There are two nice posts up by Ben Strubel at New Economic Perspectives. He addresses commodity speculation. There’s currently quite a bit of discussion about this in Germany because food speculation, for instance, drives up prices and adds volatility.
Strubel is not coming from this direction though – instead he looks at how commodity speculation is sold and shows that the sales pitch is wrong, that commodities by now track other speculation objects like stocks rather closely, and that the returns are subpar. So this sounds like a perfect lose-lose (but then, so do a lot of speculative endeavours to me). This was in the first post.
In reaction to this he’s been attacked by someone who basically claims that he has no idea, that commodities outperform other speculation objects etc. Apart from the fact that Strubel proposes in his reply a kind of trader’s Randi challenge: Strubel invests in non-commodities of his choice, his critic invests in commodities, winner after ten years gets 10k from the other one, he also offers this:
The other question is how will an index perform once there are a large amount of assets tracking that index? History is littered with investing strategies that appeared to work on paper, backtested against data sets, or tracked as a model. Only when actual investments were made have they found out that the effect of making the prescribed trades or investments has erased some or all of the predicted returns. When discussing returns, I prefer real, tradable, investable products rather than paper indexes when at all possible.
This is basically my standard response/question every time I see a Machine Learning/Data Mining paper that claims they can beat the market – so I am glad to see that I am not just paranoid.