Yikes thats a lot of responses. Nothing super top secret at play, stop worrying. Growing up my older brother would greet my public displays of analyticity by jokingly staring my way, sighing, and saying "God you're annoying". With lucid speculation I am comfortable annoying a wider group of people, when it is more "brainstormy" I prefer to narrow the range. It is just a kind of politeness and consideration (or abject cruelty if you happen to be finbar99).
I am trying to find a way to analyze the data about shares sold by Van Eck Global to try to determine the way they are sold, redemption or discretion. At first, it seemed impossible without knowledge of the rate of redemption in the fund. Without a source it can't be known directly, and due to complicating factors such as the rate of creation and discretionary rebalancing, it can't be known inferentially without those other facts. So I am trying to find another way, redemption involves unbundling creation units of 50,000 shares each which might result in identifiable patterns. Suppose we take the NAV of the fund and multiply by 50,000 to get the value of a bundle for a given day and if we multiply that by the proportion of GBGLF in the fund to get the proportion of value GBGLF contributes, we could then divide that by the NAV contributing value for GBGLF for that day which is the closing price. The result should be approximately the number of GBGLF shares in a creation unit for a given day. And with that one could determine if the total amount of shares distributed was neatly divisible by the per creation unit amount. Thereby, suggesting unbundling. That might work - I don't know. The biggest problem with trying that line of reasoning is that the proportion data is clearly inadequate - ".02" for GDXJ and ".00" isn't enough information. I need maybe six spaces behind the decimal and adjusting microsoft excel doesn't help - it just adds zeros. If we look at the distribution for 2/15/13 when it was low, there might have been one redemption that day and the number of shares distributed by GDXJ might be the number of shares per bundle for that day. If we take the GDXJ NAV for that day of 17.32466 times 50,000 times .0002126564 (as an inferred proportion of .02126564) we get 184.2099 and if you divide that by .016 the closing price and therefore NAV contributing value for GBGLF, we get 11,513, the number of shares distributed from GDXJ that day. And so, if we had appropriately specific and fluctuating proportion numbers we might be able to confirm unbundling.
That actually came out a little more lucid than I thought it would. It is OK to say "God you're annoying" I can't hear you.