A person manipulates gold. Photographer: SeongJoon Cho/Bloomberg
A person manipulates gold. Photographer: SeongJoon Cho/Bloomberg

If you were writing a paranoid fantasy of gold price manipulation you'd be hard pressed to come up with something more on the nose than the U.K. Financial Conduct Authority's order against Barclays. It has everything; it is the benchmark of manipulation by which all future manipulation will be measured. Well, this or Libor. Delightfully, this manipulation occurred on June 28, 2012, the day after Barclays was fined 290 million pounds for manipulating Libor. They just really wanted to perfect their manipulating technique.1

And oh did they! For starters, they manipulated the most manipulable thing imaginable: Digital options. Former Barclays precious-metals exotics trader Daniel Plunkett, who also settled with the FCA (for a £95,600 fine and an industry ban), sold a customer a digital option that would pay $3.9 million if the 3 p.m. London gold fix on June 28, 2012, was above $1,558.96, and zero if it was at or below that barrier.2 So right there the temptation is obvious: If you're around the barrier, you can save $3.9 million by pushing prices down just a little bit.3 As it happens, the gold price on June 28 was just above the barrier, so a bit of selling by Plunkett could move his client out of the money and make him a bunch of money.

In normal liquid markets, doing a bunch of selling to push the price down is an uncertain thing -- there might be buyers! -- but this was not a normal market. This was the London gold fixing process, which everyone thinks is always manipulated. And for good reason: It's five banks, getting on the phone, talking about what the price will be, and adjusting their trading based on that information. And the description of what went on during the June 28 call will not improve your view of the process.

Basically the way the fixing works is that the chairman of the little five-bank committee announces a price, and every bank announces its buying or selling interest, and the chairman moves the price up or down until supply and demand balance. But as he does this, the banks can keep trading and changing their interest, so they pretty much informally chat it out until everyone is satisfied. Here's a summary of how that went on the day in question:4

  • The fix opened at 3:00 p.m. with a price of $1,562.
  • That quickly dropped to $1,556, then $1,555, due to Comex gold futures selling unrelated to Barclays.
  • Then it rose to $1,558.50, just below Plunkett's $1,558.96 barrier.
  • So at 3:06 Plunkett, who knew this, put in a big sell order, making Barclays a seller of 130 bars of gold at $1,558.50 -- and making it less likely that the price would rise.
  • After everyone had declared their interest at $1,558.50, sell orders exceeded buys by 190 bars, meaning that the call would continue until there was a balance (or no more than 50 bars of imbalance).
  • So at 3:07 Plunkett, who knew this too, withdrew his sell order, reducing the imbalance to 60 bars -- making it more likely that the call would end quickly and give him the fix he wanted.
  • This almost worked, but then Comex futures ticked up and more buying interest came in, leading to an imbalance in favor of buying (and likely higher prices).
  • So at 3:09 Plunkett, who again knew this, put back a 150 bar sell order.
  • This worked, supply and demand balanced at $1,558.50, and the price was fixed there at 3:10 p.m.

This left Plunkett having saved Barclays $3.9 million, at the cost of selling short 150 bars (60,000 ounces) ($93.5 million) of gold. That is not self-evidently a good trade -- if prices rose by more than about 4 percent before he could cover, Plunkett would have a loss5 -- but it's a pretty good idea. As it happens, Plunkett covered his position almost immediately -- by trading with Barclays' gold spot book -- at a loss of $114,000.6

If you do the math,7 that works out to a price of $1,560.40 per ounce, well above the barrier that Plunkett had just avoided.

Plunkett's customer, who despite buying digital options on gold was no idiot, had some suspicions about a 3:10 p.m. fixing that was $3.50 below the 3:00 p.m. starting price, $1.90 below the price at which Plunkett covered,8 and just 46 cents below the barrier. So the customer called to complain, and Plunkett "provided an explanation that referred only to the significant selling in August COMEX Gold Futures" and somehow omitted his own selling. Then he had a change of heart: "After the weekend, on the morning of Monday 2 July 2012, Mr Plunkett sought out his line manager and informed him that he had traded during the 28 June 2012 Gold Fixing." And so eventually this was unraveled, Plunkett was fired, the customer was reimbursed, the FCA got involved, and Barclays ended up being fined £26 million today.

So, what did we learn? One, come on, this process is ridiculous. A big problem in Libor was that the cash traders who submitted Libor for banks were talking too much to the derivatives traders who wanted Libor manipulated for their own purposes. Much the same happened here -- Plunkett emailed his commodities colleagues that he was hoping for "a mini puke to 1558 for fixing," and that the "ideal" fix would be $1,558.759 -- but the problem here is almost worse. Plunkett, an exotic options trader, pretty much sat at the table for the spot gold fix: He knew the supply and demand at each level, and could put in his own orders at the fix, so he could manipulate it himself rather than having to persuade anyone to help him out.

