Moving Average with Constant Weights

The equation
#
1
𝜎 = % 𝑅"!
!
𝑚
"$%

represents the calculation of the variance and indicates that, each of the known yields, has
equal weight in the estimation of the variance, deriving from there the name of constant
weights. From now on we will leave aside the assumption of constant volatility and the
developed models will assume that the volatility varies from one point of time to another.
The simplest methodology of this type of estimation is known as moving averages. In a
moving average the estimation of variance is based on the equation:
#
!
1 !
𝜎&,# = % 𝑅&("
𝑚
"$%

Where the estimate considers in addition to m values, a point of time t.

%
For example, we can have equation 𝜎!,#$ indicating that this is the variance estimate for
day three, using the last 10 days of information. The equation indicates that we have a
variance at each point of time and that variance is the average of the yields elevated to
the square of the last m days. Usually m takes values of 5,10, 20 and 40.

To determine which value of m better describes the behavior of the variance, a measure
of goodness of fit known as the Root of the Squared Error (RMSE) that is calculated

#
𝑅𝑀𝑆𝐸 = (& ∑(')#(𝑅'% − 𝜎'% )%
Where H corresponds to the total number of windows, that is to the number of
occasions in which with the data that are calculated the variance, and the best
estimate will be that which provides the smallest RMSE. Note that, in the estimation
of variance with the moving average, it is also considered a constant weight, since
#
the last m days of information used have a weight of * each one of them. However,

in this case the oldest performance value is 40 days ago which is not considered
widely historical.

Exercises
1. What weight does the estimating of the variance has in each of the yields when
m=40?
2. Copy the following data table into Excel and using the mobile average definition
completes the missing data. Shaded areas do not require information. Note:
yields were obtained using the percentage yield equation
Volatility Volatility
Time Price Yield
with with
m=5 m=10
74.78000
73.91000 -0.01163
74.72000 0.01096
75.19000 0.00629
74.06000 -0.01503
74.15000 0.00122
75.06000 0.01227 1.02%
75.23000 0.00226 1.04%
77.56000 XX 0.92%
77.72000 0.00206 1.64%
74.87000 XX 1.50%
77.44000 0.03433 XX 1.73%
74.39000 -0.03939 2.64% 2.01%
74.52000 0.00175 3.17% XX
73.34000 -0.01583 2.86% 2.33%
4 days ago 74.36000 0.01391 2.94% 2.33%
3 days ago 74.28000 XX XX 2.38%
2 days ago 76.99000 0.03648 2.00% 2.34%
 Yesterday 75.77000 -0.01585 1.89% 2.61%
Today 76.62000 0.01122 2.01% 2.47%
Tomorrow XX 2.50%

3. The above table shows that the estimated volatility with m=10 for tomorrow is
2.5%. Determine an estimate for the volatility of tomorrow with m= 5.

4. If for certain time series, models of variance have been estimated using moving
averages with m=5,10,20,40 And the following table of RMSE was obtained:
What value of m is the most suitable to use? Why?