Why the most expensive inefficiency on a City trading desk is the one nobody measures
At a time when geopolitics can be completely upended over the weekend, trades that have been agreed on a Friday but aren’t booked ’til Monday have become one of the biggest vulnerabilities in a desk’s profit and loss swing, says Oliver Blower
Since the dawn of time, City trading desks have prided themselves on their surgical precision. Duration is hedged, exposure is modelled to the nth degree, and capital is obsessively optimised. Why is it then, that one of the most persistent sources of loss making still sits outside every formal risk framework? That is, what happens when a trade is agreed, but not booked.
Sure, on a quiet day, a failed trade booking is nothing more than an irritating inconvenience to “get sorted next week”. But late on a Friday afternoon in today’s Trump sensitive world, it is a genuine risk. As prime case in point, it could be closing in on beer o’clock time in London, say 5:25pm, just when liquidity is thinning faster than the President’s tan. A rates trade is agreed verbally over the phone, but the trader already has a cool one waiting for him in the East Indian Arms and leaves the booking hanging until Monday.
The trouble is, in a world currently riddled with political unpredictability, Monday no longer resembles Friday. Take Saturday evening’s Truth Social post from President Trump hinting at sweeping 10 per cent tariffs on European goods in a negotiation tactic to capture Greenland. An unexpected escalation in geopolitical posturing like this can all be enough to reprice inflation and growth expectations and jolt the front end of the rates curve before Asia has even finished breakfast. By the time London comes back online, the price rate futures have dropped, the yield curve has shifted upwards, and the value of that unbooked trade has moved with it.
This is when the questions start flooding in from the risk department. Was the trade a payer or a receiver, and crucially, at what level? Was the size agreed cleanly or conditionally? Did both sides understand it the same way? What was said on the call? Every rates desk has lived some version of this story recently. The trade that sat unbooked over the weekend and reappeared on Monday carrying an unexpected profit and loss (P&L) swing. Sometimes the investment bank benefits, but quite often, it does not. More often than anyone cares to admit, the difference is quietly given back to preserve a relationship. That is not market risk, that is operational ambiguity on an unprecedented scale.
What makes this inefficiency so costly is that it hides in plain sight. It is rarely logged as an error. It does not show up neatly in capital metrics. Instead, it surfaces later as friction, rework and unexplained drift in desk performance.
Word of mouth
And yet, the information needed to prevent it already exists. Rates trading remains heavily voice driven. Even as clearing expands into US treasuries later this year, and execution becomes more electronic, intent is still formed in conversation. Those conversations capture context that booking systems never will. What was agreed, when it was agreed, and most importantly, under what conditions.
The problem is not so much lack of information, more a lack of visibility. Most desks still rely on memory, goodwill and reconstruction after the fact. Heads of desks sign off on risk without knowing which trades are sitting unresolved as the weekend begins. Compliance teams may see communications, but desk heads rarely see them in a way that helps them run risk proactively.
In calmer markets, this gap was tolerated. But in the current environment, it is increasingly exposed. We are now operating in a world where risk moves outside market hours and liquidity returns unevenly. Where a late Friday development in the US can reshape expectations before Europe wakes up.
In that world, unbooked trades have become one of the biggest vulnerabilities to the P&L of any rates desk. The irony is that improving this does not require faster trading or smarter algorithms. It requires the ability to connect what was said to what was booked, to surface unresolved trades before the world has a chance to move against them, not to mention replacing reconstruction with evidence.
This is less about surveillance and more about control. After all, the most effective desks are not the fastest, they are the ones that remove ambiguity early. They know what risk they are running before Monday morning arrives. In today’s unpredictable markets, the most expensive inefficiency on a rates desk is not speed. It is the cost of not knowing what you already agreed to.
Oliver Blower, ex-Merrill Lynch and Barclays’s rates trader, now CEO of VoxSmart