Two great investment strategies
I often get asked how best to make money on the stock market. In this article I show the two best ways that I know. The most reliable way is derived from Burton Malkiel's perennial classic A Random Walk Down Wall Street. It works in 3 steps:
- Start eight investment newsletters. In four of them, forecast that the stock market will rise. In four of them, forecast that the stock market will fall. At the end of a year of enthusiastic proclamations, four of the newsletters will prove to have picked wrong. Close them down.
- Now four newsletters remain. In two of them, forecast that the stock market will rise. Maintain the opposite prediction in the other two. At the end of the year, close down the two that went wrong.
- Now two newsletters remain. Apply the same policy, leaving one newsletter alive at the end of three years.
At this point, the surviving newsletter will be a rare bird indeed -- it will have forecast the direction of the market correctly for three years running! The writer of this newsletter will be a celebrity; he will appear on television analysing events, and millions of people would hang on to his every word. It would now be time to write a book Twenty easy steps to unimaginable wealth. The circulation and advertising revenues for the newsletter itself should also grow dramatically.
Dilbert cartoon from 22 August 2009:
There is no catch -- this idea really works. This business plan illustrates what economists call the survivorship bias in fund evaluation. The pool of funds or fund houses that survive for extended periods of time tend to be biased towards those that were plain lucky in preceding time--periods; the unlucky funds tend to close down. Owing to this, when naive performance evaluation procedures are applied to all funds in an economy, the statistical procedures are biased towards discovering that many fund managers do outperform.
Remarkably enough, selection bias shows up in myriad situations, going beyond the newsletter scam and the statistical problems of performance measurement.
- Major fund houses in the US, such as Fidelity, run hundreds of products. This pool of products is large enough to ensure that each month, a few of the products look very good when compared with their benchmarks. This allows Fidelity to write a monthly newsletter with a cover story each month Product X by Fidelity is the best in its category. For any one house which has a sufficiently large number of products, it is quite easy to produce such gimmickry.
- A brokerage firm has 1000 customers in a portfolio management
scheme (PMS). A highly volatile stock is chosen, and 500 customers
are asked to buy while 500 are asked to sell. The newsletter scam is
repeated, and at the end of three months, the firm has 125 customers
who are ecstatic at having been pointed to winners for three
consecutive months. These customers can now be talked into almost
anything.
In markets with market makers, this scam would be particularly profitable if the firm in question was the market maker for the three stocks chosen. In this case, the bid--ask spread over all the trading volume generated in the implementation of this scam would additionally accrue to the firm.
- In any country where 100 schemes exist, five funds will (on
average) seem to outperform at a 95% level of significance. These
error rates are intrinsic to statistical testing. The very meaning
of a 99% level of significance is that 1% of all funds tested will
report excess returns even if all funds are truly mediocre.
In this sense, now that we have dozens of funds in India, it would not be surprising to find a few exhibiting positive alphas, as compared with the situation of previous years where the population of schemes that existed was smaller.
Survivorship bias is hence a pervasive problem in fund evaluation. If a fund manager actually adds value, it takes a multi--year track record and clarity in performance evaluation to correctly detect it. In a population of mediocre fund managers, it is easy for performance evaluation to incorrectly highlight a few `superstars'.
Competition between funds is often considered a powerful device for improving fund quality. Competition will undoubtedly help in weeding out gross failures in the fund industry. However, the ability of competition to push product quality is limited by the inability of investors to observe and measure fund performance. Mistakes in performance assessement will lead investors to be unable to discriminate good managers from poor managers. This will cause money to flow into mediocre funds (which happened to fare well by chance) or to flow out of good funds (which happened to fare poorly by chance).
Debates about reforms to the insurance and pension sectors always emphasise a competitive structure of asset management companies which plan sponsors should be able to select between. This scenario will generate poor outcomes if plan sponsors choose to back poor managers. If plan sponsors do use fund managers who claim to do active management, they will need to be extremely knowledgeable about using performance evaluation procedures wisely.
What does this imply for fund managers who wish to convey their own special qualities to investors? Performance evaluation procedures are the only answer. Good fund managers should invest in refining ideas in performance evaluation, and in communicating these ideas to investors.
What does this imply for investors? Investors can try to be sophisticated, and understand performance evaluation results and use them with care. Alternatively, investors can just stick to index funds. It does not take a great deal of cleverness to harness the advantages of equity investment -- it just takes an index fund. Over twenty--year periods, in the US, there is not a single fund which has beaten the S&P 500 index fund in risk--adjusted performance evaluation.
Index funds pay low management fees, avoid wasting money on transacting, and do not require clever performance evaluation. A person who buys Rs.3,000 of the NSE-50 index each month from age 25 till age 60 is likely to end with Rs.14 crore on average in 1998 rupees (he will beat Rs.1.4 crore with a 95% probability). A rickshaw driver who invests Rs.10 each trading day into Nifty from age 20 to age 60 is likely to end with Rs.2.4 crore on average in 1998 rupees (he will beat Rs.20 lakh with 95% probability). Index funds harness the gains from equity investment, without incurring the problems of active management.
The investment strategy of regularly buying equities, regardless of ups and downs of the market, implemented via index funds, is the second best investment strategy known in economics. It probably fares even better than newsletter scams when long time--periods are considered. It does not require understanding the economy, a "feel of the market", cleverness in performance evaluation, successful speculation, or insider information. It just requires steadily buying Nifty for decades on end.
Back up to Ajay Shah's media page