Fixing the Fixed Fix - Barclays Case

The gold blogosphere is generally not known for its nuance - its a you're with us or against us black and white world. I suppose this is a result of the need for click baiting headlines to drive traffic to your site and I'm sure ambiguity doesn't survive the brutal A/B testing Darwinian selection process that is modern social media (and something I'll probably find out for myself when the Perth Mint gets some proper website software in a year's time and I select myself out of job if I persist with my current ways, such as not getting to the point quickly in the first filled-with-SEO-friendly-words paragraph).

So it is with the two silver Fix stories, being its closure and Barclay's manipulation of. The main example of nuancelessness was confusing manipulation with suppression and thus seeing Barclay's actions as proof of the latter. There is a big difference between the two, as I discussed here:

"I believe in manipulation but not suppression. One is short term, the other long term. Many of the manipulation and suppression theories are simplistic comic book stuff."

The fact that the manipulation in this case was downward fed the confirmation bias. It will be interesting to see the response if the next case (I would not be surprised to see another) has a bullion bank trader manipulating upwards, which, if you don't look at charts with one eye, you would have to say is equally probable given the evidence.

Anyway, below I quote from the Final Notice for Barclays and Daniel James Plunkett which can be found here and make comments after each bit I find interesting. The one good thing about this case is it gives us an insight into the fix which we have never seen before.

"Gold Fixing Members are required to declare their interest in increments of five bars, but there is no such requirement in relation to their underlying customers, i.e. their underlying customers can place their orders for any amount, not only in increments of five. ... At any time a Gold Fixing Member, or their underlying customers, may increase, decrease or withdraw a previously-declared selling or buying order or place a completely new order."

This is one aspect of the fix that very few understand - that the customers (big ones dealing direct with a bullion bank) can also change their orders during the fix. Even someone like Matt Levine in this article falls into the trap of seeing the fix as "five banks, getting on the phone, talking about what the price will be, and adjusting their trading based on that information".

Now I'm not saying that the current fix process is perfect (as the buy/sell balance I believe is not communicated and a bank can change its position in response to customer changes) but the fact that a bank's customers can change their orders as the fixing price changes means that manipulating the fix is uncertain as you don't know beforehand what customers will do, making it a risky proposition (even if you can collude beforehand with traders from other banks). Whatever the fix's problems, it is not some market where the banks just set the price amongst themselves, as Matt and others portray.

"On 28 June 2011, Barclays entered into the Digital with Customer A ... The Digital had a notional amount of approximately USD43m ... customer A paid a premium of 8.18% of the notional value, USD4.4m, to Barclays ... if the price fixed in the 28 June 2012 Gold Fixing at 3:00 pm exceeded USD1,558.96 (the Barrier), a payment of 9% of the notional amount, or approximately USD3.9m, would accrue to Customer A"

Based on PM fix of $1499 on 20/6/11, that puts the notional as around 29,000oz, or nearly a tonne of gold. This is no small customer. What I find interesting is that a client of this size could have "listened in" to the fix on the 28/6/12 and put its own orders in to influence the price. Of course it would have to have done that with another bullion bank, not Barclays.

Given the money on the table, the customer could have justified losing money on a large fix trade, just as Plunkett did (which was only USD 114,000). It certainly would have been an interesting fix, with the customer countering each of Plunkett's orders. Given that Plunkett would not earn all the $3.9m ("Mr Plunkett’s book thereby profited by USD1.75m (excluding hedging") it could be argued that the customer would have won out as it would be prepared to lose more than the $1.75m that Plunkett would have earned (assuming Plunkett would not have included his share of the initial $4.4m).

It may not necessarily have been naivety on the part of the customer to not protect its interests and maybe more to do with the fact it was already down $4.4m and didn't want to reduce its profit on the first option date given the gold market had peaked and it was unlikely to make a profit on the second date.

The above does raise the question I tweeted, namely: "Would it have been OK for client on other side of Barclay's digital gold option to manipulate gold price up by buying on the Fix?"

People's views on this matter differ, as I noted in this blog post:

"Manipulation is a continuum with differing views on what constitutes unlawful or unethical behaviour. Traders I’ve spoken to see most of it as just part of the “game”, like a boxing match to see who is stronger. I tend more towards the ethical end but not naive to think that you can walk in and put all your (price) cards on the table and not get screwed."

What I find interesting about this case is the assumption that their was a principal-agent relationship. It is not like the customer was asking Barclays to broker an order on Comex - a digital option is a pure OTC product and thus clearly for me if I was the customer I would know the bank was taking the other side, and thus we has a principal to principal relationship. That view is what is behind the comments from traders in this FT article quoted at GATA.

"There's a fundamental belief that both parties can aggress or defend their book, and I would have expected my traders to do so."
"If you have Goldman Sachs on one side and JPMorgan on the other, the gloves are off"

For example, when you go to a car dealer, you know they are lying to you about how desperate they are to sell the car and what their lowest price is, just as you are lying about how desperate you are to buy it and your maximum price. If you subsequently found out that the dealer would have sold it for $1000 less, you wouldn't have any cause of action against them. Indeed, you know that the dealer made a profit on the deal. They are not acting as a broker, selling to you at their cost plus and agreed upfront fee.

Now clearly the FCA investigation found that there was a principal-agent relationship but it seems somewhat naïve of the customer to just hope that the bank would say that "pushing around a benchmark is 'not quite cricket'" (as Mr Klapwijk was quoted) when the other side of trade is not a market professional, ignore the fact that practically it was a principal-principal arrangement, and not look to protect themselves from the conflict of interest. Then again, they did in the end protect themselves and were aware of the conflict of interest in querying the trade with Barclays, so maybe that was the most ethical way to address it.