On the other hand, the problem is quite a bit less bad in that the gold fix is a market-clearing two-way auction, sort of, rather than just a made-up number like Libor. So it can't be manipulated costlessly and artificially like Libor; you gotta actually trade -- and take risk -- to move prices. So one thing about this case is that you shouldn't conclude too much from it. As the FCA says:

In the event involving Mr Plunkett on 28 June 2012, the difference between the price of gold in the Gold Fixing and the Barrier was relatively small. Were the difference between the price of gold in the Gold Fixing and the Barrier to have been larger, the likelihood of Customer A’s position in the Digital being ultimately affected by Mr Plunkett’s orders would have been lower, although still possible.

This is not quite a once-in-a-lifetime opportunity, but neither is it likely that the gold market is normally driven by large digital option expirations. But when you happen to have a digital expiration, and it happens to be big, and the price happens to be right around the barrier anyway, then yeah, I mean, how can you not manipulate it a bit? It's the unusual confluence of perfect circumstances that makes the manipulation irresistible.

And it was easy to give in to temptation, because Barclays "failed to provide staff with appropriate guidance or training with respect to their obligations concerning their participation in the Gold Fixing." If you want to be a connoisseur of modern market manipulation -- and who doesn't? -- then paragraphs 4.30 to 4.44 of the FCA's Barclays order are essential reading. Barclays didn't lack policies about gold price naughtiness. It's just that those policies were generic: Barclays employees were "expected to make every effort to avoid situations and conflicts that may compromise or give the appearance the10 they may compromise [their] ability to carry out [their] responsibilities to the firm and its clients."

Plunkett gave that passage its natural reading: Don't get caught manipulating prices too often, or too egregiously. What were his "responsibilities to the firm and its clients"? He was a trader, not a salesman or client adviser; his responsibilities were to make money for the firm without making his customers too mad. He got that balance wrong here, but those things will always be in tension.

Barclays's new policies, on the other hand, are more to the effect of "don't manipulate against customers, at all":

Since 28 June 2012, Barclays has enhanced its systems and controls relating to the Gold Fixing. On 5 February 2013, Barclays formally adopted guidelines for the Precious Metals Desk, stating, for example, that traders could not participate in the Gold Fixing for any reason if they were a party to a Digital-type exotic observation against the Gold Fixing, that the firm had in place a formal policy or procedure that sought to specifically address conflicts of interest with customers who were counterparties to products that referenced the Gold Fixing.

So there'll be no more digital option manipulation of the gold fix, at least at Barclays.11 And the casual-chat format of the gold fix may itself be under pressure. This is a perfect specimen of market manipulation, but it may be the last of its kind.

1 And they kept at it: In November, they were in trouble for manipulating some electricity prices, though I have more sympathy for that. That might just be attributable to U.S. energy regulators not knowing what they're doing.

2 The customer paid $4.4 million for this option, which had another observation date a year later, with a possible $7.8 million payment. See paragraphs 4.8 to 4.10 of the FCA's order against Plunkett.

3 From the order it seems like Plunkett only got credit for $1.75 million of this, but Barclays presumably saved the full $3.9 million of the payment. (Sort of. Again, see paragraphs 4.8 to 4.10 of the order; Barclays was in effect still on the hook for the 2013 payment so this savings was mostly in expectation.)

4 This is taken from paragraphs 4.12 to 4.23 of the Plunkett order (and paragraphs 4.11 to 4.24 of the Barclays order). Paragraph 4.7 of the Plunkett order explains the process generally.

5 Or if you take the FCA's $1.75 million number as the correct measure of his profit (see footnote 3) then that number should be 1.9 percent.

6 See paragraphs 4.24 and 4.25 of the Plunkett order.

7 The $114,000 loss divided by 60,000 ounces is $1.90 per ounce above the $1,558.50 selling price, or $1,560.40 per ounce.

8 Not that the customer knew that -- it was an internal trade -- but still.

9 That's paragraph 4.12 of the Plunkett order, and is also absurd. Why not $1,558.95? Why not $1,558.9599? Why not just send the customer a note saying "HA HA I MANIPULATED YOU 'SUP?"

10 This is a [sic], but I don't know if the error is Barclays's or the FCA's.

11 And you'd be pretty dumb to do it elsewhere, though, I mean, who knows.

To contact the writer of this article: Matt Levine at mlevine51@bloomberg.net.

To contact the editor responsible for this article: Tobin Harshaw at tharshaw@bloomberg.net.