Introduction to Financial Modeling with Gen AI in Banking

Explore the fundamentals of generative AI and financial modeling, including large language models, data generation, capabilities and pitfalls, and the typical inputs and processes of financial modeling.

Define generative AI, its capabilities and limitations, and contrast with discriminative models. Highlight banking use cases in text generation, from personalized lending to fraud detection and risk insights.

Explore the fundamentals of financial modeling, including inputs, assumptions, operations, and validation, and see how generative AI automates scenario analysis and valuation across banking, trading, fraud detection, and insurance.

Explore the fundamentals of financial modeling and generative ai, including the core models, processes, assumptions, and practical ways to apply them while avoiding common pitfalls.

Define AI literacy and explain what AI is and how it works. Explore foundation models, productivity impacts, and the three AI usefulness tiers and work modes.

Explore what generative ai is, how foundation models differ from discriminative ai, and the limits of local gains versus global impact within banking and finance.

Discover three tiers of gen ai usefulness, from text transformations and drafting to cautious decision support, with hype vs reality and limits.

Gen AI shifts work from execution to automation and redefines workers as workflow designers, emphasizing structured reasoning, subtasks, validation, and orchestration for bursty, efficient processes.

Explore foundation and multimodal models, mixtures of experts, tokens and embeddings, context windows, and hallucinations, with tuning approaches like fine tuning and retrieval augmented generation.

Identify seven core pain points when adopting gen ai in banking, from unrealistic expectations and insufficient tools to governance gaps, workflow misalignment, anxiety, and stochastic degradation.

Explore overview of financial models, use cases like valuation, pricing, and credit scoring, and four-step modeling process, with generative AI accelerating data gathering, shaping assumptions, building the model, and validation.

Explore common use cases of financial modeling in banking, including valuation with DCF and comparables, price forecasting with time series and machine learning, and credit and insurance risk models.

Explore four core model types—mathematical, statistical, simulation, and algorithmic—and their finance applications, including DCF valuation, Monte Carlo simulations, and Excel modeling.

Master the four steps of financial modeling—gather data, define assumptions and constraints, build the model, and validate results—across use cases like discounted cash flow valuation, credit risk, and insurance pricing.

GenAI and LLMS enhance the four-stage financial modeling process—gathering information, defining assumptions, constructing the model, and validating and using it—by structuring data, summarizing, generating scenarios, and producing reports.

Review the four-step modeling process and key model types: mathematical, statistical, simulation, and algorithmic, covering valuation, market pricing, and credit scoring, with generative AI support.

Explore how to model credit and credit portfolios using Gen AI, covering regression, credit scoring, and machine learning models, data sources, features, and how Gen AI augments each approach.

Explore the credit modeling process, comparing regression, credit scoring, and machine learning models, and learn how data gathering, assumptions, training, validation, and expert risk skills drive lending decisions.

Explore data sources for credit modeling, including financial, transaction, behavioral, demographic, and alternative data, and learn feature engineering like debt-to-income ratio, payment timeliness, and employment stability in regression models.

Explore regression models for credit analysis, focusing on the usual approach and logistic regression to estimate default probability from income, debt to income ratio, and credit history.

Explore regression models augmented by Genii to pre-process data, engineer features, and adapt thresholds in real time, improving predictor precision and enabling dynamic risk assessment.

Credit scoring uses weighted input data to compute a final borrower score, either standardized or bespoke. It relies on metrics like precision, recall, F1, ROC AUC, Gini, and KS.

Apply AI to augment credit scoring by personalizing weights and creating multiple tailored scores for different borrower types, with unstructured data, feature engineering, and real-time dynamic scoring.

Explore traditional machine learning approaches to credit analysis, detailing data collection, feature engineering, supervised and unsupervised training, and model evaluation, including biases, regulation, and explainability concerns.

Leverage Gen AI to augment data and features for credit modeling, debiasing and expanding datasets, generating synthetic data for testing, and delivering adaptive, explainable ML models across segments.

Conclude the module on credit analysis in banking using AI models with and without JNI, covering regression, credit scoring, and GenAI-enabled feature expansion.

Explore stress testing in banking using Monte Carlo, VAR, and CCR regulatory scenarios, enhanced by Gen AI and JNI to tailor scenarios, quantify tail risks, and prepare regulator-ready reports.

