Isn't that a Monte Carlo fallacy? One outcome in a sequence of tests does not mean that others are more likely in the future. If you flip a coin and get 10 heads in a row, your likelihood of getting a tails on the next flip (assuming a perfectly-balanced coin) is still 50%. Brazil's good year may be another case of climate change for all we know, and could very well persist for awhile; or, you might be right and next year things will balance out again. We don't know either way yet.
That isn't to say that we shouldn't be concerned regardless; you're correct in your observation that having all our figurative, coffee-flavored eggs in one basket - even if that basket is the size of Brazil - is rarely a good idea.
