The Last Look
Posted by Colin Lambert. Last updated: June 30, 2021
I am genuinely sorry to do this to you, but given events in the market yesterday and ongoing legal arguments, we need to do pre-hedging one more time.
I don’t know exactly what happened, but I can make an educated guess; someone had to sell a large amount of Swedish kroner and decided to do it at the 4pm Fix yesterday. It had to be large because the order was pre-hedged – and probably rightly so given what transpired.
Proving that the Fix can also be inappropriate away from the month end, EUR/SEK moved five big figures in the 10-13 minutes ahead of the Fix window, from the 10.1050 area to around 10.1550. It then traded fairly steadily during the window and has done ever since. I hate to think about the price action had the order been attempted in five minutes.
To put the move into perspective, that 15-minute range was the equivalent of the previous week in a currency pair that can be, to put it lightly, a little lively! Equally of interest, is that EUR/NOK, while it did react to a degree, only saw a one big figure range in the same window.
The sharper amongst you will already be pointing out that this event means that my argument for a 20-minute window would make little difference – I would of course, disagree, because of the likely impact of a five-minute execution – but I suppose there is a case to make that even in a 20-minute window, some orders might still be pre-hedged. I don’t think this takes into account the daily focus so many traders have on the 4pm window, not just month-end, however, for I would be very surprised if there wasn’t a fair bit of speculative activity as part of the move.
What continues to bother me about this is the obsession with one fixing window. WM takes its data from Refinitiv Matching and EBS Market with a bit of Currenex thrown in where necessary. Back in late 2006 I got a first look at what remains one of my favourite FX products, the Lehman Brothers’ Algo Workbench. What did it do? It allowed a customer to specify a window of execution for the order, which was executed according to a transparent TWAP methodology, and the customer was charged a spread, depending upon the time of day. Crucially, the rate the customer was provided with was the TWAP from the primary venue in that currency pair. How is this different from WM? Essentially it’s not, beyond nuances here and there, however in one important area, it is – the customer, possibly after talking with the bank, can specify the duration, and this meant what is now pre-hedging was included in the overall rate. This is how most execution algos work of course and I just can’t figure out why customers think they’re not getting a decent benchmark using an algo at a better time of day?
I would be fascinated to know the size of the SEK order because the second factor I want to discuss is deciphering when and if an order should be pre-hedged.
I was talking to somebody last week about the ECU Group case against HSBC that is currently being heard by the UK High Court. The case is one of six against banks brought by ECU, as it has pending actions against five other banks and in the HSBC case involves 52 trades between 2004 and 2006.
ECU Group highlights three stop loss orders in its claim, one each in EUR/CAD, EUR/USD and CAD/JPY. The two EUR trades were for a fraction over 167 million, the CAD/JPY for just over CAD 242 million.
As always, the case will be mired in legal-ese, but the crux of the matter is ECU is claiming it specifically said HSBC was to be an agent on the trades and that the orders were only to be executed when the market levels were hit. It has cited deal logs from HSBC that indicate that trading ahead of all three orders took place, some of it seemingly aggressive, that it believes led to the stop losses being triggered.
HSBC for its part, denies that it agreed to act as an agent in any way, says the client had not right to dictate how it acted (that’s not listening to the customer!), that it was a principal throughout, and, interestingly, it confirms the trading ahead of the orders, explicitly stating its dealers were pre-hedging. I don’t know if it should amuse me or not, but I did enjoy one riposte in the HSBC defence, where ECU Group claims that one trader bought EUR 101 million on EBS in three seconds, only for HSBC to deny that “it is in any way relevant or material that a “single trader” undertook EUR 154,000,000 of the EUR/USD Pre-Hedging or purchased EUR 101,000,000 in `just three seconds” via a programme buy order.”
Call me old fashioned but I would have thought any trades in the same market as the order, close to the level, are relevant – it may not have triggered the order but surely it has to be part of the trial?
Anyway, I digress, because a couple of things struck me about the relative arguments. Firstly, and focusing on the EUR/USD order, as the most liquid market, ECU claims that it placed the order at 13.42 UK time when EUR/USD was at 1.2119 – the order was to buy the euros at 1.2176. At around 13.52 UK time, the order was executed, but the trading logs indicate a heavy amount of buying (around EUR 175 million) in the minute prior to the order being executed (at 1.2179).
