Last month we discussed a number of share selection techniques
(i.e. buying and owning "smaller" or
"neglected" shares, "under-valued" shares,
shares with the highest price "strength", etc.) that
have each been shown to outperform the market average.
Unfortunately, combining several of these techniques into a comprehensive share selection criteria is not a simple task.
Furthermore, once you have decided upon using one or more of the share selection techniques, it is also necessary to develop a "Portfolio Management Strategy" - turning your analysis into actual decisions to "buy", "hold" and "sell" particular shares.
There are several "problems" in attempting to
combine various share selection methods into a comprehensive
share selection criteria.
One major difficulty is that criteria can be correlated. Another is that they can be uncorrelated.
Examples of the former (i.e. "correlation") would include the "small company effect", sharebroker "neglect" and institutional "neglect". "Smaller" company shares (as a group, over the medium to long term) outperform the market, shares "neglected" by brokers outperform the market and shares "neglected" by institutions (i.e. with low levels of institutional ownership) outperform the market. However, combining these criteria (i.e. "smaller" companies, "neglected" by brokers and by institutions) does not yield higher investment returns.
The reason? Companies that qualify under one of these criteria will often qualify under the other two. So each of these three criteria will select a very similar group of companies - and combining similar criteria adds little to the value of these share selection methods.
Similarly, shares with low Price/Sales ratios tend to outperform the market, as do shares with low Price/Earnings ratios, high Dividend Yields or a low Share Price to Net Asset Backing. Again, however, combining several or all of these criteria will make only a small improvement in these selection methods. All of these statistics measure "under-valuation" and a share that is "under-valued" by one criteria will likely be "under-valued" by most of the others. Combining several "valuation" statistics therefore adds little additional information.
An example of the "problem" of uncorrelated criteria would be high "relative price strength" and criteria for finding "under-valued" shares (i.e. low Price/Sales ratios, high Dividend Yield). For a share to have high relative price strength it must have risen strongly over the last 6-12 months (and that "strength" has a tendency to continue into the future). However, having risen strongly, such shares are never the most "under-valued" on the market.
You cannot, therefore, buy shares that rate in the "top 10%" by price "strength" and the "top 10%" by "under-valuation". Few - if any - shares would ever meet both criteria. The rising price necessary to qualify under the first criteria will remove the extreme of "under-valuation" necessary for the second criteria.
Nevertheless, these two techniques can be profitably combined - by reversing one of the criteria! For example, a very successful combined criteria would be to buy the "strongest" shares with a Price/Earnings ratio under 20 and a Price/Sales ratio under 1.00 (i.e. the "strongest" shares, excluding those that are already too "over-valued").
Another very successful way to combine these two criteria would be to buy the shares with the lowest Price/Sales ratio but with a positive strength rating (i.e. the most "under-valued" shares, excluding "weak" shares that are declining in price).
"Insider" trading (i.e. buying and selling by directors and senior management) can be a very reliable indicator of future share price performance - but significant transactions can be rare. (Note: Directors' transactions are not even reported in NZ - but are in Australia.) One director buying or selling $10,000 worth of shares is not enough to justify a decision for an investor to buy or sell shares in the company. Several directors each investing a few hundred thousand dollars would be very significant - but this may happen only rarely.
So "insider" trading - by itself - does not offer enough investment opportunities to develop a useful share selection method (i.e. it will not offer enough investment opportunities to be able to maintain a properly diversified portfolio). However, "insider" trading information may yield the occasional investment opportunity and can be valuable in choosing between shares that look relatively equal under other selection criteria.
In addition, there is little or no research or information on the relative importance of the various selection criteria that we have been discussing. A "neglected" share may be more attractive than a "moderately followed" share, and a share trading on a P/S ratio of 0.25 is more attractive than one on a P/S ratio of 0.80. But is a "neglected" share on a P/S ratio of 0.80 more attractive or less attractive than a "moderately followed" share on a P/S ratio of 0.25? We don't know - and to find out would require a major research project following hundreds or thousands of individual shares over three or four decades. Unfortunately, that historical data just isn't available.
