The hidden cost of a manual FX desk

The price you never see.

Every forward you book has a number that is not printed on the confirmation: the gap between the rate you were quoted and the real market mid. Banks quote both sides of the spread, and the fair midpoint sits between them, where the dealer keeps it. You see the edges. The dealer keeps the middle.

It gets harder to spot because pricing is not comparable across counterparties. Every dealer prices off their own curve, so the same six-month forward can come back from two banks fifteen basis points apart, and both look perfectly reasonable in isolation. Without an independent reference, there is no way to know which quote was fair and which one was not, until it is already booked.

The only defence is measuring every quote against a neutral mid before you click execute, not discovering the overcharge in a post-trade report when nothing can be done about it.

The confirmations typed by hand.

A trade confirmation arrives as a PDF. Someone reads it, types the currency pair, the notional, the strike, the value date, and the barriers into a spreadsheet. On a vanilla forward, that is tedious but low-risk. On a structured trade, it is where things break.

A knock-in/knock-out barrier or a TARF has terms that do not survive manual entry well. Miss a barrier level, transpose a strike, or drop a fixing date, and the error does not announce itself. It sits quietly in the book until the trade approaches that level, and by then the first person to notice is often the auditor, not the desk. Confirmations re-keyed by hand mean missed barriers and KIKO terms hit your compliance file before they hit your attention.

Reading those documents automatically, with every leg, barrier and premium extracted and confidence-scored for a human to sign off, removes the single most error-prone step in the whole workflow.

The exposure you cannot see whole.

Most teams can tell you their exposure one currency at a time, or one spreadsheet tab at a time. Far fewer can put forecasted invoices, signed contracts and existing hedges into a single view and say, in one number, where they are over-hedged, where they are under-hedged, and which months need attention before they bite.

That gap matters because FX risk is a coverage problem, not a per-trade problem. A Q3 bucket sitting at 40% hedged against a 75% policy floor is a real exposure, but it is invisible if your forecasts live in one place, your hedges in another, and nothing reconciles the two. The cost here is not a bad rate. It is the position you did not know you were carrying.

A desk that only records risk after the fact is not cheaper than one that manages it. The bill is simply paid somewhere no one is counting.

The month-end that eats a week.

When the book lives in spreadsheets, reporting is reconstruction. Someone has to gather the trades, mark them to market, work out hedge effectiveness, and assemble something an auditor will accept, all after the fact and mostly by hand. The work is slow, it is error-prone, and it produces a snapshot that is already stale by the time it is finished.

A live book inverts that. When every trade is already reconciled, already marked to market, and already logged with a timestamped decision trail, month-end stops being an event. The audit-grade record is a by-product of running the desk, not a separate project bolted on at the end of the quarter.

Manual is not cheaper. It is just uncounted.

The instinct to keep FX manual usually comes from a sense that it is the low-cost option. No software bill, no implementation, just a spreadsheet and a few relationships with banks. But the costs are still being paid. They are paid in overcharges no one benchmarked, in structured-trade errors that surface late, in exposures that stay hidden until they move, and in the analyst-days swallowed by month-end. They are just not on an invoice, so they are easy to ignore.

The alternative is not a bigger team or a bank-sized budget. It is putting the whole book, from the quote to the confirmation to the mark-to-market, into one live view where the fair price is checked before you trade, the confirmation is read for you, the exposure is netted in real time, and the record writes itself. That is institutional-grade discipline that a two-person treasury team can run from day one.