Sadly, for I think this is relevant, the ECU claim does not specify the rate at which these euros were bought, merely stating that only EUR 10 million was actually bought above the stop loss rate of 1.2176. I think it highly relevant because a 50+ point move in EUR/USD in 10 minutes is quite unusual, even back in 2006, away from data releases. As far as I can tell there were no relevant events during the 10-minute duration, however the market could have been rising thanks to a 13.30 UK time (8.30 EST) economic release from the US.
The courts will decide whether pre-hedging was allowed in the case, but either way there is a judgement call to be made here. If the buying in the one minute prior to the trigger was, for example, above 1.2070, then it could be argued that the market had moved over 50 pips in nine minutes and was therefore highly likely to go through 1.2176, hence why it was pre-hedged. If, however, the buying was from 1.2150 for example, that argument doesn’t hold up quite as well.
If this information does become public then it might help provide some guidance for the industry – personally I am not convinced that an order of that magnitude at that time of day needs pre-hedging in EUR/USD to avoid slippage (and I note the actual rate included two pips slippage and one point mark up – the latter ECU Group is also claiming was against their agreement).
Elsewhere, the CAD/JPY order provides some interesting context. It was placed at 17.15 UK time to buy CAD/JPY at 103.06, however the order was only to go live at 19.45 UK time and therefore, under current FX Global Code parlance, was an anticipated order. ECU Group claims that a similar pattern emerged, in that HSBC traders were in the market both ahead of the time and the level at which the order should have been triggered. Interestingly, however, in an early argument for a public access disclosure, ECU noted that at the time to order was placed CAD/JPY was at 102.09 and falling. It then says the market immediately reversed and rose above 1.0275 when it became at 19.45 UK time some two and a half hours later, following which it was subsequently executed. Again the timing is important because if the market had risen over 50 pips without HSBC doing anything specific in the market then it could be argued the trend was a factor in the decision to pre-hedge (whether that was allowed or not).
There are those who believe that the FX industry should move on when it comes to pre-hedging and that nothing more needs to be done – disclosures cover all activity, customers know what they’re getting into, nothing to see here.
Events like the move in EUR/SEK (and we have the month and quarter end tomorrow of course) and the claims made by ECU Group reinforce my belief that it cannot move on until there are better parameters. With pre-hedging still under a cloud in the US thanks to the Mark Johnson conviction, a decision against HSBC here could mean similar uncertainty in the UK, meaning the two largest centres in the FX world think there are issues with the practice.
The FX Global Code does a lot to provide guidance on many of the issues in the ECU-HSBC case, for example, stop loss orders are to be placed with explicit instructions as to what the triggers are and how the order should be executed. Equally, the recipient of the order has to state whether they are acting as an agent or principal, and if the client doesn’t like the answer then they should go elsewhere. Participants should also reveal whether they will pre-hedge the order, so a lot has been done to raise transparency around these types of orders.
While I can’t help but feel the ECU-HSBC case is another example of “different place, different time” in FX, however, the fact is pre-hedging is still an issue and I think more needs to be done.
Perhaps the best practice for the FX industry as things stand, would be for any pre-hedging trades to be disclosed to the client, time-stamped trade-by-trade, so that the latter can understand the “benefit” (or perhaps to understand they should give the executing party a longer window!) of the process. That way, the client can see the impact and amount of the pre-hedging (or hedging, I’m still confused on that one) and everything is clear and transparent. We do this with algo trades, all child orders are listed, why not do it for the pre-hedging?
It would potentially mean revealing (anonymised) details of other clients’ trades, for example Bank A had to buy EUR 5 billion of which Client B’s order was EUR 750 million, but this can be sufficiently post-trade for the information to have no value other than to perhaps make the client understand their order (especially at a Fix) rarely sits in isolation. There would need to be protections in place over market positioning for instance but this idea is about removing all doubt about a process, not about improving execution outcomes (although the educational process for clients should lead to a longer Fix window at the very least).
I have said before that the foreign exchange industry has a post-chat room problem in that it is seen as guilty until proven innocent. By raising transparency levels further over this particular area of controversy, it can at least provide reasonable assurance that it will not have to go through this whole mess again.