Swiss Bank Aims to Avoid Overfitting Crisis with New Strategy Testing Approach

Text
Swiss bank UBS has partnered with LGT Group to develop quantitative investment strategies (QIS) focused on Vix futures, aiming to mitigate risks associated with overfitting in algorithmic trading models. The collaboration, disclosed in a June 2026 internal memo, marks a significant step in addressing the challenges of volatility-based strategies amid shifting market conditions.

Subheading
Partnership Focuses on Vix Futures and Overfitting Risks
The partnership between UBS and LGT, a private bank based in Liechtenstein, centers on refining QIS for Vix futures, which track market volatility. According to sources familiar with the deal, the project seeks to avoid “backtest Olympics”—a term used to describe the excessive optimization of strategies against historical data, which can lead to poor real-world performance.

“Quantitative strategies that perform exceptionally in backtests often fail when deployed in live markets due to overfitting,” said a UBS spokesperson, citing internal risk assessments. “This collaboration is designed to ensure models remain robust across varying market cycles.”

Subheading
Vix Futures: A Double-Edged Sword for Investors
Vix futures, often referred to as the “fear index,” are widely used to hedge against market downturns or speculate on volatility. However, their complex pricing dynamics and sensitivity to macroeconomic shifts make them a challenging asset class.

LGT’s expertise in structured products and risk management complements UBS’s quantitative research capabilities, according to a joint statement. “The goal is to create strategies that balance innovation with prudence,” the statement said. “This requires rigorous testing and a deep understanding of both technical and market-specific factors.”

Subheading
Industry Context: Rising Scrutiny of Algorithmic Trading
The move comes as regulators and financial institutions increasingly scrutinize the risks of algorithmic trading. Overfitting, in particular, has drawn criticism for contributing to market instability during periods of stress. A 2025 report by the International Swaps and Derivatives Association (ISDA) highlighted that 34% of QIS firms had experienced significant losses due to overfitting in the previous two years.

UBS and LGT’s partnership aligns with broader industry efforts to enhance model transparency. “The challenge is not just in building sophisticated models, but in ensuring they adapt to real-world unpredictability,” said a market analyst at Bloomberg Intelligence, who was not authorized to speak publicly.

Subheading
What’s Next for the Collaboration?
While details of the partnership remain confidential, insiders suggest the teams are testing strategies on a limited scale. The focus includes integrating machine learning techniques with traditional risk management frameworks.

A UBS internal document obtained by a financial news outlet noted that the project is in its “early development phase,” with a target launch date of late 2026. The document also emphasized the importance of “stress-testing models under extreme market scenarios.”

Subheading
Why This Matters for Markets and Investors
The collaboration underscores the growing complexity of modern finance, where technological advancements must be balanced with caution. For investors, the success of this partnership could set a precedent for how institutions approach volatility-driven strategies.

“Vix futures are inherently risky, but with the right safeguards, they can offer valuable diversification,” said a portfolio manager at a European asset firm. “This partnership could signal a shift toward more resilient quantitative approaches.”

Quoted text
“Quantitative strategies that perform exceptionally in backtests often fail when deployed in live markets due to overfitting.”
UBS spokesperson

Quoted text
“The goal is to create strategies that balance innovation with prudence.”
Joint statement from UBS and LGT

Quoted text
“The challenge is not just in building sophisticated models, but in ensuring they adapt to real-world unpredictability.”
Bloomberg Intelligence analyst (unauthorized comment)