HOLD that Thought


New York – One of the critical judgments that need to be made in evaluating the recommendation accuracy of a research provider (RP) is how to deal with the hold portfolio. Some performance measurement systems assess the overall return by evaluating the buys and sells only. Others penalize the firm for having a hold portfolio that performs better or worse than expected, since this implies that the research firm should have classified some of those holds as buys or sells.

Another way to handle the hold category is to handicap its performance by an arbitrary weight structure. For example, if a company has a five point rating system, the strong buys and sells could have a weight greater than 1, the buys and sells could have a weight of 1 and the holds could have a weight of .5. In his way, the hold portfolio can be given a less important role in the overall performance commensurate with the level of conviction assigned to that classification.

By definition, holds are supposed to be those stocks that have been purchased and should not be sold. As such, they are intended to appreciate over time. In this sense the hold is akin to the “market perform” rating that some analysts assign to stocks. Here the hold is a non-actionable rating.  Buy or sell ratings, however, are actionable events, where a transaction takes place based on the recommendation. In some rating systems a buy is replaced by an “accumulate” rating indicating that the stock should perhaps be purchased in increments over time. In others, it is replaced by an “overweight” rating, reflecting the fact that the stock should have a higher weight in the investor’s portfolio.

While definitionally a hold is a passive long position, practically a hold is just a box RPs use to put everything they cover, for which they have no strong views. As such, we feel that the best way to assess the performance of buy, sell and hold recommendations is to use an artificial weighting scheme, such as the one described above, to generate a performance index for each RP.

One problem with this approach is the fact that holds have a different significance in different rating systems.  In some quantitative systems, for example, holds are sorted into deciles and the top several deciles are labeled buys, the bottom several are labeled sells and the middle band is labeled holds. When the market is in a strong rally, all of the stocks under coverage (including the sells) may be appreciating, while in a bear market all of the covered stocks may be falling. In a more qualitative system, holds are not expected to depreciate significantly under any circumstances.

Another problem with the artificial weighting scheme is that it does not indicate how much, in dollar terms, one would have made if the investor had followed all of its recommendations.  Nonetheless, when it comes to measuring the overall stock selection and sorting ability of a research provider, this system gives the most accurate picture of that provider’s performance.

Comment by Scott J. Drysdale:
The Hold rating is a cop out. In reality all stocks to a PM are either buys or sells. A stock “held” in the portfolio is simply a tansaction-cost-free buy; the alternative being a sale. Tell a client you held their stock from $20 to $10 and see if they don’t think you bought it.

The proper evaluation is to score the Hold rated stocks against the relevant benchmarks and absolute returns. For example, if a Hold rated stock outperforms the S&P 500 by 200 b.p. then it should count against the analyst. He missed a buy. Similarly a hold stock that underperforms should also be held against the analyst. He missed a short.

It is a new world and this is increasingly a big boy’s game. If an anlyst can’t add value, why pay’em?


About Author

Leave A Reply