It is one of the oldest questions facing a trader – how best to work out of a position involving two currencies that are rarely, if at all, supported by most market participants. In the deep and distant past, when FX markets were predominantly voice-traded and risk was managed on a deal-by-deal basis, this question was less important, but the advent of e-trading, more pertinently, e-risk management, has seen a growth in the amount of exposures left in what in a certain age was called a “dodgy cross”.
With more institutions using correlations for their trading and risk management and more customers accessing the benefits of multi-currency netting, the amount of exposures in these crosses has, anecdotally, risen. All of which leaves the trader or treasurer with that age-old question ‘Do I clear this risk through the legs or the cross?’.
Until very recently, the decision as to which channel to use was left largely in the trader’s hands and instinct played a major role, but now execution analysis specialists BestX, have conducted an empirical study of the issue to help provide analytical support for that decision. Dean Markwick at BestX authored the paper, Assessing the Cost of Trading a Cross, and observes that while arbitrage theory ensures that the actual difference in trading via the two different channels in minimised, the actual cost of trading a cross versus trading through the legs can be more flexible thanks to different liquidity conditions, incentives from market makers and even just convenience.
The new paper uses algo fills in BestX’ unique dataset, meaning the firm is able to compare the relative cost of trading and assess where it might be cheaper or more expensive to trade the cross. “By using algo data we are attempting to ensure that any cross recorded was actually a cross trade that was put to the market and not the combination of booking two legs of a trade,” the paper states. “However, this is not guaranteed and we know that some algo providers may report a cross trade even if traded through the direct legs.”
The research filters the size of the fills to between 0.5 to 1.5 million dollars to remove the size effects, leaving it with approximately four million trades across 191 currency pairs. With this data BestX has constructed a network that connects all the possible combinations of a trade.
“We iterate through all the different combinations, taking what the average cost paid was in a cross and then compare it via the legs, be that against euro or dollar,” explains Markwick. “This gives us three different data points, the cross cost and the two legs. For each currency pair we calculate the median cost measured as the spread to the mid at the time of execution. The use of the median helps minimise the impact of the any outliers in the data.
“On the charts the straight diagonal line means the point at which both methods are the same, arbitrage theory is working well, but above the line you want to be trading the legs, and below, you should trade the cross,” he adds. “Generally speaking, everything is close to the line, as you’d expect, but there are some that are not and we think this is a nice visualisation for it that also offers a scale to provide a number for the potential cost improvement.”
Where the analysis really comes into its own also presents a paradox, for while the differences are starker when moving into emerging market – and even Scandinavian – currencies, inevitably the more esoteric the cross, the less reliable the data is. The picture also, when it comes to the Scandis, becomes a little more complicated, with some crosses, such as EUR/NOK being significantly better to trade than through the dollar legs – as is to be expected of course – while others, GBP/NOK for example, being better through the dollar legs.
Whilst acknowledging the potential shortcoming relating to a limited dataset, Markwick is keen to stress the anecdotal value of the analysis. “We are not trying to model it or predict what the costs will be,” he explains. “We are literally just taking the results that we have seen and plotting them in a way so traders can supplement their own knowledge or the information they have, such as an axe in one of the legs of a cross.”
As an additional bolster to the analysis, BestX repeated the process in June to see if the past two months had had any discernible impact on the original analysis which ran from 2017 to April 2020. “Due to the more illiquid nature of cross trading we were aware that outliers are more likely and new data can change the results easily,” Markwick says. “We wanted to ensure the analysis wasn’t sensitive to market conditions such as we have seen with Covid-19. The good news is that the outcome has remained the same across the G10 and EM pairs – there have been no major jumps over the dividing line or significant changes between the costs.”
While indeed some of the results are intuitive – most traders will know it is easier to trade a euro-scandi cross for example – the analysis should provide a useful extra layer of assistance in firms’ decision making, nowhere more so than in firms with specific currency-based accounting systems. “We think this analysis is especially useful for those institutions and companies who are perhaps dollar or euro-based in their accounting systems,” explains Pete Eggleston, co-founder of BestX. “The treasury function in these firms often gravitates towards the a certain currency for example, because it’s easier, even though it could be more expensive to trade through an alternate. By being able to actually quantify the difference in costs and show it in an easy manner then it really offers a value add to the business.”