GLD Added 7½ Tonnes in January 2013

A quick follow-up to the GLD bar list, which had a large 600-bar addition last week, I wanted to shed some light on the inner mechanics. Inventory additions in GLD are noteworthy because the trend is currently inventory removal. This is a followup from this post when 6½ tonnes got added late last year.

The inventory addition on 17th January 2014 from the SPDR GLD trade spreadsheet finally showed up in the bars list data on 23rd January 2014 (delayed because of the US holiday). I was mildly surprised to see that all the bars added (i.e. 100%), had been seen before, i.e. they were dark bullion for a time before being 'added' back to the ledger. The average rate for dark bullion discovery is typically 30% of added gold so this is unusual. For most of the time, these bars have been sitting in the vault, from as far back as 2009, this is the breakdown by refiner:

Below is a method I am developing for GLD inventory visualization, sorry for the crudity - bear with me, I will explain it.

What you're looking at here is an approximate depiction of the history of these 600 bars. Vertically = one pixel per bar, horizontally = date, one pixel for each of the 617 GLD bar list we have on record (yes, forms a square for this sample size). All these bars show as gold in the last vertical pixel, which was the bar list from the 23rd Jan (click to enlarge).
  • The grey strip along the middle (19 bars) means that this block was not introduced to the inventory until recently (approx. August 2013).
  • The big black block in the middle right (61 bars) means that these originally went missing last year - approximately July 2013. This is different to the pallets around it which only disappeared recently.
The ordering has been normalized to be in blocks of RefinerName, then sorted by BarSerialNumber.

Some of the banding is obviously similar, and these will be bars which 'travel together' (on the basis that some leave and appear at the same time). Finally, if we arrange the bars in the physical order as they appeared on the January 23rd document, we see pallet sizes emerge (combinations of 40 and 80).

In conclusion, although we have a better view of what the bars are doing, we still don't have an explanation of why this is occurring in the vault. Here are my observations:
  1. These patterns would be difficult to fake using random data. If your belief system still maintains that the activity in GLD vault inventory is fraudulent, that idea gets stretched even more as we continue to tease out these patterns.
  2. Most of the added gold is from stuff recently removed, indicating it is nearby (i.e. not out of the vault with a range of a few weeks). Some of the added gold came from July 2013 so there's a question mark over the other gold removed since July 2013, i.e. how much else is there, in the vault? This last 1½ pallets was not evenly distributed (i.e. fragmented), does that mean anything specifically?
Please feel free to ask questions or add comment.

Best regards,
Warren James
Screwtape Files


Bron Suchecki said...

I like the visualisation by pallet, it would be interesting to see how these pallet groups hold together over time.

I guess this visualisation works best at detail rather than trying to show dark behaviour for a large amount of bars.

The GLD doesn't have gold meme I think is dead not unfortunately because of your work but because the pumpers get more value from the GLD metal going East meme now. It would be fun to find examples of people who were stridently of the view that GLD had no gold but now push the GLD redemption meme and get them to acknowledge that they were wrong, but I couldn't be bothered, they just aren't worth the effort.

GM Jenkins said...

Great stuff, Warren - I like the visualization approach you've got going.

"The GLD doesn't have gold meme I think is dead not unfortunately because of your work but because the pumpers get more value from the GLD metal going East meme now. "

Damn good point, Bron. What a bunch of clowns some of these guys are. For all their moralistic fervor -- no intellectual integrity whatsoever.

Gary said...

I note that as the GLD inventory bottoms out, we only have 2 comments on this thread to date.
Remember the good old days when dozens would gleefully claim GLD losing gold was some sort of sign.
Time will tell, but the coat check meme could soon be added to the dustbin of wishful thinking.

Warren James said...

@Gary, yes for sure - mine is a niche topic in a niche industry, plus no one wants to read a theory that runs contrary.

Binning the coat check idea is as difficult as it is purporting it. If there's anything I know from all my study; the vault workings are very much a mystery, even with these deeper glimpses from the data.

The numbers still have many more secrets I am trying to wring out, it just takes some juicing effort to get it done. For anyone keeping track, some of the recent adds in GLD Weight-List have yielded the following results:

2014-Feb-03 : 167 bars added, 25 were new signatures, 142 were Dark Bullion Returning (85%).

2014-Feb-04 : 48 bars added, 8 were new signatures, 40 were Dark Bullion Returning (83%).

I'm not sure why the sets are mixed, it seems like 'gold left over' from something or the new bars could be co-mingled with ETF Securities bars. I'll be doing more analysis on this for my next article, which is the #3 installment on the GLD series. .. takes a while to write these ;p