SOLVE Advances Pricing Precision With November 2026 SOLVE Px Release for Corporate Bonds
Key Takeaways
- •SOLVE plans to release an enhanced version of SOLVE Px for Corporate Bonds in November 2026 as part of its ongoing program of AI-driven predictive pricing improvements.
- •Preliminary testing against the current production model showed a 17.7% reduction in median absolute pricing error, to 6.8 cents, and a 32% decrease in standard deviation, to 39.2.
- •The release applies three model changes: separate handling of bid and offer quotes, greater weighting of trades with the strongest pricing signals, and removal of less valuable features to reduce noise.
- •SOLVE Px combines proprietary quote data, trade activity, and bond reference data across more than 300 feature inputs to generate real-time, size- and side-specific predicted prices with confidence scores for over 250,000 corporate bonds.
- •The reported performance figures come from preliminary testing rather than live use, and additional improvements are planned for future quarterly releases.

SOLVE, a leading provider of pre-trade data and predictive pricing for fixed income securities markets, has announced the November 2026 release of SOLVE Px for Corporate Bonds, marking the latest step in the company's ongoing program of enhancements to its AI-driven predictive pricing capabilities.
Building on SOLVE Px's established predictive pricing performance, the latest release demonstrates further gains in pricing precision. Preliminary testing against the current production model showed a 17.7% reduction in median absolute pricing error, bringing it to 6.8 cents. Standard deviation also decreased by 32%, to 39.2. Both are common yardsticks for pricing models: median absolute pricing error captures how far a predicted price typically sits from the observed outcome, while standard deviation reflects how consistent predictions are across bonds — a useful lens for desks that rely on model output for marks and risk checks.
"Accurate pricing is fundamental to the decisions our clients make every day, and these latest improvements reflect the evolution of our models through the continued innovation of our product and data teams," said Eugene Grinberg, Founder and Chairman of SOLVE. "As SOLVE continues to grow, our priority remains delivering the tools and insights that help market participants price securities and manage risk with confidence."
The latest release introduces three key model enhancements designed to further improve pricing precision and stability:
- Refined bid and offer inputs: The model processes bid and offer quotes separately to better reflect each side of the market and reduce unnecessary model reactions.
- More informative trade inputs: The model identifies and assigns greater weight to trades that provide the strongest pricing signals.
- Streamlined model inputs: Less valuable features are removed to reduce noise and improve the stability and accuracy of predicted prices.
SOLVE Px combines SOLVE's proprietary quote data with trade activity and bond reference data, driving more than 300 feature inputs to generate real-time, size- and side-specific predicted trade prices, each accompanied by a confidence score, for more than 250,000 corporate bonds. SOLVE Quotes, the company's quote data platform, processes more than 30 million quotes daily across fixed income markets. That mix of inputs reflects a structural reality of credit markets: many bonds do not trade on a given day, so quotes and reference data can offer a more current pricing signal than the last transaction alone.
According to the company, the latest enhancements are designed to provide clearer pricing context and more dependable inputs for front-office analysis. They are intended to support corporate bond traders, sales teams, portfolio managers, and wealth managers, as well as quantitative analysts and research teams that incorporate predictive pricing into their workflows. Additional improvements are planned for future quarterly releases, giving users a recurring point of reference for how the models evolve — and, since the reported figures come from preliminary testing against the current production model, how the gains hold up in live use will be the measure to watch as the November release rolls out.
"Corporate bond pricing requires more than historical trade data, particularly for securities that trade infrequently," said Michael Pellerito, Managing Director and Head of Product at SOLVE. "By refining how SOLVE Px incorporates quotes, trades, and reference data, we're improving the precision of our predictive pricing models across investment-grade and high-yield bonds. The latest enhancements demonstrate the continued evolution of SOLVE Px and the strength of our approach to predictive corporate bond pricing."
More information about SOLVE is available on the company's official website: The announcement was originally published by Global Fintech Series: https://globalfintechseries.com/trading/solve-advances-pricing-precision-with-solve-px-enhancements/