In discussing the different commodity indices, it is mentioned for several indices that they tend to “perform better when markets are in contango…because the average maturity is longer”. This is easy enough to memorize but I was still wondering why this is the case. Suppose that the commodity curve is upward sloping, i.e. in contango. It is difficult to see what’s the difference if you hold short or long maturities - your roll return is still negative. Without making any additional assumptions about the shape of the curve, I don’t get it how you can say that longer contracts are better than short ones in this setup. Another puzzling this is the comment that “indices that roll couple of days earlier” are in advantage over those that roll later. I can understand that there will be selling pressure when indices are rolling over. Are markets so thin in commodities that if index A sells contracts on Monday and index B sells them on Wednesday, everyone rushes to buy on Monday and on Wednesday there are less buyers? Does not sound like a rational behaviour from buyers’ point of view.
If we look at the contango situation, could it be that the roll return is less negative for longer maturity contracts. The rate of change of prices is more for shorter maturities as compared to longer maturities. One could draw an analogy from a government securities curve. At shorter maturities the curve is ‘more upward sloping’ than at longer maturities. By this logic indices which would do well in backwardated circumstances would be those with shorter maturity (more positive roll yield) as compared to longer maturity contracts (less positive roll yield) If you read p10 of the pdf link below (apparently the original source of the book article) it says that ‘MLCX is basically an average maturity bet) http://www.bache.com/media/managed/ComparingCommodityIndicesMultipleApproachestoReturn.pdf Hope this makes some sense
Thanks a lot for your reply and the link! Indeed, if the curve is concave, the change is bigger in the short end. Also, my second question is answered (p.11) - the earlier roll window is beneficial in contango.