• Fri
  • Jul 11, 2014
  • Updated: 10:19pm

Don't be impressed by 'survivor bias'

PUBLISHED : Monday, 25 February, 2008, 12:00am
UPDATED : Monday, 25 February, 2008, 12:00am

Like almost everyone who uses e-mail, you probably get plagued by spam. You know the sort of thing: messages offering cheap codeine or Viagra, treatments that claim to enhance various bits of your anatomy, marriage proposals from young ladies in eastern Europe, and e-mails from the widows of Nigerian generals offering you a 20 per cent share of their late husbands' millions if you first seed an offshore account for them with US$10,000.

If you are especially unlucky, alongside these you might get unsolicited e-mails from dubious-sounding finance houses offering hot investment tips. Usually, these seem to recommend stocks in obscure mining companies on the verge of announcing a major gold find, small pharmaceutical start-ups about to discover the cure for cancer, or tiny technology companies developing a 'revolutionary' new product.

If you are sensible, you probably delete these messages without a second glance.

But imagine if you were to receive an e-mail early this Monday morning forecasting that the Hang Seng Index will rise today. If the market duly rises, and you receive another message from the same source tomorrow morning correctly predicting a fall, you might take notice.

If you then get e-mails accurately forecasting a rise on Wednesday, another rise on Thursday and a fall on Friday, you might well be impressed.

And if you then get a message inviting you to pay for further research, or even to send HK$10,000 to open a brokerage account (both would be illegal), you could be tempted.

That would be a huge mistake. Just imagine if today the crooks send out e-mails to 20,000 targets, 10,000 predicting a rise in the market that day and 10,000 a fall. Then on Tuesday, they only send e-mails to the 10,000 targets who got the correct forecast the previous day. Once again half predict a rise and half a fall, and so on through the week.

By the end of trading on Friday, 625 targets would have received a full week's worth of perfect market calls. If all take up the invitation to open an account, the crooks would pocket a cool HK$6.25 million and immediately leave town.

This is a variation on the 'survivor bias' problem that accounts for so many instances of apparently brilliant investment performance. For example, if 1,000 investors decide whether to go long or short for the next year by flipping coins, after 10 years there would be just one who had called the markets correctly every time. He would have made a fortune, and would be hailed as a financial genius. But his performance would have been down to sheer blind luck.

Professional investment advisers tend to respond with fury to suggestions that market moves could be random, or that successful track records could be the result of nothing but chance.

In their defence they point to long-term trends, such as the Hang Seng Index's extended climb from 100 when it was first calculated in 1964 to 23,305 at Friday's close. Such trends, they claim, are evidence that markets are directional over time and that stock price moves can be forecast successfully.

Alas, their protests are not completely convincing. The Hang Seng Index has indeed risen 233 times over since 1964, but that performance is also skewed by survivor bias. That is because the index comprises Hong Kong's biggest listed companies. As successful companies grow, they get included. Companies that do poorly drop out, so the index selects for success.

As a result, the composition of the index has changed a great deal since 1964, even though Hong Kong's small economy and lack of competition law favours incumbent companies in any sector.

Of the original 30 companies in the index, only seven feature under the same or a similar name among the modern Hang Seng's 43 components. Swire Pacific is still there, for example, while China Light and Power has become CLP Holdings. Hongkong Bank is now HSBC.

Some, like Hongkong and Shanghai Hotels, have dropped out of the index but survive as second-tier companies. Most of the original constituents, however, have either been swallowed up or seem to have disappeared without a trace. Few investors can now remember South Sea Textile Manufacturing or China Provident as blue-chip companies.

In the meantime, a host of new stocks have entered the Hang Seng Index, including all the H shares and red chips that now make up so much of its value and have driven it to new heights.

An investor who held the original 30 stocks would have made nothing like the index's gains, however. In fact, by now he would be feeling so desperate at his underperformance he might even be tempted to try and make a quick fortune through a spam e-mail scam. Hopefully, no one would be dumb enough to fall for it.


For unlimited access to:

SCMP.com SCMP Tablet Edition SCMP Mobile Edition 10-year news archive




SCMP.com Account