Warren Buffett has never set much store by market timing. He reckoned that he drank his first Coke in 1935 or ’36 but it was not until 1988 that he took a few sips of the stock, 14,172,500 of them to be precise, for $592.5 million. It was a fine investment; over the next eight years the holding returned 961% against the 221% of the S&P 500.

I’m indebted to the excellent Jim Grant for this little gem of Buffettology, who also reminds his readers that on October 11 1996, he said sell Coca-Cola. It wasn’t immediately a great call, since the shares carried on dancing up for another year, but on the timescale that Buffett would appreciate, it was right on the money. Despite a near-trebling of sales on maintained margins, the Coke share price today is only 30% higher than it was 16 years ago (and lower than its peak in 1998). It’s underperformed the price of its key ingredient, sugar, and compared to Grant’s long-term favourite investment (and Buffett’s least favourite) it’s been a dog. That investment is, of course, gold.

This little story is a demonstration of the limits of the buy-and-hold investment philosopy. As Buffett keeps reminding us, gold produces nothing, and is only worth what the next buyer is prepared to pay for it. Coca-Cola is the world’s leading consumer brand, glugged by couch potatoes and loved by dentists everywhere. It’s also, subject to the usual rubric about the past being no guide to the future, almost unfashionably cheap. From a P/E of 57 in 1998, the shares are now valued at just 18 times earnings. It yields nearly 3%. It may get cheaper, but it’s surely a better long-term investment than the 10-year Treasury note which pays 2.3% and your money back in 2022. It may even be as good as gold. Just don’t expect Grant to put it quite like that.