STARMINE MONITOR

S TA R M I N E MONITOR

StarMine Analyst Performance Measurement Methodology

StarMine collects data on every estimate, revision, and recommendation recorded by the Thomson Financial I/B/E/S database. StarMine has worked extensively with its installed base of blue-chip institutional investors and global investment banks to perfect its metrics, which have proven to be highly predictive of future analyst performance.

Performance of Analyst Recommendations StarMine has developed two methods, based on different benchmarks, of measuring overall analyst stock-picking performance. StarMine's Absolute Return metric assumes that a "Buy" recommendation corresponds to an analyst expectation that the stock will have a positive absolute return and a "Sell" recommendation a negative absolute return; that is, cash is the benchmark. In contrast, StarMine's Excess Return builds a long-only portfolio around the analyst recommendations using an overweight / equal-weight / underweight strategy, where the benchmark is the analyst's coverage universe.

Absolute Return rewards analysts with a good sense for the overall direction of the individual stocks in their portfolios. Excess Return rewards analysts who successfully distinguish their top-performing stocks from their bottom-performing stocks, regardless of broader market or industry movements. Both measures are important for a complete understanding of an analyst's performance.

Absolute Return StarMine measures how much investors would have made had they built a non-leveraged portfolio around all of the analyst's picks, by investing 1 unit of local currency in each Buy recommendation, 2 units in each Strong Buy, 1 unit in cash for each Hold, short selling 1 unit in each Sell recommendation, and shorting 2 units in each Strong Sell recommendation. The portfolio is rebalanced each month and whenever the analyst adds coverage, drops coverage, or changes a rating.

Excess Return StarMine's Excess Return measures how well analyst recommendations distinguish among the stocks in the analyst's coverage universe. For example, do the Strong Buys outperform the Buys, and so on? It is computed as the difference between an analyst's RecommendationWeighted Return and his Coverage Return.

The Recommendation-Weighted Return is computed as a non-leveraged, long-only portfolio of all stocks in an analyst's coverage using an overweight / equal-weight / underweight strategy. StarMine invests 2 units of local currency in each Strong Buy recommendation, 1.5 units in each Buy, 1 unit in each Hold, ? unit in each Sell, and 0 units in each Strong Sell. The portfolio is rebalanced each month and whenever the analyst adds coverage, drops coverage, or changes a rating. Because

the portfolio is constructed using relative weights, it is equally effective for analysts who use three recommendation levels (buy, hold, sell) and for analysts who use five recommendation levels (strong buy, buy, hold, sell, strong sell).

Coverage Return is a personal benchmark for each analyst, indicating how a portfolio would have done had the analyst simply put a Buy recommendation on all of the stocks in his coverage. It is calculated by investing equal amounts in each covered stock during the period that the analyst covered it. The portfolio is rebalanced at the end of every month and when an analyst adds or drops coverage.

Because the Recommendation-Weighted Return and the Coverage Return are always fully invested, the difference between them, or Excess Return, is not subject to the overall direction of the market.

The Star Rating In order to fairly compare analyst skill across disparate industries with different variance and returns characteristics, StarMine provides the Coverage-Relative Score. It is derived from the ratio of the Excess Return to the cross-sectional standard deviation of the returns of the stocks in the analyst's coverage universe. The resulting measure isolates the skill of the analyst in distinguishing among the stocks in his coverage and is independent of the direction of the market or the variance across the particular stocks.

StarMine ranks all analysts in its database according to the ratio and displays the Coverage-Relative Score as a percentile ranging from 1 (low) to 100 (high). StarMine then awards the Coverage-Relative Rating of 1 to 5 stars according to the distribution in the chart below.

Star Rating

% Awarded

10 23 34 23 10

StarMine awards separate star ratings for performance of buy/ sell/hold recommendations and for earnings estimate accuracy. The stock picking ratings are based on the Coverage-Relative Score; the earnings estimate ratings are based on StarMine's Single-stock Estimate Score (SES). The top ten percent earn the 5-star rating, the next 23 percent the 4-star rating, and so on.

S TA R M I N E MONITOR

Market Cap and Equal-Weighting StarMine creates two measures for Absolute Return, Excess Return, and Coverage-Relative Score: one that puts more weight on stocks with larger market capitalizations and one that treats stocks of all market caps equally. The market cap weighting function doubles the amount of weight put on each stock for each order of magnitude increase in market cap.

Industry and Recommendation-Level Returns StarMine computes returns around various subsets of an analyst's coverage.

Industry Returns StarMine calculates industry-level returns for many countries and regions around the world. StarMine ranks analysts within industries according to its Industry Excess Return. The benchmark for this portfolio is the market-cap weighted return of all stocks in the industry.

For each Buy recommendation, the portfolio is one unit long the stock and simultaneously one unit short the benchmark. The result gives the analyst credit for the amount the stock outperformed the benchmark. Strong Buys get a larger investment of two units long the stock and two units short the benchmark. Holds invest one unit in the benchmark (i.e., for an excess return of zero). Sells are the reverse: long the benchmark and short the stock. Strong Sells get a larger investment of two units long the benchmark and short the stock.

The resulting portfolio is rebalanced each month and whenever the analyst adds coverage, drops coverage, or changes a rating.

Recommendation-Level Returns StarMine measures the performance of each stock under coverage separately for each level of recommendation used. For example, StarMine measures the average return that stocks make while an analyst maintains a Buy. These can then be compared to the average return of stocks with Holds or Sells.

StarMine measures these returns two ways: one relative to cash (absolute returns) and one relative to the analyst's coverage over the same interval.

Earnings Estimate Accuracy StarMine's proprietary metric, Overall Estimate Score (OES), measures the overall accuracy of each analyst's earnings forecasts. OES is a relative measure; that is, analysts are compared against their peers. OES ranges from 0 to 100, with 50 representing the average analyst. To get a score higher than 50, an analyst must make estimates that are both significantly different from and more accurate than other analyst estimates.

StarMine's measurement of estimates takes into account many factors: the analyst's absolute forecast error, the analyst's error compared to other analysts, the variance of the analysts' errors, the timing of the estimates, and the absolute value of the actual earnings for the stock. A Single-stock Estimate Score (SES) is computed daily for every stock an analyst covers, and then each stock's SES is aggregated to provide an OES for a group of stocks or for the analyst's entire coverage.

StarMine converts OES to a star rating by awarding the top ten percent of analysts a 5-star rating, and so on, according to the chart on the previous page.

StarMine is a registered trademark of StarMine Corporation.

Copyright ? 1999-2005 StarMine Corporation. All rights reserved. Printed in the U.S.A.

For more information on StarMine Monitor, please contact your StarMine representative:

monitor@

United States +1 (415) 777-1147

Europe +44 (20) 7680 3096

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download