“Evaluating the performance of the Motley Fool’s Stock Advisor™”

Since March 2002, The Motley Fool’s founders, David Gardner and Tom Gardner, have published monthly stock recommendations under Motley Fool’s premium Stock Advisor service. In this paper, the authors investigate whether analysts’ recommendations can add value for investors by examining the performance of portfolios con-structed based on Motley Fool’s recommendations. They evaluate the announcement effect on share price corresponding to the publication of stock recommendations. Additionally, the researchers examine holding period returns for a portfolio imitat-ing the actions of Stock Advisor. They find portfolios composed of recommendations through Stock Advisor added value initially upon recommendation and across extended holding periods. Additionally, the authors find that the Stock Advisor sample outperforms other sample portfolios on a risk-adjusted basis and over several subperiods. The findings contribute to the literature on the usefulness of analysts’ recommendations in adding value to investors’ portfolios.


INTRODUCTION
http://www.alexa.com/topsites/category/Home/Personal_Finance Analyst stock recommendations are exceedingly prevalent and accessible, which is attributable to the progression of the Internet and mobile devices. Searching for stock recommendations on Google generates approximately 151 million results. According to Google Trends (2017), numerous sources remain and are dedicated to providing their input regarding the stock market. Popular online sources for recommendations include Barron's, TheStreet, J. P. Morgan, CNBC, Forbes, and Kiplinger.
One popular investing website is The Motley Fool, which ranks sixth among home/personal finance websites in terms of monthly visits 1 .
The Motley Fool provides an extensive range of stock news and analysis on its free website, www.fool.com. Additionally, the website provides premium stock recommendations including its Stock Advisor™ service. Beginning in March 2002, brothers David and Tom Gardner, who founded The Motley Fool, began recommending stocks on a monthly basis. They only propose to sell recommendations for securities that they have previously endorsed. Additionally, each of the two discloses their five favorite stocks each month, which are securities they have previously recommended.
In this paper, we evaluate the performance of the proposed excess return-generating strategy of The Motley Fool's Stock Advisor premium service. We benchmark the performance of recommendations of the service using a matched sample of companies based on company size and book-to-market ratio, a matched sample of companies recommended by analysts, as well as the overall market. We find portfolios composed of recommendations through Stock Advisor added value initially and across extended holding periods, compared to the market and matched samples. Additionally, we find that the Stock Advisor sample outperforms other sample portfolios on a risk-adjusted basis over several periods.

LITERATURE REVIEW
Investors may rely on stock recommendations to expand their portfolio by adding an active component. Altınkılıç et al. (2016) analyze post-revision drift (PRD) following the revision of stock recommendations by sell-side security analysts. They refer to a sample from 2003 to 2010 that asserts that average PRD is relatively close to zero. They argue that high transaction costs create inefficiencies that attract arbitrage-seeking investors. The presence of high frequency trading has led to investors quickly reacting to recommendations that minimize PRD and arbitrage opportunities. Barber et al. (2001) examine the ability of investors to profit from security analysts' recommendations. They utilize consensus data from 1985 to 1996 and discover that investors who followed the consensus for recommended securities earned an 18.8 percent return, while stocks with the least favorable outlook earned 5.78 percent. As a reference, investors who maintained a value-weighted market portfolio earned 14.5 percent. Barber  Stephan and Nitzsch (2013) compare the performance of online stock recommendations between a larger population and a subset of more experienced investors within an online community. They evaluate over 60,000 stock recommendations, which include both German and U.S. stocks. The authors use cumulative abnormal returns based on the capital asset pricing model to evaluate performance before and after the publication of a recommendation and find minimal abnormal returns when considering transaction costs. We report the summary description of the Stock Advisor sample and subsamples in Table 1. About 32 percent of the recommendations (109 out of 340) belong to manufacturing segment, followed by services segment (89.26 percent), wholesale and retail services (55.16 percent) and finance, insurance, and real estate (46.14 percent). In the repeated recommended sample, stocks from services and finance, insurance and real estate segments seem to have higher probability to be recommended again following the initial recommendation. For example, about 32 percent (55 out of 167) and 19 percent (32 out of 167) of the repeated recommendations are from services segment and finance segment, respectively. Both these percentages are higher than their respective percentages in the whole Stock Advisor sample.

