Net Open Position: How FX and Crypto Desks Measure Risk

The leftover after clients leave

Every time a broker helps a customer swap one currency for another, the broker’s own books briefly hold both sides of the trade. Buy dollars from a client and you own dollars, while owing whatever currency you paid with. Sell dollars to another client and much of that position unwinds. Most days, trades roughly cancel out. But not all of them do. What remains is called an open position. Add up every remaining long and short position in a single currency and you have its net open position, or NOP.

NOP matters because it turns a messy stack of hundreds of trades into something a risk desk can actually watch. If a desk is net long dollars, it tends to gain when the dollar strengthens and lose when the dollar weakens. If it is net short dollars, the opposite is true. That exposure exists whether the trades were made for customers or for the firm’s own account. This is why regulators treat FX risk differently from some other market risks: it applies to a bank’s whole currency book, not just its flashy trading book.

What counts as a position

A currency position is broader than ordinary spot trades. The usual list includes net spot positions, net forward positions, options converted to equivalent positions, and the interest and other flows that will arrive in that currency. All of it counts from the moment the trade is agreed, not on settlement day. The reason is practical: if you sign a contract today to receive euros next month, you are exposed to the euro today.

Options deserve a special mention because they hide risk behind maths. An option does not move one-for-one with the currency. Its sensitivity is measured by something called delta: the approximate change in the option’s value for a small move in the exchange rate. A risk desk converts an option into a “delta-equivalent” position, so the option can be compared with ordinary currency trades. Two different options with very different deltas can therefore produce very different NOP numbers, even when their face values look similar.

Netting within a currency, not across currencies

It is tempting to think that a long position in one currency cancels a short position in another. It does not. A euro and a pound are different assets, and they can move in opposite directions on the same day. So the usual rule is blunt: calculate the net position separately for each currency and do not allow a long euro position to offset a short pound position. The official phrase is “no offsetting between positions of opposite sign,” which is a polite way of saying the desk must keep currencies in separate buckets.

Imagine a broker with a net long of 20 million rupees from client trades and a net short of 20 million yen. That is not a flat book. The rupee could fall while the yen rises, leaving losses on both sides. Treating currencies as independent is a conservative choice, but a sensible one, because no exchange rate promises to move in step with another.

A desk can be net flat but still have a timing gap

NOP also has a time dimension. A broker might have 10 million euros arriving next week and 10 million euros leaving in six months. In pure net terms, the euro position is zero. But the cash flows are not aligned; a lot can happen to exchange rates between next week and six months. That is why standardised approaches often place positions on maturity ladders for each currency and calculate risk within time bands. The math does not assume one big pool of currency sitting safely in a vault. It assumes inflows and outflows must actually meet each other in time.

From a risk number to capital and limits

Once a desk knows its net position in every currency, it can do something useful: compare that exposure with the firm’s capital. The rulebook of one regional supervisor, the Central Bank of the UAE, offers a simple illustration of the standardised approach: convert each currency position into the reporting currency at current exchange rates, calculate the overall net open position, then apply an 8% capital charge. Other jurisdictions set different numbers and methods, but the underlying idea is the same. An open currency book is not free; it uses up capital that could otherwise protect the firm.

That is why risk desks set limits. A desk might say it will not carry more than a certain net exposure in a single currency, or more than a certain combined exposure across all currencies, measured against the firm’s capital. Supervisors also expect these limits to be monitored continuously rather than polished once a quarter. The rules warn against “window dressing,” which means making the book look tidy on reporting day while running bigger risks the rest of the time.

The structural hedging exception

There is one important exception to the basic capital treatment: structural foreign exchange positions. A bank that reports in pounds but owns a branch or assets in dollars will see its capital ratios swing as exchange rates move, even if nothing in the underlying business changed. To smooth that effect, a bank may deliberately hold a currency position that offsets the translation risk. Regulators can permit such positions to be excluded from the normal FX capital charge, but only under strict conditions.

These positions must be genuinely structural, not trading bets. They have to be deliberately designed to hedge capital ratios, documented, segregated from trading activities, and capped at the amount that actually neutralises the exchange-rate sensitivity. Some supervisors also require the exclusion to stay in place for at least six months. In effect, a firm cannot call something structural on Monday, use it to lower capital on Tuesday, and quietly unload it on Wednesday.

What NOP cannot promise

Net open position is a powerful compass, but it does not see every risk. It measures exposure to currency price moves. It does not measure settlement risk: the danger that one side of a trade pays out and the other side never delivers. The classic example goes back to 1974 and the failure of Bankhaus Herstatt, a German bank that was closed after receiving deutsche marks from counterparties but before paying out the corresponding dollars. The memory of that episode still shapes how settlement systems are designed. NOP would have looked fine just before Herstatt failed, which is a reminder that a low NOP is not a promise that nothing can go wrong.

None of this thinking is limited to formal banks. Whether a desk trades euros, dollars or dollar-pegged crypto assets, the question remains the same: after every order is netted out, how much of each currency is actually left, where is it due, and is that amount within the firm’s tolerance? NOP is simply the honest answer to that question, written in one line.