|
|
|
|
|
|
|
|
|
 |
|
|
|
|
|||
Copyright © 2001 Zacks Investment Research, Inc. |
|||
• Current
• Dial-Up
•
Backtesting
Guide Benchmark
Return: The holding period return of the
selected benchmark over a specified period of
time. Beta:
The estimated sensitivity of a stock's return to
the returns of a broad-based index (S&P
500). Bias,
look-ahead: Using data in a backtest that
was not available as of the date that it was
used. Example: calculating a P/E ratio on
December 31 using calendar year-end earnings
that were not available until at least one month
later. Bias,
Survivor: When studying what would have
happened in the past, it is necessary to include
the companies that have disappeared as well as
those still in business. Example: screening for
money losing companies with stock price bellow
$10 as a value selection model will yield above
normal return if non-surviving companies are
excluded. Companies that "go research" through
acquisitions will also influence your results.
Use of Zacks backtesting databases (containing
both the current active and research companies)
alleviates this problem. Delta:
The difference between the mean and the
benchmark return. Excess
Return: The difference between the Portfolio
Return and the benchmark Return for one time
period. Holding
Period Return (HPR): The total return for a
stock form a base or starting date over a
specified period of time. (Zacks HPR's are
"forward-looking" - that is, the HPR value for
January 31, 20XX represents the return on the
stock that would be earned during the month of
February. In the context of backtesting, this
convention is necessary because when a stock is
purchased on January 31, the holding period
return of interest would be that earned after
the purchase date. Information
Coefficient (IC): A Spearman Rank
Correlation coefficient, which describes the
linear correlation between two ranked variables.
The ranks may be fractiles or ordinal ranks. The
coefficient can range in value from -1 (perfect
negative correlation) to +1 (perfect positive
correlation). The IC is
frequently used in backtesting because it is not
subject to the limitations imposed by assuming
that the distribution of a test variable is
normal (i.e., there are no extreme outlier
values). Outliers:
Extreme values for a database item. Outliers may
be the result of data entry errors, or may be a
consequence of the nature of the underlying
data. A common example is the P/E ratio for a
company which, for several years, has reported
EPS of 40-50 cents per share, but for the
current year reports minimal earnings (e.g.,
$0.01 per share). A market price of $15 would
then result in a P/E of 1500. Performance
Measure Item: Generally, a database item
used to calculate portfolio and stock returns (1
month HPR). Test
Variable: A database item on which a
backtest will be performed. Turnover:
A measure of the change in a portfolio's
holdings over a period of time. Portfolio
turnover has significant implications for the
performance of a portfolio; allowing high
turnover in an attempt to capture the best
performance will result in high trading
cots.
Copyright © 2001 Zacks
Investment Research, Inc.
Equity Research Products:
•
Internet-Integrated
• PC-Based
• Web-Based
Equity Analysis Data:
• Historical
Data
Delivery Options:
• CD-ROM
• FTP/Internet
Backtesting
• Glossary
• Stump
The Quant!
This
page describes some backtesting terms and
concepts. The list, while not exhaustive, should
enhance your understanding of the prompts
encountered when performing backtests with Zacks
software and the information contained in
Backtest reports.
Benchmark:
The standard of comparison for the performance
of a portfolio. The portfolio performance may be
compared against the full universe of companies
in a databases, an index (e.g., the S&P 500
index), or another portfolio.
About
Zacks | Contact
Zacks | Knowledge
Base |Feedback