GFXC hopes for Principle 17 decision on November 15

The decision on whether to tweak Principle 17 of the FX Global Code of Conduct or leave the words ‘likely inconsistent’ intact could be announced as soon as November 15, the day after the Global Foreign Exchange Committee meets in London to discuss the issue.

The GFXC has published the submissions it received in response to the consultation on the wording of the controversial principle relating to the use of client information in the last look window. The central bank-led body launched the consultation in May, at the same time as it released the final version of the 55 principles related to good practice in foreign exchange markets.

Of the 33 responses received by the GFXC, six were handed in anonymously, while others varied greatly in length – from ParFX’s half-page note to LMAX Exchange’s 15-page comment – but the majority recommended the removal of the word ‘likely’ or suggested a change to the wording, with further clarification of practices.

At the moment, Principle 17 states: ‘During the last look window, trading activity that utilises the information from the client’s trade request, including any related hedging activity, is likely inconsistent with good market practice because it may signal to other market participants the client’s trading intent, skewing market prices against the client, which (1) is not likely to benefit the client, and (2) in the event that the market participant rejects the client’s request to trade, constitutes use of confidential information in a manner not specified by the client.’

The GFXC asked market participants to comment on two questions around the principle: whether they agree with this statement, including examples of instances when the use of trading information could be beneficial to the client; and, secondly, whether the language in the principle should be modified or changed.

Non-banks attack last look, latency floors

Non-bank market-makers were notable in their vehemence against last look, with Citadel recommending a full phase-out of it, while the response of DRW Holdings – comprising DRW Holdings, Eagle Seven, Geneva Trading USA and Optiver US – encouraged the GFXC to take steps to prohibit the use of the practice.

“We struggle to think of a situation where permitting such practices would be beneficial to the client or the market as a whole,” said DRW Holdings in response to the request to identify situations where the use of client information could be beneficial to customers.

DRW’s response also attacked trading platforms that operate a latency floor, noting such venues arbitrarily delay the processing of an already received client-trade request and in some cases randomise the processing of requests received during the same time window.

“Latency floors are commonly coupled with functionality that allows resting orders to be immediately cancelled by the matching engine, thereby prioritising the ‘cancel ahead of the client trade’ request even if the cancel is receive[d] subsequent to the receipt of the client trade request,” the response stated.

“In practice, this and other delayed matching models may be abused in the same manner as last look and should be similarly prohibited,” DRW Holdings added.

XTX Markets also advocated against the use of last look, and suggested Principle 17 be changed by removing the word ‘likely’ and swapping the phrase ‘it may signal to other market participants’ with ‘is very likely to signal’.

In practice, this and other delayed matching models may be abused in the same manner as last look and should be similarly prohibited

DRW Holdings

Information leakage resulting from pre-hedging client trades in the last look window was an area that UBS felt strongly about. In the Swiss bank’s to-the-point response, it deftly sidestepped the main question and said it agrees trading activity that uses client information may have a signalling effect to the market.

“We agree with the statement that trading activity that utilises information from a client’s trade request during the last look window may have the effect of signalling trading activity in the particular currency pair in the trade request. The resulting signalling, particularly when any such activity is on a ‘lit’ market, could have the effect of moving market prices in the subject currency pair or related currency pairs,” UBS said.

“The resulting market movement generally does not benefit the client – particularly in circumstances where the client is staging the execution of a currency position via multiple successive orders,” the bank added.

In response to the second question relating to the wording of the principle, UBS recommended the GFXC “consider amending the phrase ‘the client’s trading intent’ to ‘trading intent in the subject currency pair’”.

Meanwhile, the Investment Association, which represents the UK’s asset management industry, did recommend removing ‘likely’ from the principle and went on to say: “In addition, we consider that market participants should make direct statements to their clients, indicating they do not take part in pre-hedging activities during the pre-hedging window.”

Deal-and-cover or cover-and-deal?

Trading platform Thomson Reuters was one of the institutions that argued against tightening the language around Principle 17, or at least without making an exception for so-called cover-and-deal models of trading, where a bank acts as an intermediary in the price-sourcing process and only accepts the client’s trade if it can hedge it with its own liquidity providers.

Neill Penney, global head of trading at Thomson Reuters, said an end-user could ask its relationship bank for an FX price, but as the institution is a smaller, regional dealer, it has to obtain liquidity from other banks, known as provider banks. Based on the prices quoted by the provider bank to the relationship bank, the end-user receives a quote. If the client accepts the trade, the relationship bank starts a last look process before accepting the trade, in a bid to hedge the end-user’s trade with one of the provider banks.

“If it can do so, it will accept the end-user’s offer to deal. If not, it will reject the end-user’s offer to deal. The workflow is therefore to cover and then deal,” he wrote.

If, however, the dealers have to move to a deal-and-cover model, the relationship bank would first accept the end-user’s offer to deal and then attempt to hedge its position in the market with its provider banks, and spreads would widen, Penney said.

“To conclude, we believe that if ‘cover and deal’ is precluded as a best practice, the effect would be a material reduction in the liquidity of the FX market available outside the top tiers. We therefore see an industrywide benefit in identifying this workflow as an exception within Principle 17. Not only is the workflow consistent with good practice, but supporting it eliminates a likely impediment to FX Code of Conduct adoption by smaller banks,” he concluded.

It’s alright, it’s alright, it’s alright

A number of respondents argued for further clarification of the way last look is defined and/or called for ‘client’ to be defined as a concept. There were also a handful of firms who did not want any changes to the language of Principle 17.

TD Securities said last look is a useful risk-control mechanism that prevents stale prices coming in from clients.

No modification required. TD uses last look to minimise the likelihood of a client executing trades at stale prices that are not reflective of where the market is trading

TD Securities

“No modification required. TD uses last look to minimise the likelihood of a client executing trades at stale prices that are not reflective of where the market is trading. Technical issues, latency, as well as a number of other factors, can cause a client to send a trade request with a stale price,” the company said in response to the GFXC’s second question.

Standard Chartered said there would be no need to tweak the language if the Code defined the concept of last look windows.

“SCB believes any trading activity (including skewing) that utilises information from the client’s trade request should only be performed in good faith that a transaction will be consummated. Any failure to execute after trading activity should be rare and only on the basis of technical limitations; e.g. unanticipated network delays,” the bank said.

Decisions, decisions

Market participants familiar with the latest discussions within the GFXC tell FX Week there are strong voices arguing for the word ‘likely’ to remain in the document. During the drafting of the principle relating to pre-hedging in the last look window, the largest market players waged a battle of wills and influence over how strongly the Code will come down on the practice.

Just before the so-called fatal flaw version of the second batch of principles was circulated, the suggested final version of Principle 17 was changed from ‘pre-hedging in the last look window, specific to a client’s trade request is inconsistent with good market practice…’ to its current version.

Now, some market participants expect that rather than removing words, the GFXC will add more, while others say it is possible the current language will stay as it is.

A spokesman for GFXC chair Chris Salmon confirmed to FX Week that the committee will meet on November 14 and they are hoping to make the decision public soon afterwards, possibly the next day. “Depending on how discussions go,” he added.