A reader asked: “Have you done an analysis of the Barron’s Daily Stock Alert service?” Barron’s recently described the performance of this service during November 2008 through October 2010:
“A sure way to beat the market is to pick the best stocks and avoid the duds. …It takes brains, a willingness to do the homework, plus a little bit of luck. One group that has all those attributes is the team behind Barron’s Daily Stock Alert, an electronic newsletter that delivers a stock pick at 6 a.m. every trading day of the year. Over the past two years, the Alert has picked 435 stocks, which rose 28.8% on average, compared with a 16% gain for the overall market. The Alert’s picks are also well ahead of the market for the past year, six months and three months.”
Is this representation dependable? Using the detailed list of recommendations accompanying the self-assessment, we find that:
Reservations regarding the self-assessment are:
The detailed list includes 434 recommendations on 409 separate days during the two years spanning November 2008 through October 2010, all but one of which is a trading day. However, the service description currently states: “With the Daily Stock Alert, you’ll receive an e-mail each day with…[o]ne new, credible stock pick from the editors of Barron’s… Five handpicked stock tips each week. Over 250 tips per year…” There are 502 trading days in the sample period. Are there recommendations omitted from the list, or does the service not deliver as many recommendations as stated? If there are omitted recommendations, what is the performance of the omissions?
Of the 434 recommendations, 70 (16%) are designated as closed or terminated due to acquisition. Does the service systematically issue follow-up alerts advising when to close positions?
In any case, the large majority of recommendations remain open, making portfolio-level performance difficult to assess. For example, an investor wishing to exploit all recommendations must hold a large percentage of the portfolio in reserve (as cash to buy new recommendations). Over the sample period the return on the cash is near zero. A large cash position earning approximately zero return would substantially debit performance over the sample period.
The alert prices in the detailed list are apparently closing prices for the trading day before the alert date. A subscriber may not be able to realize these prices systematically. For example, for the last 100 recommendations in the list (5/25/10 through 10/29/10), alert prices are lower than the daily lows during the alert dates 24 times. Using pre-alert prices to calculate returns on recommendations may inject a positive bias for service performance. However, the positive bias as measured by using alert-day opening prices instead of pre-alert closing prices for the last 100 recommendations is quite small.
Service performance calculation apparently do not include estimates of trading frictions, which depend on specific broker fees, trade sizes and bid-ask spreads. Trading friction can materially dent performance.
While the returns of recommendations at least sometimes include adjustments for dividends, the benchmark S&P 500 return apparently does not include S&P 500 dividends, thereby giving an advantage to the service relative to the benchmark.
Could October 2010 a lucky stop date for service performance? Is performance history available for a longer period and for other stop dates? These questions relate to the volatility of the recommendations compared to that of the broad market.
In summary, investors considering a subscription to Barron’s Daily Stock Alert service may want to seek clarifications regarding the above reservations before subscribing.
For formal research on the stock-picking performance of financial journalists and columnists, see “The Stock Picking Expertise of the Business Media” and “Slim Pickings Among Stock Picks of Columnists?”.