METHODOLOGY AND RESULTS
First, standard event methodology is utilized to produce abnormal returns regarding the announcement of stock recommendations through Motley Fool's Stock Advisor™. To be included in the sample, the recommended firms must meet the following criteria: A total of 340 buy recommendation announcements are evaluated in the event study. We consider 0 t = as the publication of the recommendation newsletter.
We report the share price reaction to the publishing of the newsletter beginning five days prior to the actual event "date". The market model is used to approximate expected returns, and expected returns are estimated during the interval (-5, 5).
Second, we investigate whether long-term effects are present by comparing holding period returns of the Stock Advisor recommendations to the performance of two matched benchmark portfolios and the S&P 500 Index.
Our first matched sample is matched based on market capitalization and book-to-market ratios in the same industry (using the/a two-digit SIC code). In order to identify our match sample, we utilize prior year-end closing price and market capitalization of all stocks with available data from CRSP for each year. We characterize BE/ME ratios as the book value of common equity from Research Insight, divided by the year-end market value of common equity of the prior year. For this study, we eliminate firms with negative book to common equity ratios. Potential selections for matching firms include securities that have not been recommended through Stock Advisor and have obtainable data from CRSP and Research Insight. To determine the most appropriate match for each firm in our Stock Advisor sample, we use Equation 1 to calculate the following matching score (MS) for each recommended stock against the remaining stocks:  The Sharpe Index (1966, 1994) examines excess return per unit of total risk, which is defined as mean monthly difference between the portfolio (market return) and the T-bill return divided by the standard deviation of the monthly return differences. The Sharpe Index provides insight on the risk-adjusted return of investors who are following the recommendations of Stock Advisor, which may not always be as diversified as the overall market. Note: Table 2 displays descriptive statistics for the Motley Fool Stock Advisor sample and matched samples. We calculate the prior year-end market capitalization and BE/ME ratio of each stock. We characterize market value of common equity (ME) as the prior year share price times the quantity of shares outstanding. We define the BE/ME ratio as the book value of common equity from Research Insight, divided by the year-end market value of common equity of the prior year.
The Treynor (1965) Index measureis more suitable to consider when an investor has a diversified portfolio. Instead of using sample standard deviation of the monthly return differences in the denominator, it uses systematic risk (i.e., portfolio beta as a proxy).
Additionally, we calculate Jensen's (1968) Alpha using Equation 2, which assesses the differential return of a portfolio compared to the expected return from a benchmark index. We compute Jensen's Alpha, α, which is the intercept term in a regression analysis, for the Stock Advisor portfolio (and matched portfolio) against excess market returns: ( A positive (negative) Alpha indicates a positive excess return of the Stock Advisor portfolio relative to the return of the market portfolio.
We also determine long-term abnormal returns by calculating buy-and-hold abnormal returns (BHARS) as outlined by Barber and Lyon (1997). BHARs are computed by subtracting simple buyand-hold returns on two matched portfolios, respectively, from simple buy-and-hold returns on the Stock Advisor portfolio. According to Barber and Lyon, this analysis eliminates potential bias from summing daily and monthly abnormal returns. In order to test the null hypothesis, which is that BHARs are equal to zero, we calculatethe t-test statistic of the BHARs.
We also test the long-run performance of the Stock Advisor sample using the Fama-French (1993) 3-factor and 4-factor models (Jegadeesh and Titman (1993)). The 3-factor (4-factor) model expands on the capital asset pricing model (CAPM) by adding size and book-to-market factors (size, book-to-market and momentum factors) to the market risk factor in CAPM. Specifically, the 3-factor model ( R -the return on one-month -bills; T mt R -the return on a value-weighted market index; t SMB -the return on a valueweighted portfolio of small stocks less the return on a value-weighted portfolio of big stocks; t HML -the return on a valued-weighted portfolio of high book-to-market stocks less the return on a valueweighted portfolio of low book-to-market stocks; t UMD -the return on the two prior high return portfolios less the returns on the two prior low return portfolios.
A positive intercept for these regressions, , α indicates that after controlling for the market, size, book-to-market ratio, and momentum factors in returns, the sample firms have higher returns than expected. To determine whether the regression intercepts, , α are significantly different from zero, we report t-statistics.

