variances σ12=0.08 and σ22=0.065, covariance σ12=−0.01.
Step 1 of 4Drop the safe security
a0=0, so a1=1−a2. The ends are known at once: at a2=0 everything is in branch A, V=σ12=0.08; at a2=1 everything is in B, V=σ22=0.065.
E(P)=0.1(1−a2)+0.075a2=0.1−0.025a2. Slope μ2−μ1=−0.025, constant term μ1.
V(P)=0.08(1−a2)2+0.065a22+2⋅(−0.01)(1−a2)a2.
With a22: 0.08+0.065−2⋅(−0.01)=0.165. With a2: −2⋅0.08+2⋅(−0.01)=−0.18. Constant term 0.08.
E(P)=−0.025a2+0.1 and V(P)=0.165a22−0.18a2+0.08. Option (d). Check at the end: at a2=1 we get 0.165−0.18+0.08=0.065=σ22. The vertex of the parabola in the picture, a2≈0.55, is the subject of the next article: there the share with the smallest variance is found.
Ask yourself: why is the constant term of the variance σ12 and not σ22?
At a0=0 the rate r enters nothing. It is given to see whether you put it in out of habit.
The statement about the Lagrange method. If one of the statements is about the Lagrange method, it is settled in the section “The Lagrange equations” of the article “Minimum variance: which portfolio weights to choose”.
Another model has μ1=0.12, μ2=0.07, σ12=0.06, σ22=0.03, σ12=0.01, r=0.03, a0=0.
E(P)=0.12+(0.07−0.12)a2=0.12−0.05a2.
With a22: 0.06+0.03−0.02=0.07. With a2: 0.02−0.12=−0.10. Constant term 0.06.
V(P)=0.07a22−0.10a2+0.06. At a2=1: 0.07−0.10+0.06=0.03=σ22.
Two functions of the share a2. Given σ12=0.03, σ22=0.07, σ12=0.01, μ1=0.05, μ2=0.09, a0=0. Write E(P) and V(P) as functions of a2.
E(P)=0.05+(0.09−0.05)a2.
Find the three numbers of V(P): with a22, with a2 and the constant term. Check with the end a2=1.
Your turn2 of 2
For two stocks μ1=0.12, μ2=0.07, σ12=0.06, σ22=0.03, σ12=0.01; the fixed-income security has r=0.03. What is the portfolio variance at a0=0?
Traps. Four mistakes and the numbers they give:
Securities swapped, option (a): E(P)=0.025a2+0.075 and constant term 0.065, as if the variable were a1. Check at the end: at a2=0 everything is in A, and V must be 0.08.
Sign of the linear term, option (e): +0.18a2. The (1−a2)2 brings −2a2, so the sign is minus. Check at the end: 0.165+0.18+0.08=0.065.
Both mistakes at once: option (b) flips the slope of E and the sign of a2 in V, option (c) swaps the securities and the sign.
Covariance not doubled: 0.155a22−0.17a2+0.08. The ends still fit, so look at the coefficient of a22: it must be 0.165.
On the formula sheet: V(a⋅X+b⋅Y)=a2⋅V(X)+2a⋅b⋅Cov(X,Y)+b2⋅V(Y) with a=1−a2 and b=a2. Multiplying out the brackets is up to you.
At a glance: return and risk of a portfolio
The topic in a minute
Return: random for a stock, equal to r for a safe security:
μi−r: how much more the stock brings on average than the fixed-rate security, whose r you give up for this part.
What it means
What role does the fixed-rate security play in V(P)?
What role does the fixed-rate security play in V(P)?
None directly: its variance is 0 and it has no covariance with any security. V(P)=a12σ12+a22σ22+2a1a2σ12 contains neither r nor a0.
What it means
What does the Lagrange method do in the portfolio model?
What does the Lagrange method do in the portfolio model?
It optimises one quantity with the other one fixed: minimum variance at a fixed return, or maximum return at a fixed variance. Optimising both at once is not possible.
Which method
Statements about E(P) and V(P) as functions of a2, with nothing in fixed-rate securities.
Statements about E(P) and V(P) as functions of a2, with nothing in fixed-rate securities.
Substitute a1=1−a2. Check: at a2=0 you must get V=σ12, at a2=1 you must get V=σ22.
Which method
You are asked for the system of equations for minimum variance at a given expected return, not for its solution.
You are asked for the system of equations for minimum variance at a given expected return, not for its solution.
∂V/∂ai=λ(μi−r) for i=1,2, and as the third equation E(P)=E0. The λ goes with the derivative of the constraint.
Several options name optimal portfolios for different target returns.
Several options name optimal portfolios for different target returns.
The ratio a1:a2 is the same in all optimal portfolios; the target return only changes how much goes into the shares in total. Options with a different ratio drop out.