A collaborative effort between SAS, Man Group, Pension Insurance Corporation, and Stanford University has resulted in a groundbreaking ML model that forecasts corporate credit rating shifts more accurately, offering investors crucial early signals in a volatile credit environment.
A new machine learning (ML) model developed through a collaboration between SAS, Man Group, Pension Insurance Corporation, and Stanford University promises to revolutionise the forecasting of corporate credit rating changes. The innovative model has demonstrated a marked improvement over traditional credit risk assessment methods in predicting credit rating upgrades and downgrades, providing early-warning signals that could enable investors and portfolio managers to act before such changes are fully reflected by the markets or official rating agencies.
The model incorporates over twenty years of historical data, including KRIS default probability metrics, bond spreads, yields, equity performance indicators, and macroeconomic variables. By harnessing ML techniques, the system forecasts whether a company’s credit rating will improve, worsen, or remain stable. Backtesting has shown this technique to rank companies more accurately by their future rating shifts than conventional tools.
Stas Melnikov, Head of Quantitative Research and Risk Data Solutions at SAS and a key member of the development team, stated that the model’s “early-warning signals give critical time to act before the market fully prices in the event,” helping investors better manage risk, reduce losses, and capitalise on opportunities. This aspect is especially crucial for stakeholders such as insurers and fund managers who operate under regulatory mandates requiring high investment-grade credit quality.
The model’s introduction comes amid rising borrowing costs and increased pressure on corporate credit profiles. Despite the Bank of England maintaining interest rates at 4%, refinancing costs remain elevated compared to recent years, escalating risks for companies needing to refinance debt and raising default concerns. This context underscores the importance of early identification of credit rating changes that could impact asset prices and credit availability.
Drawing on an extensive dataset of more than 500,000 credit events from 2001 to 2024, the model prioritises input variables such as option-adjusted spread (OAS), yield-to-maturity (YTM), and KRIS one-year default probability (KDP), the latter proving influential in spotting companies vulnerable to rating transitions ahead of market recognition.
The stakes surrounding bond ratings are high; a downgrade below investment-grade can compel institutional investors following policy to dispose of certain holdings, often triggering price declines. Hence, advanced warning could enable asset managers to adjust portfolios proactively, potentially mitigating losses or capturing gains from rating movements.
Steven Desmyter, President of Man Group, expressed surprise at the model’s performance: “The results surprised even us. The new models significantly outperformed traditional approaches. In plain English, they were much better at telling us which companies were at risk of downgrade or upgrade.”
This breakthrough aligns with broader industry research highlighting the superior predictive power of machine learning models in credit rating classification. Studies have demonstrated that advanced ML techniques, such as Artificial Neural Networks, Gradient Boosting, Random Forests, and Support Vector Machines, significantly outperform traditional statistical methods, underscoring their growing relevance in financial risk management.
The integration of this model within SAS’s risk management ecosystem could benefit insurance companies and asset managers by enhancing asset and liability management, credit risk oversight, and provisioning for expected credit losses. By delivering more accurate forecasts and early signals of credit rating shifts, the model may help these institutions navigate uncertain credit markets more effectively.
In an era marked by volatility and tightening credit conditions, the adoption of machine learning-driven predictive tools represents a critical advance. These technologies enable investors to transition from reactive responses to credit events towards a more anticipatory and strategic risk management approach, potentially reshaping the landscape of corporate credit investing.
- https://itbrief.co.nz/story/machine-learning-model-gives-early-warning-on-credit-rating-shifts – Please view link – unable to able to access data
- https://www.sas.com/en_us/news/press-releases/2025/november/model-predicts-credit-rating-changes.html – SAS, in collaboration with Man Group, Pension Insurance Corporation, and Stanford University, has developed a machine learning model that forecasts corporate credit rating changes with high accuracy. This model provides early-warning indicators for credit rating upgrades and downgrades, enabling investors and portfolio managers to act before such changes are fully priced by markets and reflected by ratings agencies. The model leverages over two decades of KRIS default probability data and outperforms traditional risk assessment tools in predicting future rating changes.
- https://www.prnewswire.com/news-releases/new-risk-model-sets-benchmark-for-accuracy-in-predicting-credit-rating-changes-302606447.html – A new AI-driven risk model, developed by SAS, Man Group, Pension Insurance Corporation, and Stanford University, sets a benchmark for accurately predicting corporate credit rating changes. The model offers early-warning signals for potential rating upgrades and downgrades, allowing investors to act before market pricing and credit rating agencies’ actions. By integrating machine learning techniques with KRIS default probability data, the model significantly outperforms traditional approaches, providing critical time for investors to manage risk and seize opportunities.
