UBS spent 2025 proving what its platform does under load, and the numbers are hard to argue against. For clients aggregating liquidity across banks, that is the difference between a price that fills and one that rejects.
Those gains came from a decision taken years earlier. UBS has spent the past decade consolidating its trading technology onto a single quant trading platform, QTP, which now underpins pricing, execution and booking across spot, NDFs, short-term interest rates (STIR), options and precious metals.
The best trading technology removes complexity, giving clients faster and more seamless access to liquidity. But performance is ultimately measured in the moments that matter most
Ciara Quinlan
The commercial logic is reuse. Improvements developed for one desk can propagate across the franchise, and gains have travelled in unexpected directions: work done in equity derivatives, for example, has been pulled back into FX. The latest step is QTP-X, a new FIX engine that replaces legacy queue-based processing with snapshot-based shared memory. Crucially, it makes latency independent of market activity, which matters most in exactly the conditions where most platforms degrade.
UBS paired that with a less glamorous fix that clients felt just as acutely. Its credit engines sit in Switzerland and the round trip was adding latency visible on API and multi-dealer channels. A new credit checking methodology went in alongside the QTP work.
Clients noticed. UBS reports that its reject ratios now rank among the lowest in clients’ aggregated liquidity pools.
The more consequential shift is in pricing itself. Core Pricing Model 2.0 moves UBS from the position of reacting to market data towards anticipating it.
Those inputs include the bank’s own flow, CME and SGX futures activity for spot price discovery in Asia-Pacific, continuous linked settlement (CLS) data for real-time market share, and signals from mid-book dealer activity. The platform is built so new components can be added as signals prove out.
Automation has also been extended across harder products. In FX options, UBS automated marking and pricing for liquid pairs improved its implied volatility surface fitting, added anomaly detection for distorted markets and cut latency in specific fitting workflows from milliseconds to microseconds.
In STIR products trading, automated overnight indexed swaps curve construction, a full forward-forward spread matrix, liquidity tapering for illiquid hours and dynamic skewing transformed manual curve management into monitored, model-driven pricing. Execution moved onto a new order management system, Cobra, with smart order routing, expanded order types and voice-to-electronic migration. UBS also quotes on non-last-look venues including LMAX, extending that to NDFs in 2025.
Almost all of this is built in-house. Owning the code means owning the security remediation rather than inheriting vendor vulnerabilities. AI tooling is deployed across quants, developers and e-trading teams rather than confined to an innovation unit, and a spot analytics chatbot converts natural language questions into queries over approved datasets. Through all of it, CLS market share held steady in the 15% to 16% range.
“The best trading technology removes complexity, giving clients faster and more seamless access to liquidity,” says Ciara Quinlan, global head of principal algo trading and flow product. “But performance is ultimately measured in the moments that matter most. During periods of market stress, our platform continues to deliver the resilience and reliability clients expect from UBS.”
Plenty of banks improved their FX technology in 2025. UBS delivered a measurable step change across the entire stack and its clients have seen the results in their own data.
