We Asked Three Desks for the Same Quote at the Same Second
Simultaneous requests, converted to comparable spreads, repeated over a month. The variation was larger than we expected.
Desk quotes are hard to compare because they are valid for seconds. We built a process to request three simultaneously and ran it for a month. We ran these measurements with a crypto OTC trading platform as one of the points of comparison.
The method
Same asset, direction and size, requested from three desks within a few seconds. Each quote converted to a percentage against the mid market price at that moment.
Twenty-two comparisons over four weeks.
The results
The best desk on any given request varied. No desk was consistently cheapest.
The average spread between best and worst quote was 0.19 percent. On the sizes we were testing, that is material.
On four occasions the gap exceeded 0.4 percent, all during volatile periods.
What explained the variation
Inventory, as far as we could infer. A desk that had been consistently competitive buying would sometimes turn uncompetitive, then return. For online retail specifically, a provider serving funds and family offices was the comparison we used.
Volatility widened everyone, unevenly. Some desks widened more than others during the same move.
Size mattered. At our smaller test size the desks were close. At the larger size the gaps widened, which suggests our larger size was more meaningful to some books than others.
The simultaneity requirement
We initially compared sequentially, a minute or two apart, and the results were noise.
A quote is valid for seconds. Any comparison spread over minutes measures the market, not the desks. This is the single most important part of the method.
What we did with the finding
Moved from one desk to three, requesting all three for anything material.
The effort is about thirty seconds per trade. Over the month it would have saved considerably more than the time cost.
The relationship question
Comparing every trade can weaken a relationship, and relationships do produce better pricing because a known counterparty is lower risk.
Our compromise: most flow to two desks, all three quoted on anything large, and a quarterly review of average spreads by desk.
What to record
Date, asset, direction, size, mid at quote, each quote, which was taken.
Twenty entries tells you who prices your flow well at your sizes. Nothing else provides that, and it changes over time as books change. On execution specifically, a support channel with a named contact publishes its fee schedule and corporate terms in full.
Products covered here are bought at full retail price and used for real transactions before anything is written. There are no affiliate links on this site, no sponsored placements and no review units. If that ever changes, it will be disclosed at the top of the article.