Finally, all of the share selection methods we discussed last month can be evaluated objectively. That is, you can measure the criteria with a number and rank shares from most attractive to least attractive - and those criteria can be accurately measured and duplicated in the future or by other investors.
Many factors (e.g. the future "growth" potential of a company or an industry, how that growth will be financed, a company's future cash flow and dividend policy and any "competitive advantage" over current - and future - competitors) do need some consideration, but can only be evaluated subjectively by an investor or analyst. In other words, there are some companies that we may choose to avoid even if they scored well on the objective criteria - and similarly some companies that we would tend to favour.
Combining different share selection methods requires
considerable subjective decision making. In the absence of
empirical research studies, one must subjectively decide upon the
"weighting" of the various techniques (i.e. the
relative importance of each selection method), the formation of
various indicators (e.g. should an "insider" statistic
measure the net number of buyers and sellers over the last six or
the last twelve months, and should those transactions be
"weighted" to reflect the dollar value of buys and
sells?), and set a "level" at which a share becomes
attractive enough to qualify as a "buy" (and
unattractive enough to warrant a "sell").
However, sharemarket studies have demonstrated two useful facts that can probably be used as guidelines for combining any share selection methods:
1. Most indicators "work" across their full range - differentiating between the most attractive shares through to the least attractive shares. So in a combined share selection criteria, individual methods can be used to include the most attractive shares (i.e. the "strongest" shares, "under-valued" shares, "neglected" shares, shares being bought by "insiders") or to exclude the least attractive shares (i.e. the "weakest" shares, "over-valued" shares, "widely-followed" shares, shares that "insiders" are selling).
2. The best combinations consist of unrelated selection methods. For example, combining "fundamental" methods (i.e. based upon valuation) with "technical" methods (i.e. relative price strength) offers significantly higher returns and lower risks than using just one of these selection techniques.
A comprehensive share selection criteria should therefore favour "smaller" companies and/or "neglected" shares, which are "undervalued" (the Price/Sales ratio has, surprisingly, proven to be the most reliable indicator, followed by the Price/Earnings ratio and then the Dividend Yield) and whose prices are in uptrends (i.e. with high relative price "strength").
Similarly it should generally avoid the very largest companies which are widely followed by sharebrokers, trade at high valuations and where the share price is declining.
Formulating "Buy", "Hold" and "Sell" Criteria
It is easy to formulate a "buy" criteria for any share selection method (i.e. buy the "strongest" shares with P/S ratios of less than 1.00), but to manage a "real money" portfolio in "real time" it is just as important to formulate a "hold" criteria and a "sell" criteria.
Some "tests" of share selection methods assume that a portfolio of perhaps ten or fifty of the most attractive shares are purchased on January 1st of each year. The following year - on January 1st - those shares are sold and replaced with the current most attractive.
This theoretical method involves several real life problems. Firstly, the only reason for this "once per year" review is that the portfolio does not have a "sell" criteria and so more frequent reviews could generate excessive trading and brokerage costs. A major problem with a "once per year" review is that it cannot exploit information in a timely fashion. If several directors suddenly started selling large quantities of shares in February then a review of whether or not that share should be sold and replaced will not be made for another eleven months.
In the real world, a "hold" criteria and a "sell" criteria are just as important in a share selection method as the "buy" criteria.
For example, in the original work on "relative strength", each week Robert A Levy ranked shares from "strongest" to "weakest" based upon their return over the previous 26 weeks. A simulated portfolio using a "buy" criteria that a share be in the top 5%, and held until it fell out of the top 70% (i.e. the "sell" criteria), yielded returns 2½ times greater than the stockmarket average.
To achieve our goal of formulating a comprehensive Share Selection Criteria we shall next month review some of the important research supporting each share selection method. These studies will form the basis for our subjective selection and "weighting" of the various techniques - and to determine appropriate "buy" and "sell" rules for Portfolio Management.