- https://blogs.sas.com/content/sascom/2025/10/24/what-happens-when-machine-learning-meets-credit-ratings/ – SAS has developed a machine learning model that combines over two decades of KRIS default probability data to forecast corporate credit rating transitions with high accuracy. This model provides early-warning signals for potential rating changes, enabling investors to act before market pricing and credit rating agencies’ actions. By integrating machine learning with KRIS data, the model offers a powerful tool for anticipating downgrades and upgrades, enhancing proactive portfolio management and capital allocation decisions.
- https://www.mdpi.com/2227-9091/13/6/116 – A study published in MDPI’s journal examines the predictive performance of various machine learning models—Artificial Neural Networks (ANN), Gradient Boosting (GB), Logistic Regression (LR), Random Forest (RF), and Support Vector Machines (SVM)—in corporate credit rating classification across different countries. The findings reveal that ANN and GB models consistently achieve high accuracy, outperforming traditional approaches like LR. The study underscores the effectiveness of advanced machine learning techniques in financial prediction tasks, highlighting their superior performance in credit rating prediction.
- https://www.mdpi.com/1999-4893/18/10/608 – An article in MDPI’s journal discusses the use of machine learning to explain corporate ratings transitions and defaults. The study evaluates the performance of supervised learning algorithms—Logistic Regression (LR), Random Forest (RF), XGBoost, and Support Vector Machines (SVM)—across six credit rating segments. The results indicate that ensemble models like XGBoost and RF consistently outperform traditional approaches, demonstrating the potential of machine learning in providing insights into credit rating transitions and defaults.
- https://www.centerforfinancialstability.org/research/ForecastingBankCreditRatings.pdf – A research paper from the Center for Financial Stability explores forecasting bank credit ratings using various models, including neural networks and support vector machines. The study compares the performance of these models in predicting credit ratings and discusses their effectiveness in the context of bond portfolio risk management. The findings highlight the potential of machine learning techniques in enhancing the accuracy of credit rating forecasts, providing valuable insights for financial institutions and investors.
Noah Fact Check Pro
The draft above was created using the information available at the time the story first
emerged. We’ve since applied our fact-checking process to the final narrative, based on the criteria listed
below. The results are intended to help you assess the credibility of the piece and highlight any areas that may
warrant further investigation.
Freshness check
Score:
8
Notes:
The narrative is based on a press release from SAS dated November 6, 2025, announcing the development of a machine learning model for predicting corporate credit rating changes. ([sas.com](https://www.sas.com/en_us/news/press-releases/2025/november/model-predicts-credit-rating-changes.html?utm_source=openai)) The earliest known publication date of substantially similar content is November 6, 2025. The report has been republished across various reputable outlets, including PR Newswire and Structured Retail Products, indicating a high freshness score. ([prnewswire.com](https://www.prnewswire.com/news-releases/new-risk-model-sets-benchmark-for-accuracy-in-predicting-credit-rating-changes-302606447.html?utm_source=openai)) The press release nature of the content typically warrants a high freshness score. No discrepancies in figures, dates, or quotes were identified. The narrative includes updated data but recycles older material, which may justify a higher freshness score but should still be flagged. ([blogs.sas.com](https://blogs.sas.com/content/sascom/2025/10/24/what-happens-when-machine-learning-meets-credit-ratings/?utm_source=openai))
Quotes check
Score:
9
Notes:
Direct quotes from Stas Melnikov and Steven Desmyter are present in the narrative. The earliest known usage of these quotes is in the SAS press release dated November 6, 2025. ([sas.com](https://www.sas.com/en_us/news/press-releases/2025/november/model-predicts-credit-rating-changes.html?utm_source=openai)) No identical quotes appear in earlier material, indicating potentially original or exclusive content. No variations in quote wording were found.
Source reliability
Score:
10
Notes:
The narrative originates from a reputable organisation, SAS, a global leader in data and AI. ([sas.com](https://www.sas.com/en_us/news/press-releases/2025/november/model-predicts-credit-rating-changes.html?utm_source=openai)) The collaboration with Man Group, Pension Insurance Corporation, and Stanford University further enhances the credibility of the report. All entities mentioned have verifiable public presences and legitimate websites.
Plausability check
Score:
9
Notes:
The claims made in the narrative are plausible and align with current advancements in machine learning applications in financial risk management. The integration of over two decades of historical data and the use of machine learning techniques to predict credit rating changes are consistent with industry trends. The narrative lacks supporting detail from other reputable outlets, which is a minor concern. The language and tone are consistent with the region and topic, and the structure is focused and relevant. No excessive or off-topic detail unrelated to the claim was identified. The tone is formal and appropriate for a corporate announcement.
Overall assessment
Verdict (FAIL, OPEN, PASS): PASS
Confidence (LOW, MEDIUM, HIGH): HIGH
Summary:
The narrative is a recent press release from SAS, detailing a new machine learning model for predicting corporate credit rating changes. The content is original, with direct quotes from reputable sources, and originates from a credible organisation. The claims are plausible and align with current industry trends. The lack of supporting detail from other reputable outlets is a minor concern but does not significantly impact the overall assessment.