The latest Global Investment Returns Yearbook, produced by ABN/Amro and London Business School, which appeared this week, provides its usual fascinating wealth of detail about the performance of the world's financial markets in the year gone by, and with it a further entry into their painstakingly developed series of long-run investment returns from 1900 onwards. It should be a starting point for any serious student of the "big picture" in investment.

The theme that runs through this year's edition is that 2004 was something of an oddity in financial market history. The global economy grew by a handsome 5 per cent in real terms, confounding virtually all the forecasters, yet the world bond markets, far from retreating at the prospect, as you would normally expect and most investors predicted, ended the year even more favourably priced than before, with yields falling slightly over the year (in the UK, by 0.25 per cent).

Stock markets in general also had a reasonable year, one that produced returns that were a bit above the long-run average, though most of the performance was sandwiched into the last few weeks of the year, following the re-election of President Bush. The year was characterised by low volatility and the absence of any clear or powerful trends, other than the weakness of the dollar. Whereas in 2003, the year which saw the market finally reverse direction after three years of bear market, the place to be was in risky assets (emerging markets, high beta stocks, junk bonds, the riskier the better), last year there was no such simple overarching theme.

True, it was another year when you wanted to tilt your holdings of equities away from large cap stocks, which in the UK returned 13 per cent, against 21 per cent for smaller cap shares as a group. Normally, size is a simple and dominant factor in determining investors' performance, in the sense that its effect gets more marked the further up or down the capitalisation scale you go. Thus, in the 1995-2000 bull market, the largest capitalisation stocks did best, followed by mid-cap stocks, then smaller cap shares and finally the minnows of the market; and in the bear market of 2000-2003 the effect was reversed, with the smallest doing best and the largest the worst.

Last year presented a rather more confusing picture, with mid-cap stocks doing well, and the minnows best of all, while the other smaller capitalisation sectors lagged a bit behind. If you look at beta, the academics' measure of a stock's sensitivity to overall market movements, the results were also somewhat perverse. Surprisingly, the lowest-risk stocks on this measure (often known as "defensives") did best of all, and the highest-risk stocks did least well, the reverse of what you might have expected in the second year of a new bull market.

Turning to dividend yield, 2004 saw the continuation of a trend that had surprised the authors of the study the year before as well. Value stocks, defined as those with the highest dividend yields, topped the performance tables, while, more oddly, those in the middle of the market on this measure did worst.

This shapeless picture is consistent with the view that the market has been polarised between different groups of investors, all looking for the emergence of a clear-cut trend they can follow, but none emerging as dominant. The lack of "momentum" in the markets also helps explain why hedge funds, most of which operate on trend-following strategies, had a relatively poor year in 2004.

Standing back from one-year data and looking at broader trends continues to underline features of the current environment. One is that the seismic shift in 2000, away from "growth" to "value" dividend-paying stocks, has yet to lose its force.

Anyone who had followed a simple strategy of buying only the 20 per cent of highest-yielding stocks in the UK market since the bull market peaked in March 2000 would have achieved a five-year return of 57 per cent, in a period when the market did nothing and the lowest-yielding stocks lost 53 per cent; they could rightly ask: "What bear market?" A value investing strategy has also proved historically the most profitable of all over the entire period covered by the ABN/Amro study (1900-2005). There is a strong correlation between the level of real dividend growth in stocks and their long-run total return. The downside is that nothing endures forever. There will be periods of up to five years or so when value investing fails to work - and most investors do not have the patience to ride out those years when other investors are doing so much better.

Another point that comes out clearly from the 2004 data is that while investing appears simple with hindsight, not everyone finds it so in practice. The range of returns achieved by professionally managed UK equity funds in 2004, notes the study, was strikingly wide for a second year running. The average professionally managed fund underperformed the market as a whole in 2004, this time by about 0.5 per cent, but the range of returns last year extended from 36 per cent to minus 3 per cent - which, for a year when volatility was so low, is again something of a surprise.

Given that the UK stock market is so lop-sided at the top end, with just four sectors (banks, oils, pharmaceuticals and telecoms) accounting for 50 per cent of its value, if you got the sectors wrong (pharmaceuticals and telecoms bad, banks and oils good), you had little chance of beating the market unless you were in smaller stocks. In the past two to three years, stockpicking has been the key to doing well - which is encouraging, in theory, for active investors - but also a potential source of disappointment, as doing it well for a sustained period has proved notably difficult.