# Predictive AI for SME and Large Enterprise Financial Performance Management

**Authors:** Ricardo Cuervo  
**arXiv:** [2311.05840](https://arxiv.org/abs/2311.05840)  
**Date:** 2023-09-22  
**Summary page:** https://www.summarizepaper.com/en/paper/2311.05840v1/

## Key points

- Importance of financial performance management in business
- Limitations of traditional methods such as financial ratio analysis
- Introduction of new financial and macroeconomic ratios to complement standard ratios
- Use of supervised learning models including ML Regressors, Neural Networks, and Bayesian models to predict company performance and assess risks
- Incorporation of new variables improves model accuracy when combined with industry-standard ratios
- Feedforward Neural Networks (FNN) found to be simpler to implement and perform best among predictive tasks examined (ROA, ROE, Net Margin, Op Margin, Cash Ratio, and Op Cash Generation)
- Bayesian Networks (BN) can outperform FNN under specific conditions and provide a probability density function alongside predicted value
- Potential benefits for CFOs, CEOs, lenders, and investors in assessing risks of financial underperformance and analyzing financial statements accurately
- Contribution to advancing financial performance management by introducing new variables and predictive models that enhance accuracy in assessing company performance and potential risks.

## Summary

The study titled "Predictive AI for SME and Large Enterprise Financial Performance Management" by Ricardo Cuervo focuses on the importance of financial performance management in business and the limitations of traditional methods such as financial ratio analysis. The author introduces a new set of financial and macroeconomic ratios that complement the standard ratios found in Balance Sheets and Income Statements. To predict company performance and assess the risks of financial underperformance, Cuervo utilizes supervised learning models including ML Regressors, Neural Networks, and Bayesian models. The study concludes that incorporating these new variables improves model accuracy when combined with industry-standard ratios. Among the predictive tasks examined (ROA, ROE, Net Margin, Op Margin, Cash Ratio, and Op Cash Generation), Feedforward Neural Networks (FNN) are found to be simpler to implement and perform best. However, Bayesian Networks (BN) can outperform FNN under specific conditions. Notably, BNs offer an additional benefit by providing a probability density function alongside the predicted value. The findings of this study have significant potential in assisting CFOs and CEOs in assessing the risks of financial underperformance to guide companies towards more profitable directions. Additionally, lenders can better evaluate a company's condition while investors gain tools to accurately analyze financial statements of public companies. Overall, this research contributes to advancing financial performance management by introducing new variables and predictive models that enhance accuracy in assessing company performance and potential risks.