Despite Controversy, Last Look in FX Is Here to Last

The ability to reject quoted trades divides industry participants.

While a key practice for arbitrage trades, it might deter the buy-side.

Last look remains a controversial topic among non-bank liquidity providers in particular, but even its staunchest critics accept it isn’t going away any time soon.

Speaking at the Tradetech FX conference in September 2021, Guy Debelle (who was then chair of the Global Foreign Exchange Committee) referred to Last Look as a topic that generates ‘strong and sometimes diverse views across market participants’ – a comment that many would view as something of an understatement.

He also noted that the guidance paper published by the GFXC on the topic earlier in the year generated a larger volume of discussion and feedback than any other part of the review and update of the FX Global Code and said the GFXC intended to continue to monitor the application of Last Look.

In this context, it is hardly surprising that opinion is divided on the merits of allowing market participants receiving a trade request a final opportunity to accept or reject the request against its quoted price.

“We believe certain customers – particularly the buy-side – are entitled to execute their important FX transactions on venues like ours that bar last look practices,” says Jamie Singleton, the Chairman & CEO Cürex Group. “The reasoning is simple. Buy-side FX traders are operational and directional. Their goal is to get rid of risk, and their trades can be large with significant signaling risk and the potential for negative market impact if their trading intention becomes known in the marketplace.”

But, he also recognises that on a macro level, a customer using an ECN should get the liquidity their behaviour warrants and that no one size fits all. “So the use of last look is an option that may be important to certain liquidity providers who face unknown customers whose trading objective is to use speed and market data to make money,” adds Singleton.

A spokesperson for LMAX Group confirmed its view that Last Look should not exist on public multi-dealer platforms has not changed.

“We recognise that it may still have its place in disclosed bilateral trading relationships (bank to specific client) if both counterparties prefer to trade with it,” she says. “However, we would always recommend ‘no last look’ execution given the complexities and diverse nature of disclosures.”

On the pro side, as long as there continues to be transparency with accurate disclosures and a fair adoption of Last Look under principle 17 of the FX Global Code there is a place for it according to Hugh Whelan, the Head of EBS Direct at CME Group.

“ECN-style venues are unique in that one price with limited inventory is presented to many clients across many venues,” he says. “Last look is designed to facilitate timely credit checks and validation of prices and inventory.”

Last Look is fair when it is purely symmetric. It is normal for a liquidity provider to check that a proposed price remains within the tolerance range and does not lose money from stale prices or latencies.

That is the view of Mohamed Hajibe, the Head of the Global Institutional desk at Swissquote, who says pricing protection is necessary to support investment in bank pricing engines with connectivity to multiple data centres.

Eric Donovan, the Global Head of Institutional FX at StoneX Group acknowledges that using Last Look as a means for liquidity providers to generate additional profits on normal trading flows raises ethical concerns. However, he reckons most reputable market makers are using it as a protective measure against other market participants seeking to capture latency arbitrage.

Cutting Down Hold Time

“We are okay with last look as long as there is no additional hold time,” says Christian Lønborg Thomsen, the Team Lead of e-Trading Client Services at Saxo Bank, adding that in his experience hold times have come down substantially of late. “In some cases, liquidity providers have handled these reductions proactively, and at other times, we have had to request them to come down,” he adds.

Swissquote has observed a reduction in the hold time for trades subject to Last Look, especially those executed through direct API connections, says Hajibe. “There is still hold time for clients across secondary sources of liquidity, which is justified by less transparency between makers and takers in such venues.”

Swissquote’s average hold time for spot across institutional platforms stands at 7ms this year compared to 12ms in 2021.

“We have also seen hold times come down significantly and expect this trend to continue,” says Whelan. “Rather than completely disappear, we believe hold times will come down close to zero as further investment is made in credit check technology. As a consequence, we would expect that there would be no impact on spreads.”

Hajibe reckons Last Look could never be completely removed because of the effect that would have on bid/offer spreads, suggesting that this is particularly true for larger executions traded full-amount style. “The price discovery process and price distribution are expensive for the bank, and there is an increased risk of loss in case of mispricing for large amounts,” he says. “Removing last look completely would naturally force liquidity providers to widen spreads, which in the end would not necessarily result in better execution for the client.”

According to Donovan, all you have to do is look to the futures market or EBS as an example of what true CLOB FX trading would look like – the pricing is much wider than what we would typically see in the OTC market.

When an FX market maker distributes their best stream to a new user, they watch the deals very closely. If they see arbitrage activity, the first thing they might do is start rejecting some of the deals via Last Look and communicate that to the broker or user they had dealt with. Eventually, they will widen the spread or shut off the stream.

Defensive, Not Predatory

“Removing last look takes away one of the primary mechanisms that dealers have for handling predatory trade flows, so this would almost certainly result in wider spreads for market participants who do not have toxic flows and rarely – if ever – experience trade rejections,” says Donovan, who explains that when StoneX engages with a new liquidity provider it looks very closely at their rejection rate starting on day one.

“My expectation is that it will be zero or very close to zero,” he continues. “If not, I will discontinue that relationship very quickly, and I expect our clients to do the same if we were to reject them.

“Flow has to be earned – nobody is beholden to a single stream these days, and I believe this more than anything has pushed last look into being used as a defensive mechanism rather than an unethical profit driver.”

However, Singleton rejects the suggestion that removing Last Look would destroy bid/offer spreads and says clients who execute using Cürex-only liquidity have better outcomes on average, in terms of market slippage/implementation shortfall at the parent order level than when they use multi-venue alternatives.

“We have found that clients who use last look venues with the expectation of spread savings actually experience minimal or no benefit in return for the rejection risk they assume,” he concludes. “In our no-last look environment our top liquidity providers make good profit margins when interacting with our buy-side clients.”