"If the price fixed during the 20 June 2013 Gold Fixing exceeded USD1633.91, a payment of 18% of the notional amount would accrue to Customer A, less any accrued percentage payment related to the 28 June 2012 Gold Fixing."

I note that the PM fixed at $1292.50 on 20/6/13, so the customer net lost $500,000 on this trade.

"the proposed price quickly dropped to USD1,556.00, following a drop in the price of August COMEX Gold Futures (which was caused by significant selling in the August COMEX Gold Futures market, independent of Barclays and Mr Plunkett"

As Nanex ask, "How does the FCA know the drop at 10:00:23 was unrelated to Barclays or Mr. Plunkett? Do they have access to COMEX audit trail data? If so, why was there no mention of cooperation with the exchange or the CFTC?"

That early Comex move doesn't look like a coincidence. I don't read "independent of" as implying that the FCA actually investigated Comex trading, just that the price move occurred before Plunkett's actions. If you look at it from FCA's point of view, they already have a closed case, with Barclays having voluntarily done an internal investigation and bringing it to FCA's attention. Once they had their man on the illegal trading done 6 minutes later, what's the point of spending more time and money looking into trading on an exchange in another country before that?

I would also note here this quote from a Bloomberg article: "While commodity derivatives are regulated by the FCA, the London gold fixing isn’t. As a result, the trader’s actions fell outside the regulator’s criminal jurisdiction" so the FCA doesn't even have oversight over OTC gold trading, let alone US exchanges, and they only got him on "breaching the regulator’s principles of integrity".

What I think is interesting is that the CFTC should be looking into the 10:00:23 Comex trading, which we don't hear anything about. Seems like a good chance it will payout (in fines) for CFTC, certainly more of a sure thing than some other investigations they could spend their limited time on. Maybe it is because Plunkett contacted the Fed's go-to gold man who all the bullion bank traders have on speed dial to front their manipulative trades (sarcasm). Seriously, the CFTC should be looking into trading at this time.

While we are on Nanex, I would note that the timing of this case shows that Comex trading influenced the Fix, not the other way around as it is often presented, and as it was misinterpreted in this case. This is not surprising as an LBMA Alchemist article showed that price influence worked both ways between London and New York, shifting over time.

Indeed, the other observation is that Plunkett's subsequent Fix actions 6 minutes does not seem to have had much impact on Comex, which is not surprising considering how small it was relative to Comex volumes in this case.

Nanex also asked a few questions in their article, which I will have a stab at answering:

1. "Do poker players show everyone their hand at the beginning of a round?" Did Nanex actually read the FCA document, Plunkett was emailing internally, he wasn't showing his hand.
2. "Is Mr. Plunkett really that lucky?" Not sure why Nanex asks this, the whole case proves he wasn't and manipulated it down.
3. "Mr. Plunkett made no attempt to manipulate prices during the crucial first 6 minutes" He waited because he had the luxury of doing so as he could see how the fix was progressing. They normally take a couple of minutes, so 23 seconds in he had plenty of time to step in, but why do so and risk losing (as he did) on your fix trade if you don't have to and the market moves your way?
4/5/6. "How does the FCA know the drop at 10:00:23 was unrelated to Barclays or Mr. Plunkett?" Agreed, CFTC needs to look further into this.

"placed a large sell order of between 40,000 oz. (100 bars) and 60,000 oz. (150 bars) ... which led to Barclays declaring itself to be a seller of 52,000 oz. (130 bars)."

Not sure why it says "between", I mean don't they know exactly? Anyway, I love all this detailed stuff, which while only showing one day and not representative of all Fix trading, is interesting for me as to what it says about the volume of trading done on the Fix.

The key is the statement that after Plunkett's first order, the Fix was at "155 bars buying/345 bars selling", which is 2 tonnes buying, 4.3 tonnes selling. That selling is only $200m or so, which doesn't seem like a lot.

Subsequent Fix positions were "155 bars buying/215 bars selling" and finally "155 buying/145 selling". That is only two tonnes or $100m worth of trades. Not a lot and thus easy for a bullion bank or hedge fund to influence, which is probably why Plunkett was successful.

"before the price was fixed, there were a number of further changes in the levels of buying and selling in the 28 June 2012 Gold Fixing, which coincided with an increase in the price of August COMEX Gold Futures."

This bit is important because there are those that don't know the Fix is constantly being arbitraged to OTC and other market exchanges (which is obvious to any professional) like academics Caminschi and Heaney who "found" that "information from the fixing is leaking into markets prior the fixing results being published, and there exist economic returns for trading on these information leaks". Wow, you don't say, and given that customers can also adjust their fix orders during the process, they too can get economic returns, but if any serious player can do it, is it really an unfair leakage (as their work was presented in the blogosphere)? Caminschi and Heaney - you are just observing arbitrage here.

"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. He also subsequently reported his trading to Barclays’ Compliance. During Barclays’ internal investigation, Mr Plunkett provided an account of his trading during the Gold Fixing that was untruthful, in that he did not disclose the true rationale for his trading, or the reasons why he failed to disclose his trading to the Sales Desk on 28 June 2012. In giving this account, Mr Plunkett intended to give the impression that he placed orders in the 28 June 2012 Gold Fixing for reasons other than to increase the likelihood that the price of gold would fix below the Barrier."

So it wasn't a case of the FCA uncovering the illegal behaviour, it was only because the client complained, that set off a chain of events. It does make you wonder how many other derivatives that were close to the Fix which did not pay out will have customers reviewing and complaining. I do find it surprising that Plunkett, after realising that his trading on the Fix would be found out, then persisted it lying about why, given that surely Barclay's investigators would look at his whole book and find the digital option.

Next post I'll have a look at the closure of the silver Fix, and all whether that will fix the fixed Fix.