Why waiting for a better exchange rate rarely pays for a business: what the research says about forecasting FX, what delay costs, and what to do instead.
Usually not. For a payment with a due date, waiting for a better exchange rate is a bet with an expected gain close to zero and a cost that's real: a late supplier, an annoyed contractor, idle cash. Research going back to 1983 finds short-term exchange rates are hard to beat with any forecast. Pay on schedule with a live quote, and hedge only exposures big enough to matter.
That's the short answer. The longer one is about when timing is fine, what to do with large exposures, and why a rate alert isn't a price.
Key takeaways
Not reliably over the horizons that matter for payouts. In a famous 1983 paper, Richard Meese and Kenneth Rogoff compared the out-of-sample forecasts of the leading exchange rate models with a simple random walk, which predicts that tomorrow's rate equals today's. The random walk performed as well as any model at one- to twelve-month horizons for the dollar against the pound, mark, and yen, even when the models were fed the actual future values of their inputs (Meese and Rogoff, Journal of International Economics, 1983).
Forty years of follow-up work softened that result without overturning it. Barbara Rossi's survey in the Journal of Economic Literature concludes that predictability depends on the predictor, horizon, sample, and method, and that the random walk without drift remains the toughest benchmark (Rossi, "Exchange Rate Predictability", 2013).
If the models built by economists struggle, a payments team checking a chart before lunch won't do better. That's not a criticism. It's the starting assumption to build policy on.
Waiting has a cost even when the rate doesn't move. The rate side is a coin flip; the delay side is not.
| What you give up by waiting | Who feels it |
|---|---|
| On-time payment to a supplier or contractor | The recipient, then your relationship |
| Early-payment discounts or late fees in the contract | Your P&L |
| Cash sitting unconverted instead of doing its job | Treasury |
| Someone watching a chart instead of working | The team |
| A clear audit trail of why you paid when you did | Finance and auditors |
The rate side is symmetric. If you wait a week, the market can move for you or against you. Who carries FX risk in a stablecoin payout maps where that risk sits at each step.
You owe a supplier in Brazil and plan to send 50,000 USD. Today's rate is 5.20 BRL per USD. Assume that over a week the rate is equally likely to end 2% higher or 2% lower (illustrative numbers, not a forecast).
| Scenario | Rate | BRL delivered for 50,000 USD |
|---|---|---|
| Pay today | 5.20 | 50,000 x 5.20 = 260,000 |
| Wait, rate up 2% | 5.20 x 1.02 = 5.304 | 50,000 x 5.304 = 265,200 |
| Wait, rate down 2% | 5.20 x 0.98 = 5.096 | 50,000 x 5.096 = 254,800 |
| Expected value of waiting | (265,200 + 254,800) / 2 | 260,000 |
The expected outcome of waiting is exactly what you'd get today: 260,000 BRL. What waiting adds is a 10,400 BRL range of outcomes and a payment that's a week late.
Now add the delay cost. If the contract has a 1% late fee (illustrative), paying a week late costs 2,600 BRL with certainty, against an expected rate gain of zero. Waiting loses money on average. That's the general shape: a certain cost against a gamble with no edge.
If the supplier invoice is in reais, the picture flips but the logic doesn't. You owe a fixed BRL amount, the dollar cost moves with the rate, and waiting is still a bet. Paying an exact local amount with receiver-side quotes shows how to quote that case.
When there is no deadline and you write the rule down first. Examples:
The rule matters more than the target. "Convert 25% each Monday" or "convert when the rate crosses X, or on the 25th, whichever comes first" are policies. "Wait until it looks better" is not. A written rule stops the team from moving the target every time the rate moves.
They help as a trigger, not as a price. An alert watches an indicative rate, the kind you see on Google or a market screen. When it fires, you still need a binding quote, and that quote includes the provider's margin and fees. Why the Google rate differs from your quote explains the gap.
So the flow is: alert fires, request a quote, check the delivered amount, execute inside the window. A BlindPay payout quote locks the rate and fees and expires 5 minutes after creation (payout quotes docs). Indicative rate vs binding quote covers how to show both in a product.
Make timing boring. Four habits beat watching charts:
If you want to see how much a currency typically moves in a month before you size that range, the Federal Reserve's daily series on FRED is free, for example Brazilian reals per US dollar. How to read an FX quote helps you read the direction right before you compute anything.
Write a one-paragraph timing rule for your team, "payouts run on schedule, quoted at execution", and request one live quote for your next scheduled payout using the payout quotes docs.
Seven places blockchain payments beat bank rails: contractor payroll, remittances, B2B suppliers, marketplaces, 24/7 treasury, bill pay, and PSP payouts.
Push means the payer sends an ACH transfer. Pull means you debit a connected account. How each funds a USD to USDC on-ramp, the fees, and the return risk.
AP2, ACP, and x402 each verify that an AI agent had permission to spend. Here is what every protocol covers, who backs it, and the reconciliation gap none of them close.