XTX sounds death knell for last look

XTX Markets has released a set of analytical tools that allow counterparties to quantify the impact of trading with each of their liquidity providers (LPs) and put a dollar amount on execution costs – a development the LP claims will end the use of last look. 

The free tool, dubbed XTX-ray, analyses trades based on spreads, round-trip times and fill ratios, as well as quantifying the dollar cost of rejects – defined as the amount by which the market mid-price moved between the time of sending the order and 100 milliseconds later.

Trades are also analysed for the immediate 100ms market impact, which gauges the behaviour of the counterparty by looking at the rate of decay. While transaction cost analysis (TCA) tools routinely examine metrics including fill ratios, the crucial missing elements such as immediate market impact and the cost of rejects is something only banks and sell-side providers could access until now.

Jeremy Snart

"At the moment, counterparties only have limited data points given to them by market-makers, so the extent they can measure counterparty behaviour or execution quality is limited, and not necessarily objective," says Jeremy Smart, global head of distribution at XTX Markets.

"What we have done is something much more objective than TCA, because we give people objective data and it's up to them how they interpret the findings based on that. We don't make a judgement on whether an outcome is good or bad; we just tell them how much it cost to execute a trade in dollar terms and what the immediate decay looked like 100ms after the transaction," he adds.

Rather than TCA tools purposed for analysing algorithmic execution, XTX-ray focuses on the risk-transfer business, which still represents the overwhelming majority flows.

Matt Clarke, also part of the distribution team at XTX, says rejection rates and response times do not give market participants the full picture. This is because a LP could have a 100% fill ratio and offer the tightest spreads consistently, and still turn out to be damaging to the overall quality of execution, without the counterparty noticing.

"Say a counterparty is trading in an aggregated last look environment, trying to get $5 million done in $1 million clips. The first LP accepts the trade, but immediately slams the primary market to accelerate the move," he says.

If you don't look at the cost of rejects and the immediate 100ms market impact, you would think that LP one, who accepted your trade, is the best counterparty in the world, but in reality that's completely untrue
Matt Clarke, XTX Markets

Meanwhile, the remaining $4 million is still waiting for responses from other LPs, who – in reaction to the sharp price move after the first LPs accepted the deal – then widen their spreads or reject the trade.

The client, in turn, is left with all the remaining clips to do, which means a replacement cost – in other words the cost of rejects – and a market that is going against them due to the significant signalling problem.

"If you don't look at the cost of rejects and the immediate 100ms market impact, you would think that LP one, who accepted your trade, is the best counterparty in the world, but in reality that's completely untrue," Clarke adds.

The cost of rejects can also vary significantly across venues, not just LPs. On anonymous ECNs, the presence of aggressive market-makers who hedge in the last look window can push up the cost of rejects many times – a factor that needs to be taken into account when measuring the cost of execution.

"A sharp decay in the 100ms immediate aftermath of trading inevitably means a counterparty is hedging during the last look window," says Smart.

XTX has published a disclosure policy on XTX-ray, which states the market-maker is not active in the last look window based on a trade request and, additionally, does not use any information from rejected transactions in its trading.

Bad behaviour

"When people talk about toxic flow, this is what they mean. No flow is intrinsically toxic; it becomes toxic because of the behaviour of one counterparty, whose actions result in a steep and sharp price move for a short period of time, which then recovers relatively quickly," Smart says.

Spreads quoted by big market-makers today are all determined by this level of analysis, with spread, response times, roundtrip times, fill ratios, the cost of rejects and the immediate 100ms market impact all taken into account.

The ability for liquidity takers to access the same data and tools used by market-makers for years means the problem of last look may become obsolete.

"When market participants start looking at the dollar cost of their execution, which takes into account the immediate 100ms market impact, the problem of last look disappears," Smart says.

"Some banks have expressed an interest in open-sourcing it – an idea we would be open to. We want the market to use it, so the industry becomes more transparent as a whole. We believe XTX-ray will likely signal the death knell for last look as a practice," he adds.