Explore stress testing in banking through Monte Carlo simulations, value at risk, and CCR regulatory models; gather data, set assumptions, build and validate models, and assess Basel capital implications.

Identify data sources for stress tests, including loan data, trading book, and how PD, LGD, and EAD drive Monte Carlo, VAR, and CCR models used in CCAR and DFAST.

Explore Monte Carlo simulations for stress testing in banking, modeling multiple input variables, sampling from probability distributions, and balancing accuracy with explainability through iterations, backtesting, and sensitivity analysis.

Use GenAI to generate diverse Monte Carlo stress scenarios, calibrate parameters beyond history, reveal correlations, and update simulations rapidly with new data.

Explore the value at risk (VaR) model for banking stress testing, covering historical, variance-covariance, and Monte Carlo methods, confidence intervals, and limitations such as tail risk and biases, with backtesting.

Explore how to enhance value at risk models with Gen AI to stress test banking systems, capturing tail risks and emerging scenarios through synthetic data and unstructured text.

Explore how CCR and DFAST stress testing uses composite models—Monte Carlo simulations, time series, and regression—chained to estimate losses, assess risks, and recalibrate for regulatory needs.

Explore how gen ai enables composite models for ccar/dfast stress tests, enabling institution-specific scenarios, qualitative data incorporation, parameter calibration, and automated regulatory documentation.

Explore the eight topics of the stress testing module, including Monte Carlo simulations with and without JNI, VAR and CCR or Dfast models, data features, and regulator-ready outputs.

Explore how banking fraud prevention uses rules-based, anomaly detection, and network graph models, augmented by JNI and GenAI, across transaction, user, and merchant data.

Explore the process, models, and expertise in fraud prevention, including rules-based, anomaly detection, and network analysis, with a four-step modeling workflow and cross-functional collaboration.

Explore data and features for fraud modeling, including transaction data, authorization data, platform data, historical fraud records, velocity, baselines, and network relationships for detection.

Apply rules-based models to detect banking fraud using thresholds, boolean logic, and condition cascades, while managing velocity, geographical, and channel checks and acknowledging maintenance, limitations, and binary outputs.

Leverage rules-based models enhanced by JNI to detect fraud, dynamically calibrate thresholds, personalize rules across seasons and cohorts, and translate between human language and precise rules.

Anomaly detectors use statistical and machine learning baselines across multiple features to flag deviations for fraud detection, incorporating transaction amounts, locations, time of day, devices, and data drift.

Explore how Gen AI enhances anomaly detectors for banking fraud by adding contextual data, multi-dimensional behavior profiles, and human-readable explanations for detected anomalies.

Represent entities as nodes and relationships as edges to detect coordinated fraud across accounts, devices, and ip addresses, using network models to reveal complex patterns and clusters.

Leverage GenAI to augment network models for fraud detection by resolving entities, capturing network evolution, and translating complex metrics into human-readable explanations.

Apply AI-driven fraud prevention techniques to detect and prevent fraud in banking. Explore rules-based, anomaly detectors, and network models, enhanced by Genai for clearer insights.

Learn to build anti-money laundering financial models with traditional methods and Genii, including transaction monitors, client risk scoring, and network models, using transactions, client profiles, payment networks, and relationships.

Learn how to build anti-money laundering models using transaction monitoring, risk scoring, and network analysis, guided by a four-step data process and essential regulatory expertise.

Explore data and features for anti-money laundering, including transaction data, customer information, sanction and watch lists, politically exposed persons, velocity and anomaly indicators, and network-based relationship mapping.

Explore transaction monitors that apply predefined rules and thresholds to detect money laundering, raising tiered alerts across amount, geography, and timing. Assess false positives, alert fatigue, and rule tuning.

Enhance transaction monitoring with Genai to dynamically refine rules and thresholds, contextualize alerts, and improve explainability for money laundering investigations.

Leverage Gen AI to dynamically tune risk factor weights in customer risk scoring, integrating structured and unstructured data, contextual information, and geography-based cohorts for a holistic, regulator-friendly view.

Discover how graph network models map entities as nodes and transactions as edges to reveal money-laundering patterns across networks, including centrality, clusters, and hub-and-spoke structures.

Explore how generative AI enhances financial modeling, covering LLM basics, the modeling process, and how AI gathers data, shapes assumptions, applies formulas, and checks errors.