Why analysts can get it wrong

Another possibility is that analysts all sign up to the same view for reasons of self-preservation. For them, there is safety in numbers. While they could go against the crowd, the kudos they'd get if they were proved right is outweighed by the fear of looking silly by getting it wrong when everyone else was right. If all analysts get it wrong together, they can defend themselves more easily from the criticism of their employers and their clients.

The UK evidence

Inspired by the US research, we've built a portfolio of UK equities every year since 2004. We scour the largest FTSE 350 companies in a bid to find the shares that the analyst community loves and hates the most. We then suggest buying the most unpopular shares and selling the most adored ones. Over time, this approach has worked a treat, outperforming the rest of the stock market three years running from 2004.

The most recent portfolio assembled in December 2007 did not match its predecessors, however. All 10 shares (see table 'Our last selection' below) fell, albeit against the backdrop of a savage bear market. The average loss was 42 per cent. But this was worse than UK shares in general – the average underperformance of each share was 7.9 per cent. So, what went wrong?

As it happens, we volunteered some reasons why the strategy might not do as well at the time when we originally wrote the article. Seven of the 10 shares were members of the FTSE 250 index. We argued that a UK economic slowdown was likely to hurt mid-cap shares such as these much more than FTSE 100 shares. In the event we were correct, with the FTSE 100 underperforming the FTSE 250 by a thumping 15 per cent during the period.

We also said that a lack of takeover activity might count against our portfolio. In prior years, a couple of portfolio members were bought out, boosting the portfolio's overall performance. Admittedly, one of the class of 2007 did get taken over. But Banco Santander's acquisition of Alliance & Leicester was sealed during the blackest days of the credit crunch, and at a price far below the one the shares traded at when they entered our portfolio.

Still, it wasn't all bad news. Our collection of analysts' most hated shares did better than our collection of analysts' most adored shares, with the dogs beating the darlings by 3.1 per cent. In addition, the dogs did much better on average than the sectors to which they belong. On average, they were up 10 per cent compared with their industry groupings. By contrast, the darlings underperformed their direct peers by an average of 15.5 per cent.

The fact that analysts' favourite shares generally did worse than the sectors to which they belong is significant. Analysts are assigned to follow specific sectors and to identify their best and worst members. While their record may be better for other shares that didn't make it into our survey, their achievements here suggest that they have huge room for improvement.

Strongest sells in December 2007

Strongest buys in December 2007

Our selection (from the above two tables)