In recent years, this has been a great forecaster of annual changes in the All-Share index. Since January 1996, variations in the 12-month running total of foreign buying of US equities have explained 45 per cent of the variation in the All-Share returns over the following 12 months. The dividend yield, by contrast, can explain only 25 per cent. And if we combine foreign buying with the dividend yield, it's possible to account for half the variance in annual returns.
For example, big foreign buying of US equities in 2000 and 2007 led to the All-Share index falling in the following 12 months, while low buying in 2003 and 2008-09 led to good returns in the next 12 months.
The only occasion when this indicator has led us seriously astray was in 2002-03, when the market fell, even though foreign buying was predicting a rise. (The All-Share index did fall by more than foreign buying predicted in 2008, but because an investor who had heeded the message of the indicator in 2007 has gotten out of the market already, this wasn't an expensive error.)
There's a simple reason why foreign buying predicts returns. It's a measure of global investors' sentiment. Ordinarily, investors are prone to the home bias: we prefer to buy local shares because these fell more familiar to us. It's only when we are exceptionally confident that we'll venture into overseas markets. Such high confidence, however, is often a sign of excessive exuberance and hence a precursor to falling share prices. Conversely, low foreign buying is a sign of weak sentiment. This often leads to share prices rising simply because - from a low base - sentiment is likely to improve in the following months.
And herein lies some good news. Foreign buying has recently been quite low. In the 12 months to September (the latest month for which we have data), foreigners bought just $24.9bn of US equities - much less than the post-1996 average of $83.8bn. This is a sign that they lack confidence, which augurs well for 2014.
If post-1996 relationships between the dividend yield, foreign buying and subsequent returns continue to hold, this points to UK share prices rising by 13 per cent between now and September 2014, with a less than one in six chance of them falling.
Of course, there's no guarantee that past relationships will continue to hold; remember the story of Bertrand Russell's chicken. Goodhart's law reminds us that statistical regularities tend to break down when relied upon. Nevertheless, foreign buying of US shares is the least bad predictor of returns I know. And - for now at least - it is sending a mildly bullish message.