EVENT STUDY RESULTS
The results from the event study for the publication of Stock Advisor security recommendations are shown in Table 3. Using the newsletter publication date as t = 0, we find that the announcement of recommendations through Stock Advisor generates a significant positive market reaction. While there is a negative abnormal return on date -1, which is statistically significant at the 5 percent level, abnormal returns on date 0 and date 1 are 0.64 percent and 0.46 percent, respectively, and statistically significant at the 1 percent level. Moreover, the Stock Advisor portfolio generates statistically significant positive abnormal returns over all evaluated intervals. Specifically, the portfolio produces a cumulative abnormal return of 0.41 percent from dates -1 to 0. Over the interval from dates -5 to 5, the portfolio earns cumulative abnormal returns of 1.51 percent. Both of these interval results are statistically significant at the 1 percent level. Comparing the cumulative abnormal returns (CARs) across different subperiods, the Stock Advisor sample yields a highest CAR of 1.97 percent (statistically significant at the 1 percent level) in years 2007-2011. Over the (-5, 5) event window, the repeated recommended stock sample only yields 1.30 percent CAR (statistically significant at the 10 percent level), which is lower than the CARs for the whole Stock Advisor sample. A possible explanation could be that market only reacts to the new information, so a repeated recommendation may not be considered as a surprise to the market and therefore a less announced effect.
Overall, the event study results in Table 3 indicate that the Stock Advisor sample does generate statistically significant abnormal returns around the event dates, which suggests that The Motley Fool, and specifically Stock Advisor, possess a substantial following of investors who value their analysis and react to new recommendations quickly.   Table 3 displays the outcome of the event study for the Motley Fool Stock Advisor sample and sub samples. We examine the reaction of the share price to the release of the recommendations beginning five days prior to the event date by calculating abnormal returns (ARs) and cumulative abnormal returns (CARs). Expected returns are approximated during the interval (-5, 5). ***, **, * indicate statistical significance at 0.01, 0.05 and 0.10 level, respectively.

LONG-TERM EFFECTS
Jensen's alpha for the Stock Advisor sample (statistically significant at the 1 percent level) exceeds Jensen's alpha for the matched sample but has a lower alpha when compared with the IBES matched sampleover the whole period. Overall, the repeated recommended stock sample yields higher raw returns and risk-adjusted measures when compared with their matched samples and S&P 500 index. For example, the repeated recommended stock sample yields a monthly raw return of 1.30 percent, which is higher than the monthly raw returns of 1.13 percent for the whole Stock Advisor sample. It is also higher than the raw monthly returns of both matched samples and the S&P 500 index. Table 5 displays the results of the two regressions for four multi-year periods. We only report the regression intercepts for brevity. Our results indicate that after controlling for additional factors covered by the Fama-French models, we do observe statistically significant superior performance (at the 1 percent level) for our Stock Advisor sample for the whole period. By comparing the results of different sub-periods, we only find statistical significant intercepts (at the 10 Our repeated recommended stock sample yields statistically significant intercepts (at the 1 percent level) for both Fama-French 3-and 4-factor models for the entire time period and for the first subperiod (at the 10 percent level).  in the years 2011, 2013, and 2015, but these figures are not statistically significant.

BUY AND HOLD ABNORMAL RETURNS RESULTS
We find similar but less statistically significant results when compared with the IBES matched sample. Specifically, the results show that for seven out of fifteen years in our sample period, the Stock Advisor sample produces higher buy-and-hold returns compared to its IBES matched sample. In addition, over the whole sample period, the Stock Advisor sample yields statistically significant (at the 10 percent level) BHARs of 0.851 percent compared to its matched sample. These indicate that in our overall sample period, the Stock Advisor sample outperforms both its matched samples.
Taken together, the Stock Advisor portfolio generates significant BHARs over the whole period, which signifies the exceptional stock-picking ability by the Stock Advisor service. Given that securities in the Stock Advisor sample and matched samples are similar in terms of market capitalization and BE/ME ratio, the service recommended the overachieving stock with relative consistency. This performance is attributable to exceptional performance during the initial sample years of the Stock Advisor service, specifically in 2004. The performance of the sample declined substantially from 2010 to 2016, as indicated by both low and negative BHARs values during this period.

CONCLUSION
In this paper, we have examined the performance of securities recommended through Motley Fool's Stock Advisor service. We find that the Stock Advisor recommendations do statistically outperform the matched samples and S&P 500 index, since the creation of Stock Advisor in 2002 regarding both

BHAR: (1) -(3) t-stat
short-term and long-term holding periods. Event study results indicate a statistically significant market reaction on the day the recommendation is announced and the subsequent two days, which indicates a favorable reaction by investors to the recommendation. Over a longer holding period, the Stock Advisor portfolio repeatedly outperforms the S&P 500 index and matched samples in terms of monthly raw returns and risk-adjusted measures. Additionally, regression results for Fama-French 3-and 4-factor models reveal statistically significant abnormal returns for the Stock Advisor portfolio over the whole period. The performance of the Stock Advisor portfolio also exceeds the matched samples in generating buy and hold abnormal returns. Although the overall performance of the Stock Advisor portfolio benefits from remarkable recommendation performances between 2002 and 2006, the portfolio still exceeds the benchmarks regarding risk-adjusted measures during the subsequent period between 2007 and 2011. It is evident that investors who follow Stock Advisor's recommendations to build their portfolio outperform the S&P 500 index and the matched samples to an extent over the whole period, although the portfolio benefits from particularly favorable investments during the initial sample years. Additionally, the results indicate that investors react favorably to the release of recommendations through Stock Advisor.