Articles
July 23, 2025
AI Accuracy: AtlasFX Model Trounces Manual FX Exposure Forecasts
# Foreign Exchange
# AI
# Sponsored Content
Scott Bilter of AtlasFX on FX market volatility, using AI to tackle forecast problems, and what he’s hearing from Europe.
Hedging currency risk effectively and the FX exposure forecasting it requires got more complicated for many treasury teams in the first half of 2025 as the U.S. dollar slumped the most in more than 50 years against a basket of other currencies. But it’s not completely clear yet how much of a price multinationals will pay for the currency market’s gyrations.
- “Sometimes you can get away with not having a great process when there’s not that much volatility to sort of call you out on it,” said Scott Bilter, co-founder and CFO of AtlasFX, in a Strategic Finance Lab podcast you can listen to now on Apple and Spotify. But the onset of volatility often exposes flaws with hedging programs or processes.
- “A lot of FX risk managers are kind of sweating bullets right now,” he said. “Folks will be going through their Q2 results and may be getting some nasty surprises.”
- Mr. Bilter knows what they’re going through. Before starting AtlasFX—a risk management and analytics platform—he and two other AtlasFX co-founders worked in treasury and FP&A roles at Hewlett-Packard.
- Their frustration relying on massive Excel spreadsheets is one reason they began innovating at HP and later started AtlasFX—which today is enhancing its offerings with artificial intelligence.
Scott Bilter, CFO, AtlasFX
A great AI use case. Mr. Bilter in the podcast describes to NeuGroup Insights editor Antony Michels how an AI model developed by AtlasFX helped a client struggling with forecasting intercompany FX exposures, a process complicated by the FP&A team not sharing treasury’s objectives as well as the unreliability and difficulty of obtaining forecasts from business units spread across the globe.
- The client ran the AI model in parallel with its manual process. “You get a scorecard every month—it’s like, all right, which one is more accurate?” Mr. Bilter said. “And we’re forecasting a lot of different things. And the AI model was just trouncing the manual just time and time again for many exposures.”
- Using the AI model helped reduce forecasting errors, saving money and time. “So it ended up being a great success story,” Mr. Bilter said. He and a colleague told the story with a member of the client’s treasury team at a NeuGroup session in April.
- “That’s just one example; but there can be a lot of different ways to apply AI to the split of FX at a company,” he said in the podcast.
The view from Europe. In other comments about the current market environment, Mr. Bilter said the dollar’s decline has prompted more outreach from European companies to AtlasFX. “Their dollar exposures or revenues are worth a lot less, so [there’s] a lot more discussion in Europe around maybe hedging a higher percentage or starting a hedging program when they weren’t at all,” he said.
- Some U.S. companies are also taking a closer look at exposures not considered core. “When there’s more volatility, maybe the next step is exposures that weren’t quite the level of needing to be addressed are maybe now getting to that that level,” he said. As a result, some clients are asking AtlasFX to run their unhedged exposures though a value at risk analysis to weigh the costs and benefits of additional risk reduction—i.e., hedging.
- Mr. Bilter also recommends companies stress test their unhedged exposures. “If you don’t like the results, then maybe that’s a good reason to tweak the hedging program a little bit,” he said. “You want to be prepared for everything. And I wouldn’t expect things to go back to a low-volatility environment anytime soon.”
Dive in