My New Textbook on Performance Evaluation and Attribution, which Includes Examples of Usage of the DGTW Benchmarks (published December 2012)


Corrected Bibliography


End of Advertisement

Download Page for Daniel, Grinblatt, Titman, and Wermers (DGTW; 1997) Characteristic-Based Benchmarks

Monthly Benchmark Returns & Annually Reconstituted Benchmark Portfolio Assignment Identifiers

-- Covering all NYSE/AMEX/Nasdaq (United States) Stocks From 1975 to 2012 Having Complete Data (for a Given Year) in the CRSP/Compustat Merged Database --

(Excludes REITS, Closed-End Funds, ADRs, Canadian Firms (and Other Non-U.S. Incorporated Firms), Primes and Scores, and HOLDRs)

Maintained by Russ Wermers

University of Maryland

Updated October 8, 2013


I.  Rules for Use of These Data

A) You must have a legal subscription to the CRSP/Compustat Merged Database from the Center for Research in Security Prices (CRSP)

B) You must cite Daniel, Grinblatt, Titman, and Wermers (1997) and Wermers (2004) (both available below) in your working papers and published papers that use any of these data

C) You must place the following text in a footnote of your paper (to help other researchers find the benchmarks):  "The DGTW benchmarks are available via"

- If you cannot comply with the above, then please exit this page.  Otherwise, your downloading of these data signifies your acceptance of these terms -

- I hope these data are helpful for your research! -

II. Notes About Recent Changes to DGTW Procedure From Prior Years

Changes Made to the Entire Time-Series Starting with the September 2008 Vintage (which covered 1975-2007) and Continued Through Current October 2013 Vintage (which covers 1975-2012)

1.  If a fiscal year ends during January through May, I use this fiscal year-end book value to form the book-to-market variable for the following June 30 (previously, I followed Fama and French in using the one-year lagged fiscal year-end book value for such cases). This data should be available to the market by June 30.

2.  All stock returns are now delisting-adjusted, using CRSP delisting returns--when available. This assumes, therefore, that a stock is held through the delisting event and is not sold (even if the delisting is publicly known beforehand). In the case of a performance-related delisting without a CRSP delisting return, I use the procedures recommended by Shumway (1997, 1999).

3.  All CRSP/Compustat information is drawn from the "Fundamentals" file of CRSP/Compustat, with the exception of Post-Retirement Benefit Asset (PRBA), which is drawn from the Compustat "Pension Annual" table. Both of these datasets are available via Wharton Research Data Services (WRDS).

4.  (Starting with the July 23, 2011 Vintage):  I limit to stock-years having a CRSP/Compustat currency code of "USD" to more accurately exclude Canadian companies. In the past, I screened out Canadian companies by limiting to stock-years having a sharecode of 10 or 11, but a minority of Canadian companies slipped through this screen.

III.  Description of Data & Hyperlinks to Datasets (All Excel Datasets)

These data cover the period 1975 to 2012 (inclusive).  The benchmark assignments are conducted as described on page 7 of Wermers (2004), which also refers to the DGTW (1997) paper.  The main innovation in Wermers (2004) is that the book-to-market ratio is industry-adjusted in a different way--please carefully read that page to see the details.  Also, note that industry portfolios are formed using the 48 Fama-French industry classifications available on Ken French's website (  Benchmark assignments are reconstituted each June 30th (not quarterly, as in one version shown in the Wermers (2004) paper), starting June 30, 1975 and ending June 30, 2012.

1) Benchmark Returns for 125 DGTW Portfolios (1975 to 2012) (In .CSV [Comma-Separated Values] format--Can Be Opened Directly in Excel 2007 or Other Programs) 

Note that each row has benchmark returns that begin during July of the sort year, then continue through June of the following year.  For example, all rows (portfolios) (e.g., size_jun=1, book_m_jun=1, and mom_jun=1) corresponding to the year "1975" have the July, August, September, October, November, and December 1975 benchmark returns, followed by the January, February, March, April, May, and June 1976 benchmark returns.  Then, proceed to the 1976 row for the same benchmark cell (e.g., size_jun=1, book_m_jun=1, and mom_jun=1) to obtain the July-December 1976 and January-June 1977 benchmark returns for that cell (which is now populated with potentially different stocks).

2) Stock Assignments to Benchmark Portfolios (June 30 of Each Year, 1975 to 2012)

    - Stock Assignments (In One .CSV [Comma-Separated Values] Dataset--Can Be Opened Directly in Excel 2007 or Other Programs)   

3) Benchmark Cell Breakpoints (Through June 2010--will be updated later)

    Defined as Maximum Value, in Each Cell During Each Year, of: 

        i) Market Capitalization (in $Thousands)

        ii) Industry-Adjusted Book-to-Market Ratio (as Defined in Wermers (2004), Page 7)

        iii) Book-to-Market Ratio (Unadjusted, for Reference Only)

        iv) Momentum (Average Monthly Return During June 1 of Year t-1 to May 31 of Year t)

4) Total Number of NYSE Stocks in Each Cell During Each Year (Through June 2010--will be updated later)

5) Total Number of NYSE/AMEX/Nasdaq Stocks in Each Cell During Each Year (Through June 2010--will be updated later)

6) DGTW (1997) Paper

7) Wermers (2004) Paper


UNIQUE Visitors since January 2005 (Defined as a Visitor who has not visited the site during the past week)

*** Long-term statistics (click here) ***