This worksheet has been designed and tested using Netscape 3.0. It should work with later versions of Netscape's browsers but will not work with other browsers. It uses only JavaScript for computations and should cause no problems when used with the appropriate browsers. The operative word here is "should" -- the author cannot guarantee fault-free operation.

You should be able to use the worksheet when not connected to
the internet. Save the program file (ws_***.htm) and the
accompanying instruction file (wi_***.htm) on your disk using the
browser's command to **F**ile **S**ave.
At a later time you may retrieve the file using the browser's **F**ile
**O**pen file command; you may then use the page as
you would if you were on the network.

When using the worksheet, you may change any inputs. To do so, click inside the appropriate box, then make your changes. When finished, click any area outside the boxes on the form.

You may copy inputs from other sources such as spreadsheets,
word processing documents, and other worksheets in this series,
then paste the results into the appropriate boxes on this form.
To copy an area from an Excel spreadsheet, select it, then select
**E**dit **C**opy. In the browser,
select a position in the input box, then select **E**dit
**P**aste. To copy an area from a box in the browser
to an Excel spreadsheet, select the text in the browser and
select **E**dit **C**opy. In the Excel
spreadsheet, select a position, then select **E**dit
**P**aste. This brings each row into the spreadsheet
as text. To convert to a matrix, select the column in which the
information is located (the left-most one shown), then select **D**ata
**T**ext-to-columns. Choose **Delimited**
and **Spaces** as delimiters and the information
will appear in the requisite number of cells.

When you save a page on your own disk (using the browser's **F**ile
Save **A**s command), only the original material in
the form will be saved. To overcome this, you may click the **MAKE
RECORD** button. This will create a new page with the
relevant inputs and outputs from your most recent case. You may
print this or save it on your own disk. At a later time you may
open this file in your browser and copy information from it to a
regular worksheet, if you wish. When you are through with the
record page, choose either the **F**ile **C**lose
command or click the X in the upper right corner (or equivalent
on your platform) to return to the worksheet.

There are two other ways to save and retrieve worksheet information. You can copy the information you wish to save to some other document, such as a spreadsheet, word processing document or text file. You can also load the source (ws_***.htm) file in a word processor and edit it to include your inputs. You will find the default information in blocks marked TEXTAREA and in the VALUE attributes of INPUT tags. Simply replace the default values with your information, then save the page as a file on your disk under any desired name.

Whenever you change an input, the output area will be cleared
to avoid having old outputs appear simultaneously with new
inputs. To produce new outputs, click the **PROCESS**
button.

The **Asset Range and Returns** box should start
with a row giving the identifiers for the assets to be used in
the analysis. Each identifier should use six or fewer characters
(of any type), but no spaces. The next row should indicate the
minimum proportion for each asset class in the resulting style
and the following row should give the maximum proportion for each
class. It must be possible to construct at least one allocation
among the classes that lies within or at the border of each of
the specified ranges and adds to 1.0. The remaining rows contain
monthly returns for the desired asset classes with each month in
a separate row. In each row the first column contains the monthly
in the form YYYYMM (for example, 199301) and the remaining
columns the returns for the asset classes in that month. The
months covered by the rows must be sequential. Within each row
the entries must be separated by one or more spaces and/or tabs.
Blank rows may be included in the table and it is not necessary
for your data to "line up" in columns as long as it
conforms to the rules given above.

The **Fund Name **can contain any
desired characters except quotation marks or apostrophes. It
should be limited to fifty characters or less.

The **Fund Returns** box provides returns for the
fund to be analyzed. The first row contains only the single
identifier (Return). Subsequent rows provide the fund returns for
each month to be covered. The first item in each row is the
month; the second is the fund return. For the first observation
the month must be entered in YYYYMM format. Subsequent rows can
use any desired identifier (e.g. 1) since they will be considered
to be sequential observations in any event. The first fund return
must be for a month that is contained in the asset returns box.
The range of months covered will be the longest possible, given
the periods covered by the asset and fund returns.

The output is in the form of three tables, all contained in a single box.

The first two rows in the output box show the name of the fund and the range of months used for the analysis.

The first table shows the **Style **of the fund
for the period covered. Of all the **feasible**
combinations of asset classes, this is the one that
"tracked" the fund's returns most closely during the
period. In this connection, feasible combinations of asset
classes are those for which every asset proportion lies either at
its required minimum, at its required maximum, or at some point
between the minimum and maximum. Of all such feasible
combinations, the one shown in the table had the lowest **Selection**
**Standard Deviation** during the period.

The next table shows annualized values for two measures of
performance. The first row shows the annualized value of the
arithmetic **Mean** monthly returns, obtained by
multiplying each average monthly return by 12. The second row
shows the annualized value of the **Standard Deviation **of
monthly returns, obtained by multiplying each standard deviation
of monthly returns by the square root of 12. The monthly standard
deviation of the selection return is adjusted for degrees of
freedom based on the number of non-zero asset exposures in its
style -- the actual value is multiplied by the square root of
nobs/(nobs-numpos-1) where nobs if the number of observations
(months) and numpos is the number of assets with positive
exposures.

In the table, the first column shows the values computed from
the monthly returns for the** Fund**. The second
column shows the values computed from the monthly returns for the
combination of asset classes shown in the **Style**
table. The third column shows the values computed for the monthly
**Selection** Returns, where the selection return
for each month is the difference between the fund return and the
style return for that month.

The final table provides statistics for the analysis. The **Percent
Active **is computed by dividing the selection variance
(standard deviation squared) by the fund variance (standard
deviation squared), then multiplying the ratio by 100. This
provides a measure of the proportion of the fund variance due to
active management. The **Selection Sharpe Ratio (SSR)**
is computed by dividing the annualized mean selection return by
the annualized standard deviation of the selection returns. This
provides a measure of value added through active management per
unit of added risk. The **T-statistic** is computed
by dividing the mean monthly selection return by (the standard
deviation of monthly selection return divided by the square root
of the number of months analyzed). This provides a measure of the
statistical significance of the value added through active
management. The **Percentile** is the approximate
location of the T-statistic in a cumulative normal distribution.
For example, if the Percentile is 80, one might assume that in a
group of managers with zero skill, approximately 80% would have
poorer performance due to luck and 20% would have better
performance due to luck.

You may enter any desired text in this box to describe the source of the input data, etc..

This version January 2, 1997